22
Homesickness among students in two cultures: Antecedents and consequences Margaret Stroebe 1* , Tony van Vliet 1 , Miles Hewstone 2 and Hazel Willis 3 1 Utrecht University, The Netherlands 2 Oxford University, UK 3 Cardiff University, UK Review of the theoretical and empirical literature on homesickness showed that despite recent advances, scienti c understanding of the impact on students of leaving home for college is still limited. Further empirical investigation using standardized measures, structural equation models and including additional mediating/moderating variables is needed. Two studies were thus conducted, one in the Netherlands, and one in the UK. Homesickness was investigated among recent-intake students, using a newly developed instrument, the Utrecht Homesickness Scale. Variables investigated in relationship to homesickness included depression, personality factors (self-liking, competence, self-esteem) and family situation (attachment to family). Homesickness was found to be a common though differentially prevalent phenomenon (approximately 50% in the Netherlands; 80% in the UK). Structural equation models showed that students missed family and friends and had dif culties adjusting to college life. These dif culties were associated with ruminations about home and loneliness, which themselves were associated with depression. There were differences in intensity (UK students were more homesick) and there were gender differences (UK females experienced more homesickness). Both the personality and family situation factors had an impact on homesickness. The results supported the conceptualization of homesickness as a ‘mini-grief’, to be viewed from theoretical perspectives in the eld of loss and bereavement. Empirical and theoretical overview of student homesickness There is a limited scienti c literature on the topic of homesickness. This is surprising, since departure from the home environment, particularly for young persons, can be a 147 British Journal of Psychology (2002), 93, 147 –168 © 2002 The British Psychological Society www.bps.org.uk *Requests for reprints should be addressed to Margaret Stroebe, Department of Clinical Psychology, Utrecht University, PO Box 80140, 3508 TC, Utrecht, The Netherlands.

res_6

Embed Size (px)

Citation preview

Page 1: res_6

Homesickness among students in two culturesAntecedents and consequences

Margaret Stroebe1 Tony van Vliet1 Miles Hewstone2

and Hazel Willis3

1Utrecht University The Netherlands2Oxford University UK3Cardiff University UK

Review of the theoretical and empirical literature on homesickness showed thatdespite recent advances scienti c understanding of the impact on students of leavinghome for college is still limited Further empirical investigation using standardizedmeasures structural equation models and including additional mediatingmoderatingvariables is needed Two studies were thus conducted one in the Netherlands andone in the UK Homesickness was investigated among recent-intake students usinga newly developed instrument the Utrecht Homesickness Scale Variables investigatedin relationship to homesickness included depression personality factors (self-likingcompetence self-esteem) and family situation (attachment to family) Homesicknesswas found to be a common though differentially prevalent phenomenon (approximately50 in the Netherlands 80 in the UK) Structural equation models showed thatstudents missed family and friends and had dif culties adjusting to college life Thesedif culties were associated with ruminations about home and loneliness whichthemselves were associated with depression There were differences in intensity (UKstudents were more homesick) and there were gender differences (UK femalesexperienced more homesickness) Both the personality and family situation factorshad an impact on homesickness The results supported the conceptualization ofhomesickness as a lsquomini-griefrsquo to be viewed from theoretical perspectives in the eldof loss and bereavement

Empirical and theoretical overview of student homesicknessThere is a limited scientic literature on the topic of homesickness This is surprisingsince departure from the home environment particularly for young persons can be a

147

British Journal of Psychology (2002) 93 147ndash168copy 2002 The British Psychological Society

wwwbpsorguk

Requests for reprints should be addressed to Margaret Stroebe Department of Clinical Psychology Utrecht UniversityPO Box 80140 3508 TC Utrecht The Netherlands

highly stressful experience (see Fisher 1989 van Tilburg 1998 van Tilburg Vinger-hoets amp van Heck 1996 for recent reviews of the literature) Homesickness is a morewidespread phenomenon than is usually believed As many as 50ndash75 of the generalpopulation have experienced homesickness at least once in their life (Fisher 1989)Furthermore 10ndash15 of the homesick have experienced it to such an extent that itinterferes with their daily activities (Fisher 1989) While there have been some previousstudies on homesickness among college students (eg Archer Ireland Amos Broad ampCurrid 1998 Brewin Furnham amp Howes 1989 Carden amp Feicht 1991 Fisher ampHood1987 1988 van Tilburg 1998) these only yield information on a limited number ofpsychological variables and employ a variety of homesickness measures some of themnon-standardized In fact even a precise denition of homesickness is hard to nd in theliterature Generally it is understood that homesick children or students miss theirparents and family friends and other familiar persons their familiar surroundings andhome comforts and they feel extremely insecure (cf Baier amp Welch 1992 Eurelings-Bontekoe Vingerhoets amp Fontijn 1994 Fisher 1989) With respect to student home-sickness some dimensions include lsquodisliking the universityrsquo as well as lsquoattachment to thehomersquo (cf Archer et al 1998) These dimensions reect the varied descriptions that canbe found which range from lsquodepression as a result of absence from homersquo (The OxfordDictionary) to lsquolonging for home and family while absent from themrsquo (WebsterrsquosDictionary)

Empirical overviewWhat is the state of knowledge about prevalence predisposition for and impact ofhomesickness among students thus far Wide differences in the prevalence of home-sickness have been reported although it is generally considered by researchers toaffect people from all cultures and all age groups (van Tilburg et al 1996) Research hasindicated that homesickness occurs in large numbers of students who leave homeAmong student populations in Great Britain for example between 30 and 60 ofboth men and women reported being affected by homesickness during their rst yearat university For example Brewin et al (1989) found that 40of the students in theirsample reported feeling homesick 20 were unsure and 40 reported not feelinghomesick By comparison also among students Fisher and Hood (1987 1988) found30ndash35 to be slightly to very homesick while 65ndash70 were not On a dichotomoushomesickness measure Fisher Murray and Frazer (1985) found that 60 of the rstyear students in their sample reported being homesick Vastly different percentageshave been reported for other countries Carden and Feicht (1991) for example foundthat 19 of American and 77 of Turkish students attending universities in their owncountries could be classied as being homesick in their rst year (based on a cut-offpoint of 1 standard deviation above the mean rating for the group on a homesicknessquestionnaire) The diversity in percentages both within and between countries islikely to be due at least in part to differences in the scales used and in the duration oftime since leaving home Indeed it is sometimes difcult to compare results acrosssamples particularly with respect to prevalence of homesickness because investigatorsuse their own specic homesickness scales (cf Archer et al 1998 Eurelings-BontekoeTolsma Verschuur amp Vingerhoets 1996 Eurelings-Bontekoe Verschuur Koudstaal ampvan der Sar 1995 van Tilburg Vingerhoets amp van Heck 1997)

The psychological literature on predispositional factors risk factors and mediatorsof homesickness has mainly studied personal characteristics which predispose persons

148 Margaret Stroebe et al

to become homesick and characteristics of the environment or situation that areassociated with greater intensities of it Highly relevant for student populations is thequestion of gender differences as a risk factor for homesickness While lay assump-tions might lead to expectations of greater female homesickness on leaving homefor college the literature does not support this Brewin et al (1989) reported thathomesickness was similarly prevalent among male and female students a ndingconrmed by Fisher for a variety of samples (eg Fisher et al 1985 Fisher amp Hood1987 1988) Ways of coping however differed between the genders with womenseeking more social support than was the case for men Similarly Archer et al (1998)found that female students showed higher levels of intrusive thinking about home-sickness mediated by their higher scores on the attachment factor of a homesick-ness questionnaire whereas there was no gender difference in avoidant responsesto homesickness There are gender differences in coping with and consequences ofother negative life-events notablymdashin this contextmdashbereavement (eg health detri-ments are relatively more extreme for widowers than for widows see Stroebe ampStroebe 1987) Thus further examination of patterns of gender differences in home-sickness would seem timely

Van Tilburg et al (1996) have recently emphasized the need for further examinationof cultural differences in the prevalence and experience of homesickness noting thatthis had not been studied systematically Carden and Feicht (1991) studied home-sickness among Turkish and American students in their own countries but limitedinvestigation to females Hojat and Herman (1985) studied a sample of Iranian andFilipino individuals but these were all resident in the USA and the sample wascomposed of physicians Ward and Kennedy (1993) examined homesickness amongstudents in different cultures but these were all from New Zealand residing in 23different countries of the world Lu (1990) also examined students who had moved tostudy in a different culture but again the study was conned to one cultural groupnamely Chinese students all of whom had moved to a university in the UK

Particularly important too in the context of our investigation are ndings to dowith personal vulnerability factors Both Carden and Feicht (1991) and Brewin et al(1989) reported associations between dependency and homesickness In the Brewinet al (1989) study lsquodependency on othersrsquo was a predictor of homesickness in theCarden and Feicht (1991) study dependency on family and parents was found to becharacteristic of homesick but not non-homesick students Personality and dispositionalcorrelates of homesickness have also been identied among student samples Fisher(1989) found that introversion depression and obsession were related to homesicknessbut that self-esteem was no different among homesick versus non-homesick studentsIn other samples lower self-esteem has indeed been reported for homesick samples(Eurelings-Bontekoe et al 1994 Hojat amp Herman 1985) Furthermore Eurelings-Bontekoe et al (1994) also found higher levels of introversion rigidity and negativismand lower levels of dominance among homesick conscripts Rigidity proved to be thebest predictor for homesickness These patterns are not however always conrmedfor other samples (cf van Tilburg et al 1996)

Turning from personal to situational factors lsquogeographic distancersquo is also importantfor the current investigation Among students Fisher et al (1986) found homesicknessto be higher for those more distant from home (although notably distance fromhome did not make a difference among boarding school children) Needing furtherinvestigation are related factors such as psychological distance and possibilities forcommunication with home

Homesickness among students in two cultures 149

Finally with respect to the consequences or outcomes associated with homesick-ness among students studies have also found a relationship between homesicknessand a greater number of cognitive failures poor concentration handing in work latedecrements in work quality and higher scores on anxiety and depression measures(Burt 1993) This has recently been conrmed in a meta-analysis of four studiesconducted by Archer et al (1998) As Brewin et al (1989) emphasized such ndingssuggest that homesickness is a potentially important phenomenon that may exercisea considerable inuence on academic performance at least in the short term

A theoretical approachThere has been surprisingly little exploration of the theoretical underpinnings ofthese phenomena and manifestations of homesickness a notable exception being theanalysis of Fisher (eg 1989) In fact from a psychological perspective there are goodreasons to consider homesickness a lsquomini-griefrsquo experience (Stroebe Stroebe amp Schut1993) and to draw on theoretical formulations in the eld of bereavement to derivehypotheses and explain phenomena (cf Stroebe amp Stroebe 1987) It is well establishedthat bereavement has severe mental and physical consequences for some but not allindividuals It is also known that the time course and symptomatology of grief arecomplex and varied (Stroebe Stroebe amp Hansson 1993) If homesickness is understoodas the emotional reaction to (temporary) loss of signicant persons then somemdashperhaps less extrememdashparallels with reactions of grief following losses through deathwould be expected Thus insofar as homesickness is a grief experience one wouldexpect more homesickness the greater the degree of (perceived) separation from home

A recent bereavement-specic model the Dual Process Model of Coping withBereavement (DPM Stroebe amp Schut 1999) provides a framework for integratingsome of the central propositions of two very different approaches cognitive stress(Lazarus ampFolkman 1984) and attachment (Bowlby 1980) theories both of which havebeen applied to the specic stressor of bereavement The DPM identies two typesof stressor that are relevant to the prediction of outcome (in terms of variables suchas health and well-being) namely those that are loss- versus restoration-orientedLoss-orientation refers to concentration on coping with the loss experience itselfand restoration-orientation refers to efforts to adjust to the concurrent changed anddemanding situation The model postulates the necessity of lsquooscillationrsquo betweenthese two tasks identifying the need both to come to terms with loss and to adaptto the different environment

In the case of loss through death loss-orientation would include missing thedeceased person in the case of homesickness it would include missing the absentfamily and friends Elements of attachment theory are clearly relevant here One wouldexpect more homesickness among those whose attachment is less secure (cf Bowlby1980) Following the reasoning of attachment theory as applied to grief experiences(cf Parkes Stevenson-Hinde amp Marris 1991) an insecure style of attachment wouldbe a predictor of homesickness which would lead to health consequences

Restoration-orientation following a death would include developing a new identityand new roles independent of the deceased After relocation it would seem likely thatsomewhat similar adjustments are needed (becoming a lsquocollege studentrsquo member ofnew social housing and academic groups) as are necessary following a loss throughdeath In line with cognitive stress theory (Lazarus amp Folkman 1984) the DPMpredictsthat those individuals experiencing a severe stressor (relocation) who lack the resources

150 Margaret Stroebe et al

(eg social support from home) would be unable to cope with the demands of the newsituation (eg academic performance) and would suffer homesickness and possiblydetrimental health consequences

Thus the DPM predicts that attention to and oscillation between these two tasksloss- and restoration-oriented are necessary for successful adaptation Similarly wewould predict that studentsrsquo well-being is dependent not only on the fact of havingleft home but also on successful adjustment to the new situation Thus the DPM pro-vides a theoretical basis for introducing two classes of variables into a structuralequation model for empirical investigation namely those to do with missing homefriends and so on and adjustment difculties in the new environment InterestinglyArcher et al (1998) recently identied two similar factors on a questionnaire measureof homesickness representing attachment to home and adjustment to the university

Purposes of the empirical studyTaken together what are the implications of this empirical and theoretical overviewfor further investigation of homesickness among students The review of empiricalstudies identies some emergent patterns of homesickness in relationship to personaland situational factors but some conicting results There is a need for replication ofndings and extension of the variables under investigation We have emphasized theneed for further investigation of the origins manifestations and consequences ofhomesickness among vulnerable groups From the outline above it is evident thatstudents leaving home for college belong in this category of vulnerable groups Lackingstill is not only information on predictors and outcomes but also on component partsof homesickness derived from the implementation of standard validation techniquesThus the rst aim of this study was to provide a detailed examination of homesick-ness among college students To this end two studies of homesickness were conductedone in the Netherlands and one in the UK

This two-study investigation enabled some examination of cultural patterns ofhomesickness To our knowledge this research provides the rst cross-cultural exam-ination of homesickness in both male and female students in their own countries Thusthe additional factor of acculturation (ie moving abroad and adjusting to the newcountry) does not occur in the examination of homesickness here Our research isalso unique in that it focuses on two cultures similar in terms of general societal back-ground in that they are both western European but different with respect to a centralattachment theory-related variable noted above namely accessibility to home (as willbe discussed later accessibility is greater in general in the Netherlands) As the abovereview indicated it is important to clarify the relationship between gender and home-sickness among student samples

A further advantage of the two reported studies is that they used the samestandardized homesickness measure In addition in each of the studies the relationshipof homesickness to a number of potentially important variables could be exploredincluding depression self-liking competence or self-esteem and secure or conictedattachment to the family The indices were standardized and thus provide usefulwithin-culture information which was the main focus of the analyses Between-culturecomparisons are to be made cautiously since some of the measures employed differedbetween the two countries (eg one standardized depression scale was used in theUK a different one in the Netherlands) However given the paucity of research todate and the intrinsic interest in comparing these phenomena across the two cultures

Homesickness among students in two cultures 151

as well as within each culture results are presented comparatively In addition anumber of demographic variables are analysed (eg gender age living arrangements)

Following the theoretical predictions outlined above it was expected that home-sickness would be prevalent and associated with distress and poorer mental health inthe student samples of both cultures but that prevalence would be greater in theculture with lower accessibility to home (UK) Following the DPM it was predictedthat missing home and difculty adjusting to the new situation would be antecedentto distress (rather than vice versa) Also that those who missed home more would bethe ones to have more difculty adapting to the new college situation (subscales onthe homesickness questionnaire) Furthermore it was predicted that homesicknesswould be greater among those living further from home irrespective of culture sincefollowing attachment theory the possibility of meeting up with signicant othersshould reduce separation anxiety and distress Again also following attachmenttheory it was predicted that students with greater parental conict1 or insecure attach-ment patterns would feel more homesick Homesickness was expected to occur in bothmale and female students of both cultures but it was predicted that female rateswould be higher (they could feel more comfortable admitting to homesickness) Finallyit was predicted that a longer time period since leaving home (or being at university)would be associated with lower homesickness (ie an adjustment effect) but that somestudents would still suffer homesickness even after being away for a longer period

Method

Study 1 The Netherlands (NL)

Respondents and procedureAll new-intake students entering the Faculty of Social Sciences of Utrecht University were mailed aquestionnaire containing the homesickness scale This took place in January 1995 Four weekslater a reminder was mailed to non-respondents This resulted in 482 completed and returnedquestionnaires (response rate 65) Eighty per cent of the respondents were female which issimilar to the proportion of females entering this faculty The mean age was 196 years (SD= 21)Twenty-nine per cent still lived at home with their parents 489had moved from the parentalhome to lodgings 196had recently moved from one place of lodging to another) and 26hadmoved back from lodgings to the parental home Most of the data analyses were conducted on the489relocated students while those living at home were included in specic analyses for controlpurposes (see below) The Utrecht students are referred to here as the NL sample

Measures

Socio-demographic informationRespondents were asked questions about their age gender previous living situation presentliving situation and the duration of their present living situation

The Utrecht Homesickness Scale (UHS)Designed to assess the extent of homesickness the UHS was developed in the following manner(see van Vliet Stroebe ampSchut 1998) Items were initiallyderived from previous research (Carden ampFeight 1991 Eurelings-Bontekoe et al 1994 Fisher Elder ampPeacock 1990 Fisher Frazer ampMurray

152 Margaret Stroebe et al

1 This prediction was tentative given that the evidence on con icted relationships is mixed (cf Archer 1999 pp 170ndash172)

1984 1985 1986 Fisher amp Hood 1987 1988) Items derived from a pilot study in which 100respondents were asked to give a denition of homesickness supplemented the original list

This preliminary questionnaire was given to 300 social science students who were asked torate the extent to which 51 items were associated with homesickness This resulted in a list of 45items This version of the questionnaire was then given to 117 third year psychology studentsThese students were asked to rate the extent to which theyhad experienced the 45 homesickness-related aspects in the past four weeks This led to a third version of the questionnaire which wasused in the Utrecht Homesickness Project The 45 items assess the extent to which studentsexperienced these various aspects of homesickness in the past 4 weeks (using the answercategories lsquonotrsquo lsquoweakrsquo lsquomoderatersquo lsquostrongrsquo lsquovery strongrsquo)

Aprincipal component analysis (minimum eigenvalue gt 1) with a varimax rotation over the 45items was computed for students who had felt homesick in the past four weeks (N = 151)resulting in nine factors On the basis of a scree plot we used the rst ve factors which wereclearly interpretable and explained 57of the variance These ve factors correspond closely tothe features that according to most authors constitute homesickness missing family missingfriends having adjustment difculties ruminations about home and feeling lonely

Based on these ve factors a reduced set of items (20) was selected to ensure that thedeterminant of the correlation matrix would be gt 0 This is necessary to avoid co-linearity of setsof items Asecond principal component analysis with a varimax rotation over these 20 items andthe restriction that ve factors should be extracted revealed that the items loaded as expected onthe relevant ve factors This last procedure was repeated with all the students (N = 439) andrevealed the same factor loading pattern Within this set of 20 variables the ve factors explained73 of the variance On the basis of the ve-factor structure scales were computed resultingin very acceptable Cronbachrsquos alphas (see Table 1) This set of subscales will be referred to as theUtrecht Homesickness Scale (UHS) Note that if there were missing values lsquolist-wise deletionrsquo wasused

Additional measuresStudents were also asked how often they had experienced homesickness in the past 4 weeks(never rarely sometimes often very often) Adiscriminant analysis whereby the ve UHS scalesare used to predict whether or not students endorsed feeling homesick (never versus rarelysometimes often very often) in the past 4 weeks revealed that 83 were correctly classiedFurthermore there was a strong correlation between the total homesickness score and thefrequency of homesickness in the past 4 weeks (r = 71) which indicates that the UHS scale isclosely related to a direct measure of homesickness even though the scale does not mentionhomesickness specically The Dutch version of the Symptoms Check List (SCL-90 Arrindell ampEttema 1975) was used as an indicator of mental health status of the NL sample Only thedepressive mood subscale (a = 92) and the total health score (a = 97) were used for this part ofthe investigation

Personality factors were measured by means of a Dutch translation of the Golberg Bi-polar Big5 (De Jong van Eck ampvan den Bos 1994 eg Goldberg 1992) Only emotional stability (a = 85)was used for this investigation

A Dutch translation of the Attachment Style Measure (Hazan amp Shaver 1987) was used toassess insecure (that is avoidant and anxiousambivalent see Hazan ampShaver 1987) versus secureattachment This particular scale was chosen because of its conciseness and applicability for usewith these student samples

Statistical analysesComparable to what was done in the pre-test described previously we rst conducted factoranalysis on the homesickness scale which identied the ve subscales A principal componentanalysis (minimum eigenvalue gt 1) with a varimax rotation over the 45 items resulted in ninefactors On the basis of a scree plot we used the rst ve factors which were clearly interpretable

Homesickness among students in two cultures 153

and explained 57of the variance These ve factors correspond closely to the factors identiedin the pre-test on the UHS namely missing family missing friends having adjustment difcultiesruminations about home and feeling lonely

Then we carried out a multivariate analysis of variance (MANOVA) to test differences inhomesickness scores between the UK and NL samples We then computed path analyses (usingLISREL81) in order to examine multiple relationships simultaneously We hypothesized identical

154 Margaret Stroebe et al

Table 1 Utrecht Homesickness Scale factor structure and reliabilities (NL sample)

Factor

(1) (2) (3) (4) (5)

Percentage of explained varianceItem

62 15 9 8 6

Factor loadings

(1) Missing family (a = 90)Missing your parents 89 12 20 06 16Missing your family 87 13 19 09 14Missing home 74 12 21 18 37Feeling missed by your family 74 15 19 15 04

(2) Loneliness (a = 85)Feeling lonely 21 78 22 25 09Feeling unloved 12 75 22 15 07Feeling isolated from the rest of the world 09 72 14 33 22Feeling uprooted 08 64 21 35 22

(3) Missing friends (a = 87)Longing for acquaintances 16 22 83 19 16Searching for familiar faces 19 18 73 23 12Missing people whom you trust and can talk with 25 31 71 07 20Missing your friends 36 14 70 15 14

(4) Adjustment dif culties (a = 88)Finding it dif cult adjusting to a new situation 23 27 15 77 16Feeling uncomfortable in a new situation 15 44 06 71 23Feeling lost in the new situation 17 46 22 69 21Having dif culties in getting used to new customs 03 14 38 68 26

(5) Ruminations about home (a = 80)Having thoughts that an old situation was better thanhere and now 16 15 18 28 81Regretting the decision to leave an old situation 10 21 08 31 79Continuously having thoughts about home 41 08 24 13 68Repeatedly thinking of the past 18 45 28 plusmn 01 52

Total homesickness (a = 94) (20 items)

Note The questions were formulated in the following manner Could you please indicate to what extentyou have experienced the following in the past 4 weeks Missing your parents notndashweakndashmoderatendashstrongndashvery strong N = 439

models for both cultures assuming that the patterns of associations would be similar Weexamined models of the relationships between homesickness and its putative antecedents andconsequences in terms of both overall scores and subscale scores Precise details of these modelsare given in the results section

Several t indices as described by Jaccard and Wan (1996) are appropriate for our investigationof path models To demonstrate the overall degree of correspondence between predicted andobserved covariance matrices the x2 statistic with p gt 05 indicates that a perfect t model doesexist in the population The goodness-of-t index (GFI gt 90) indicates goodness of t Thestandardized root mean square residual (SRMR lt 10) indicates a deviation of less than 10correlation units on average between predicted and observed correlations The root mean squareerror of approximation (RMSEA lt 08) implies an adequate model and lt 05 implies a good modelwith the p value for close t (CFit) being non-signicant (gt 05) CFI gt 90 implies a good t

Study 2 The United Kingdom (UK)

Respondents and procedureDuring the autumn term of 1996 new-intake students at Cardiff University were askedto participate in a student survey Questionnaires were administered in October and Novemberand 280 completed questionnaires were returned representing approximately a 60 responserate Seventy-two per cent of the respondents were female The mean age was 188 years(SD = 21) All respondents had moved from the parental home to lodgings on campus TheCardiff students are referred to here as the UK sample

MeasuresAn English translation of the Utrecht Homesickness Scale (UHS) was used to measure the extent ofhomesickness The reliabilities of the scales were comparable to the NLsample missing the family(a = 85) loneliness (a = 84) missing friends (a = 78) adjustment difculties (a = 84)ruminations about home (a = 86) and the total score for homesickness (a = 93) Againrespondents were also asked to indicate how often they had felt homesick during the past 4weeks The Beck Depression Inventory (BDI Beck Ward Mendelson Mock amp Erbaugh 1961)was used to give a reliable measure of current level of depression (a = 82) The Self-likingSelf-competence Scale was used to assess self-esteem (a = 91) (cf Tafarodi amp Swann 1995) Anadapted version of the Conict Tactics Scale (cf Straus Hamby Boney-McCoy ampSugarman 1996)was used to ascertain type of family background of the UK respondents The Conict Tactics Scaleconsists of six subscales with similar acceptable alpha levels (on average 72) which assess theextent to which parents are perceived as handling conicts with the respondents in certain waysThe items used here reected three factors that is verbal reasoning (three items) verbalaggression (six items) and physical aggression (three items) of either father or mother Selectionof these items from the original much longer scale was thought to be less intrusive and easier forrespondents to answer while still providing a measure of the three factors The students were alsoasked to state the name of their home town On the basis of this information distances could becalculated between their home and their university lodgings

Statistical analysesThe analyses described above for Study 1 apply here too

ResultsThe results section is structured in three parts In the rst section results on the NLsample are reported In the second section the same is done for the UK sample In thelast section a comparison is made between the two samples

Homesickness among students in two cultures 155

Study 1 NL

HomesicknessAlmost 50of the students reported having felt homesick at least some of the time afterhaving started their new education (N = 230) There were no differences betweenmales and females with respect to homesickness intensity The intensity of home-sickness as measured by the UHS subscales (range 1 (not experienced) to 5 (verystrongly)) revealed that on average the highest scores were obtained for missing friends(M = 238 SD = 102) and missing the family (M = 215 SD = 095)2 The otherdimensions scored were adjustment difculties (M = 180 SD = 088) ruminationsabout home (M = 152 SD = 071) and loneliness (M = 179 SD = 087) Theseaveraged to a total mean score of 193 (SD = 071)

Antecedents and consequences of homesicknessThe NL data set offered the possibility of comparisons between students still living athome and those who had relocated3 This provided the opportunity to test whetherrelocation (R) would indeed be associated with depression (D) which in and of itselfwe assumed would be associated with homesickness (H) We must be cautious giventhe cross-sectional design of our studies in assuming mediation between the RndashDndashHvariables Based on Baron and Kenny (1986) a requirement for mediation with this dataset would be that relocation and depression are signicantly associated in other wordsthere must be an effect to be mediated In our data we nd that relocation anddepression are not signicantly associated and therefore a mediation analysis is notappropriate Nevertheless a path analysis is possible if we regress depression on re-location and homesickness and regress homesickness on relocation (see Fig 1) Ourresults show a signicant path from relocation to homesickness (b = 21) a signicantpath from homesickness to depression controlling for relocation (b = 62) and a non-signicant direct path from relocation to depression (b = plusmn 05) Showing that thetwo component paths of the indirect effect are each signicant is not the same asshowing that the indirect effect is signicant Testing for an indirect path (Sobel 1982)revealed that the indirect effect of relocation on depression is signicant (b = 13) Itseems reasonable to conclude then that the data are consistent with an indirect effectof relocation on depression by means of the intervening variable homesickness Whilethere are good reasons to argue that there is an indirect effect of relocation ondepression through homesickness it is evident that this effect though signicant isquite small and not strong enough to produce an overall effect of R on D

The model shown in Fig 1 was also tested using the SCL-90 total score as the healthmeasure instead of using the depression subscale This showed an almost identicalsolution The analysis provides further evidence that the assumed pattern is justiedand that it is not just due to some correspondence between depression and home-sickness The pattern suggests that relocation may be an antecedent for homesicknessbut not for depressive moods or general ill-health and that homesickness may bean important antecedent for ill-health Having found evidence for this relationshipstructural equation models could be tested with respect to the effects of emotionalstability gender and duration of stay on homesickness and depression (see Fig 2) The

156 Margaret Stroebe et al

2 Controlling for age and duration of stay in order to make comparisons with the UK sample possible3 Some students who had not moved also reported being homesick (15) in the past 4 weeks The vast majority of them(90) reported that they experienced homesickness only lsquorarelyrsquo or lsquosometimesrsquo

literature suggests that emotional instability is a good predictor for both homesickness(Eurelings-Bontekoe et al 1996) and depression (cf Watson ampPennebaker 1989) Thusone could assume that emotional stability (the opposite of emotional instability) wouldexplain the variance of depression However we nd that even though reducedhomesickness retains an effect on depression This suggests that irrespective of thelevel of emotional stability homesickness is a precursor for depression Furthermorethe emotionally unstable as expected have higher levels of homesickness and depres-sion What we also nd is that the longer one is away from home the less intense thehomesickness reaction which suggests that adaptation to the new situation takes place

In order to create a more comprehensive picture a model was formulated whichincluded the subscales of the UHS instead of the total homesickness score Theassumptions were that (1) adjustment difculties would be associated with missingfamily and friends ruminations about home and loneliness (2) missing family andfriends in turn would be associated with ruminations about home and loneliness (3)ruminations would also be associated to loneliness and depression and (4) lonelinessin turn to depression Results with respect to this model revealed that in principle theassumptions were correct if two modications were incorporated These wereestablishing a link between missing family and friends and removing the path frommissing family to loneliness The choice was made to let the errors of missing friendsand family correlate because no prior hypotheses were made as to a causal relation-ship between both features and to delete the non-signicant path4 Figure 3 indicateshow the homesickness subscales are related and how these are linked to the measuredantecedents and consequences To enhance the visual impact of the model the thick-ness of paths corresponds to the size of the path coefcients

Figure 3 suggests that to become depressed one has to experience adjustment

Homesickness among students in two cultures 157

Figure 1 Model of relocation homesickness and depression for NL students

4 For the sake of comparison with the forthcoming UK model the non-signicant paths between missing friends and lonelinessgender and adjustment dif culties and duration of stay and missing family were not removed from the model

difculties and consequently loneliness and to ruminate about home If the relocatedstudents experience severe adjustment difculties these do seem to go hand in handwith missing family and friends ruminations about home and loneliness Specicallyloneliness is most strongly associated with depression Furthermore the analysessuggest that the best predictor for adjustment difculties in this study is emotionalstability (b = plusmn 41) Emotional stability also has a strong relationship to depression(b = plusmn 45) and to a lesser extent to missing friends (b = plusmn 20) and loneliness(b = plusmn 15) The slight association of duration of stay with total homesickness dis-appeared when its relationship was tested on the ve separate homesickness subscales

Study 2 UK

HomesicknessMore than 80of the students reported having felt homesick at least some of the timeafter having started their new education (N = 280) The intensity of homesicknessas measured by the UHS subscales (likewise controlling for age and the number of daysaway from home) revealed that on average the highest scores in the UK were obtainedfor missing family (M = 293 SD = 097) and missing friends (M = 272 SD = 094)The other dimensions scored were adjustment difculties (M = 205 SD = 084)ruminations about home (M = 166 SD = 086) and loneliness (M = 195SD = 092) These averaged to a total mean score of 226 (SD = 073)

Antecedents and consequences of homesicknessAlthough the hypothesis could not be tested with the UKsample it seems reasonable toassume based on the NLdata that relocation would be associated with homesicknesswhich in turn would be associated with depression and not vice versa (of course similarcautions about mediating variables to those expressed for Study 1 are in order here too)

158 Margaret Stroebe et al

Figure 2 Model of NL sample

Accepting this premise and despite the differences in the depression and personalityscales used in the UK and the NL what one would expect is comparable patterns ofcorrelations since these should be the same irrespective of the particular instrumentused We formulated a structural equation model analogous to the one used in the NLstudy examining the effects of self-esteem gender and duration of stay based onhomesickness and depression for the UK students (see Fig 4) This analysis revealedthat analogous to the NL data (Fig 2) homesickness is associated with depressionand that low self-esteem can be considered a risk factor for becoming homesick anddepressed To see whether or not homesickness is the precursor for self-esteem insteadof self-esteem being the precursor as hypothesized the alternative model was testedResults did not show a signicant effect of homesickness on self-esteem Thus low self-esteem like low emotional stability can rather be considered a vulnerability factorfor homesickness Again we see that after controlling for self-esteem the relationshipbetween homesickness and depression remains signicant This suggests that home-sickness is associated with depression irrespective of the level of self-esteem while ifone has low self-esteem the risk of becoming homesick and depressed is greater TheUK results also reveal the slight negative effect of duration of stay the longer studentslive in a new situation the lower the level of homesickness In contrast to the NLdata the UK data do reveal a gender difference Female students (M = 241 SD = 074)report higher levels of homesickness (F(278 1) = 1585 p lt 001) than male students(M = 204 SD = 060)

Analogous to the NL study we formulated a more detailed model including theUHS subscales (Fig 5) This revealed a similar pattern of correlations although thereare some obvious differences Adjustment difculties are assumed to be the crux of

Homesickness among students in two cultures 159

Figure 3 Detailed model of NL sample

the model and the data conrm this From adjustment difculties there are two pathsthat are connected indirectly to depression one via loneliness and the second viaruminations about home Self-esteem (a personality trait which was assumed to havesome commonality with emotional stability) has a similar relationship to adjustmentdifculties and depression in the UK sample as emotional stability in the NLsample Asdemonstrated previously with respect to the total homesickness score when analysedat the subscale level females in the UK study report having more adjustment difcultiesthan males This would indicate that the difference in homesickness mainly lies infemales reporting more adjustment difculties

Study 1 and Study 2 comparisonThere were a number of signicant differences between the two national samplesDutch students were older (M = 192 years) than British students (M = 188 yearst (478) = 295 p lt 01) and Dutch students (M = 1988 days) had spent a longer numberof days away from home than British students (M = 501 days t (224) = 1098 p lt 001)Furthermore there were proportionately more females in the NLsample than the UKsample (82versus 72) The implications of these differences are discussed later

HomesicknessAfter controlling for age and duration of relocation it was still found that the NLand UKstudents differed considerably with respect to homesickness with the UK studentsreporting more intense homesickness Univariate differences were found for three ofthe ve subscales (missing family missing friends adjustment difculties) and the totalhomesickness score Adjusted means for the subscales and for total homesickness forthe two samples are presented in Table 2

Furthermore when asked in the separate question about general feelings of home-sickness the UK students more frequently endorsed having felt homesick (see Table 3)

160 Margaret Stroebe et al

Figure 4 Model of UK sample

These analyses indicate not only that the UK students are more often homesick butthat they are so to a greater extent

Antecedents and consequences of homesicknessThe global models (Figs 2 and 4) demonstrate that homesickness is related to depres-sion in both cultures and that low self-esteem or emotional instability are linked tohigher levels of self-reported homesickness and depression More possibly howeverthe longer the stay in the new situation the less homesick and consequently lessdepressed one tends to be However the evidence also suggests that there may becultural differences with respect to gender differences in homesickness

The more detailed models (Figs 3 and 5) have the same basic structure althoughthere are some differences In the UK females are more likely to experience adjustmentdifculties than are males (b = 15) and the longer a UK student has experienced thenew situation the less he or she is inclined to miss the family (b = plusmn 12) effects whichare non-signicant in the NLstudy To enhance comparisons between the cultures thegeographical distance to the parental home (data not available in the NL sample) wasnot included here in the UKmodel However if included the pattern remained the sameand revealed that distance from home was only reliably associated with missing thefamily (b = 19)

Attachment style and family backgroundWe conducted further analyses to explore the association between attachment style andfamily background variables with homesickness In the NLstudy the effects of insecureattachment on relocation the homesickness features depression and emotional stability

Homesickness among students in two cultures 161

Figure 5 Detailed model of UK sample

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 2: res_6

highly stressful experience (see Fisher 1989 van Tilburg 1998 van Tilburg Vinger-hoets amp van Heck 1996 for recent reviews of the literature) Homesickness is a morewidespread phenomenon than is usually believed As many as 50ndash75 of the generalpopulation have experienced homesickness at least once in their life (Fisher 1989)Furthermore 10ndash15 of the homesick have experienced it to such an extent that itinterferes with their daily activities (Fisher 1989) While there have been some previousstudies on homesickness among college students (eg Archer Ireland Amos Broad ampCurrid 1998 Brewin Furnham amp Howes 1989 Carden amp Feicht 1991 Fisher ampHood1987 1988 van Tilburg 1998) these only yield information on a limited number ofpsychological variables and employ a variety of homesickness measures some of themnon-standardized In fact even a precise denition of homesickness is hard to nd in theliterature Generally it is understood that homesick children or students miss theirparents and family friends and other familiar persons their familiar surroundings andhome comforts and they feel extremely insecure (cf Baier amp Welch 1992 Eurelings-Bontekoe Vingerhoets amp Fontijn 1994 Fisher 1989) With respect to student home-sickness some dimensions include lsquodisliking the universityrsquo as well as lsquoattachment to thehomersquo (cf Archer et al 1998) These dimensions reect the varied descriptions that canbe found which range from lsquodepression as a result of absence from homersquo (The OxfordDictionary) to lsquolonging for home and family while absent from themrsquo (WebsterrsquosDictionary)

Empirical overviewWhat is the state of knowledge about prevalence predisposition for and impact ofhomesickness among students thus far Wide differences in the prevalence of home-sickness have been reported although it is generally considered by researchers toaffect people from all cultures and all age groups (van Tilburg et al 1996) Research hasindicated that homesickness occurs in large numbers of students who leave homeAmong student populations in Great Britain for example between 30 and 60 ofboth men and women reported being affected by homesickness during their rst yearat university For example Brewin et al (1989) found that 40of the students in theirsample reported feeling homesick 20 were unsure and 40 reported not feelinghomesick By comparison also among students Fisher and Hood (1987 1988) found30ndash35 to be slightly to very homesick while 65ndash70 were not On a dichotomoushomesickness measure Fisher Murray and Frazer (1985) found that 60 of the rstyear students in their sample reported being homesick Vastly different percentageshave been reported for other countries Carden and Feicht (1991) for example foundthat 19 of American and 77 of Turkish students attending universities in their owncountries could be classied as being homesick in their rst year (based on a cut-offpoint of 1 standard deviation above the mean rating for the group on a homesicknessquestionnaire) The diversity in percentages both within and between countries islikely to be due at least in part to differences in the scales used and in the duration oftime since leaving home Indeed it is sometimes difcult to compare results acrosssamples particularly with respect to prevalence of homesickness because investigatorsuse their own specic homesickness scales (cf Archer et al 1998 Eurelings-BontekoeTolsma Verschuur amp Vingerhoets 1996 Eurelings-Bontekoe Verschuur Koudstaal ampvan der Sar 1995 van Tilburg Vingerhoets amp van Heck 1997)

The psychological literature on predispositional factors risk factors and mediatorsof homesickness has mainly studied personal characteristics which predispose persons

148 Margaret Stroebe et al

to become homesick and characteristics of the environment or situation that areassociated with greater intensities of it Highly relevant for student populations is thequestion of gender differences as a risk factor for homesickness While lay assump-tions might lead to expectations of greater female homesickness on leaving homefor college the literature does not support this Brewin et al (1989) reported thathomesickness was similarly prevalent among male and female students a ndingconrmed by Fisher for a variety of samples (eg Fisher et al 1985 Fisher amp Hood1987 1988) Ways of coping however differed between the genders with womenseeking more social support than was the case for men Similarly Archer et al (1998)found that female students showed higher levels of intrusive thinking about home-sickness mediated by their higher scores on the attachment factor of a homesick-ness questionnaire whereas there was no gender difference in avoidant responsesto homesickness There are gender differences in coping with and consequences ofother negative life-events notablymdashin this contextmdashbereavement (eg health detri-ments are relatively more extreme for widowers than for widows see Stroebe ampStroebe 1987) Thus further examination of patterns of gender differences in home-sickness would seem timely

Van Tilburg et al (1996) have recently emphasized the need for further examinationof cultural differences in the prevalence and experience of homesickness noting thatthis had not been studied systematically Carden and Feicht (1991) studied home-sickness among Turkish and American students in their own countries but limitedinvestigation to females Hojat and Herman (1985) studied a sample of Iranian andFilipino individuals but these were all resident in the USA and the sample wascomposed of physicians Ward and Kennedy (1993) examined homesickness amongstudents in different cultures but these were all from New Zealand residing in 23different countries of the world Lu (1990) also examined students who had moved tostudy in a different culture but again the study was conned to one cultural groupnamely Chinese students all of whom had moved to a university in the UK

Particularly important too in the context of our investigation are ndings to dowith personal vulnerability factors Both Carden and Feicht (1991) and Brewin et al(1989) reported associations between dependency and homesickness In the Brewinet al (1989) study lsquodependency on othersrsquo was a predictor of homesickness in theCarden and Feicht (1991) study dependency on family and parents was found to becharacteristic of homesick but not non-homesick students Personality and dispositionalcorrelates of homesickness have also been identied among student samples Fisher(1989) found that introversion depression and obsession were related to homesicknessbut that self-esteem was no different among homesick versus non-homesick studentsIn other samples lower self-esteem has indeed been reported for homesick samples(Eurelings-Bontekoe et al 1994 Hojat amp Herman 1985) Furthermore Eurelings-Bontekoe et al (1994) also found higher levels of introversion rigidity and negativismand lower levels of dominance among homesick conscripts Rigidity proved to be thebest predictor for homesickness These patterns are not however always conrmedfor other samples (cf van Tilburg et al 1996)

Turning from personal to situational factors lsquogeographic distancersquo is also importantfor the current investigation Among students Fisher et al (1986) found homesicknessto be higher for those more distant from home (although notably distance fromhome did not make a difference among boarding school children) Needing furtherinvestigation are related factors such as psychological distance and possibilities forcommunication with home

Homesickness among students in two cultures 149

Finally with respect to the consequences or outcomes associated with homesick-ness among students studies have also found a relationship between homesicknessand a greater number of cognitive failures poor concentration handing in work latedecrements in work quality and higher scores on anxiety and depression measures(Burt 1993) This has recently been conrmed in a meta-analysis of four studiesconducted by Archer et al (1998) As Brewin et al (1989) emphasized such ndingssuggest that homesickness is a potentially important phenomenon that may exercisea considerable inuence on academic performance at least in the short term

A theoretical approachThere has been surprisingly little exploration of the theoretical underpinnings ofthese phenomena and manifestations of homesickness a notable exception being theanalysis of Fisher (eg 1989) In fact from a psychological perspective there are goodreasons to consider homesickness a lsquomini-griefrsquo experience (Stroebe Stroebe amp Schut1993) and to draw on theoretical formulations in the eld of bereavement to derivehypotheses and explain phenomena (cf Stroebe amp Stroebe 1987) It is well establishedthat bereavement has severe mental and physical consequences for some but not allindividuals It is also known that the time course and symptomatology of grief arecomplex and varied (Stroebe Stroebe amp Hansson 1993) If homesickness is understoodas the emotional reaction to (temporary) loss of signicant persons then somemdashperhaps less extrememdashparallels with reactions of grief following losses through deathwould be expected Thus insofar as homesickness is a grief experience one wouldexpect more homesickness the greater the degree of (perceived) separation from home

A recent bereavement-specic model the Dual Process Model of Coping withBereavement (DPM Stroebe amp Schut 1999) provides a framework for integratingsome of the central propositions of two very different approaches cognitive stress(Lazarus ampFolkman 1984) and attachment (Bowlby 1980) theories both of which havebeen applied to the specic stressor of bereavement The DPM identies two typesof stressor that are relevant to the prediction of outcome (in terms of variables suchas health and well-being) namely those that are loss- versus restoration-orientedLoss-orientation refers to concentration on coping with the loss experience itselfand restoration-orientation refers to efforts to adjust to the concurrent changed anddemanding situation The model postulates the necessity of lsquooscillationrsquo betweenthese two tasks identifying the need both to come to terms with loss and to adaptto the different environment

In the case of loss through death loss-orientation would include missing thedeceased person in the case of homesickness it would include missing the absentfamily and friends Elements of attachment theory are clearly relevant here One wouldexpect more homesickness among those whose attachment is less secure (cf Bowlby1980) Following the reasoning of attachment theory as applied to grief experiences(cf Parkes Stevenson-Hinde amp Marris 1991) an insecure style of attachment wouldbe a predictor of homesickness which would lead to health consequences

Restoration-orientation following a death would include developing a new identityand new roles independent of the deceased After relocation it would seem likely thatsomewhat similar adjustments are needed (becoming a lsquocollege studentrsquo member ofnew social housing and academic groups) as are necessary following a loss throughdeath In line with cognitive stress theory (Lazarus amp Folkman 1984) the DPMpredictsthat those individuals experiencing a severe stressor (relocation) who lack the resources

150 Margaret Stroebe et al

(eg social support from home) would be unable to cope with the demands of the newsituation (eg academic performance) and would suffer homesickness and possiblydetrimental health consequences

Thus the DPM predicts that attention to and oscillation between these two tasksloss- and restoration-oriented are necessary for successful adaptation Similarly wewould predict that studentsrsquo well-being is dependent not only on the fact of havingleft home but also on successful adjustment to the new situation Thus the DPM pro-vides a theoretical basis for introducing two classes of variables into a structuralequation model for empirical investigation namely those to do with missing homefriends and so on and adjustment difculties in the new environment InterestinglyArcher et al (1998) recently identied two similar factors on a questionnaire measureof homesickness representing attachment to home and adjustment to the university

Purposes of the empirical studyTaken together what are the implications of this empirical and theoretical overviewfor further investigation of homesickness among students The review of empiricalstudies identies some emergent patterns of homesickness in relationship to personaland situational factors but some conicting results There is a need for replication ofndings and extension of the variables under investigation We have emphasized theneed for further investigation of the origins manifestations and consequences ofhomesickness among vulnerable groups From the outline above it is evident thatstudents leaving home for college belong in this category of vulnerable groups Lackingstill is not only information on predictors and outcomes but also on component partsof homesickness derived from the implementation of standard validation techniquesThus the rst aim of this study was to provide a detailed examination of homesick-ness among college students To this end two studies of homesickness were conductedone in the Netherlands and one in the UK

This two-study investigation enabled some examination of cultural patterns ofhomesickness To our knowledge this research provides the rst cross-cultural exam-ination of homesickness in both male and female students in their own countries Thusthe additional factor of acculturation (ie moving abroad and adjusting to the newcountry) does not occur in the examination of homesickness here Our research isalso unique in that it focuses on two cultures similar in terms of general societal back-ground in that they are both western European but different with respect to a centralattachment theory-related variable noted above namely accessibility to home (as willbe discussed later accessibility is greater in general in the Netherlands) As the abovereview indicated it is important to clarify the relationship between gender and home-sickness among student samples

A further advantage of the two reported studies is that they used the samestandardized homesickness measure In addition in each of the studies the relationshipof homesickness to a number of potentially important variables could be exploredincluding depression self-liking competence or self-esteem and secure or conictedattachment to the family The indices were standardized and thus provide usefulwithin-culture information which was the main focus of the analyses Between-culturecomparisons are to be made cautiously since some of the measures employed differedbetween the two countries (eg one standardized depression scale was used in theUK a different one in the Netherlands) However given the paucity of research todate and the intrinsic interest in comparing these phenomena across the two cultures

Homesickness among students in two cultures 151

as well as within each culture results are presented comparatively In addition anumber of demographic variables are analysed (eg gender age living arrangements)

Following the theoretical predictions outlined above it was expected that home-sickness would be prevalent and associated with distress and poorer mental health inthe student samples of both cultures but that prevalence would be greater in theculture with lower accessibility to home (UK) Following the DPM it was predictedthat missing home and difculty adjusting to the new situation would be antecedentto distress (rather than vice versa) Also that those who missed home more would bethe ones to have more difculty adapting to the new college situation (subscales onthe homesickness questionnaire) Furthermore it was predicted that homesicknesswould be greater among those living further from home irrespective of culture sincefollowing attachment theory the possibility of meeting up with signicant othersshould reduce separation anxiety and distress Again also following attachmenttheory it was predicted that students with greater parental conict1 or insecure attach-ment patterns would feel more homesick Homesickness was expected to occur in bothmale and female students of both cultures but it was predicted that female rateswould be higher (they could feel more comfortable admitting to homesickness) Finallyit was predicted that a longer time period since leaving home (or being at university)would be associated with lower homesickness (ie an adjustment effect) but that somestudents would still suffer homesickness even after being away for a longer period

Method

Study 1 The Netherlands (NL)

Respondents and procedureAll new-intake students entering the Faculty of Social Sciences of Utrecht University were mailed aquestionnaire containing the homesickness scale This took place in January 1995 Four weekslater a reminder was mailed to non-respondents This resulted in 482 completed and returnedquestionnaires (response rate 65) Eighty per cent of the respondents were female which issimilar to the proportion of females entering this faculty The mean age was 196 years (SD= 21)Twenty-nine per cent still lived at home with their parents 489had moved from the parentalhome to lodgings 196had recently moved from one place of lodging to another) and 26hadmoved back from lodgings to the parental home Most of the data analyses were conducted on the489relocated students while those living at home were included in specic analyses for controlpurposes (see below) The Utrecht students are referred to here as the NL sample

Measures

Socio-demographic informationRespondents were asked questions about their age gender previous living situation presentliving situation and the duration of their present living situation

The Utrecht Homesickness Scale (UHS)Designed to assess the extent of homesickness the UHS was developed in the following manner(see van Vliet Stroebe ampSchut 1998) Items were initiallyderived from previous research (Carden ampFeight 1991 Eurelings-Bontekoe et al 1994 Fisher Elder ampPeacock 1990 Fisher Frazer ampMurray

152 Margaret Stroebe et al

1 This prediction was tentative given that the evidence on con icted relationships is mixed (cf Archer 1999 pp 170ndash172)

1984 1985 1986 Fisher amp Hood 1987 1988) Items derived from a pilot study in which 100respondents were asked to give a denition of homesickness supplemented the original list

This preliminary questionnaire was given to 300 social science students who were asked torate the extent to which 51 items were associated with homesickness This resulted in a list of 45items This version of the questionnaire was then given to 117 third year psychology studentsThese students were asked to rate the extent to which theyhad experienced the 45 homesickness-related aspects in the past four weeks This led to a third version of the questionnaire which wasused in the Utrecht Homesickness Project The 45 items assess the extent to which studentsexperienced these various aspects of homesickness in the past 4 weeks (using the answercategories lsquonotrsquo lsquoweakrsquo lsquomoderatersquo lsquostrongrsquo lsquovery strongrsquo)

Aprincipal component analysis (minimum eigenvalue gt 1) with a varimax rotation over the 45items was computed for students who had felt homesick in the past four weeks (N = 151)resulting in nine factors On the basis of a scree plot we used the rst ve factors which wereclearly interpretable and explained 57of the variance These ve factors correspond closely tothe features that according to most authors constitute homesickness missing family missingfriends having adjustment difculties ruminations about home and feeling lonely

Based on these ve factors a reduced set of items (20) was selected to ensure that thedeterminant of the correlation matrix would be gt 0 This is necessary to avoid co-linearity of setsof items Asecond principal component analysis with a varimax rotation over these 20 items andthe restriction that ve factors should be extracted revealed that the items loaded as expected onthe relevant ve factors This last procedure was repeated with all the students (N = 439) andrevealed the same factor loading pattern Within this set of 20 variables the ve factors explained73 of the variance On the basis of the ve-factor structure scales were computed resultingin very acceptable Cronbachrsquos alphas (see Table 1) This set of subscales will be referred to as theUtrecht Homesickness Scale (UHS) Note that if there were missing values lsquolist-wise deletionrsquo wasused

Additional measuresStudents were also asked how often they had experienced homesickness in the past 4 weeks(never rarely sometimes often very often) Adiscriminant analysis whereby the ve UHS scalesare used to predict whether or not students endorsed feeling homesick (never versus rarelysometimes often very often) in the past 4 weeks revealed that 83 were correctly classiedFurthermore there was a strong correlation between the total homesickness score and thefrequency of homesickness in the past 4 weeks (r = 71) which indicates that the UHS scale isclosely related to a direct measure of homesickness even though the scale does not mentionhomesickness specically The Dutch version of the Symptoms Check List (SCL-90 Arrindell ampEttema 1975) was used as an indicator of mental health status of the NL sample Only thedepressive mood subscale (a = 92) and the total health score (a = 97) were used for this part ofthe investigation

Personality factors were measured by means of a Dutch translation of the Golberg Bi-polar Big5 (De Jong van Eck ampvan den Bos 1994 eg Goldberg 1992) Only emotional stability (a = 85)was used for this investigation

A Dutch translation of the Attachment Style Measure (Hazan amp Shaver 1987) was used toassess insecure (that is avoidant and anxiousambivalent see Hazan ampShaver 1987) versus secureattachment This particular scale was chosen because of its conciseness and applicability for usewith these student samples

Statistical analysesComparable to what was done in the pre-test described previously we rst conducted factoranalysis on the homesickness scale which identied the ve subscales A principal componentanalysis (minimum eigenvalue gt 1) with a varimax rotation over the 45 items resulted in ninefactors On the basis of a scree plot we used the rst ve factors which were clearly interpretable

Homesickness among students in two cultures 153

and explained 57of the variance These ve factors correspond closely to the factors identiedin the pre-test on the UHS namely missing family missing friends having adjustment difcultiesruminations about home and feeling lonely

Then we carried out a multivariate analysis of variance (MANOVA) to test differences inhomesickness scores between the UK and NL samples We then computed path analyses (usingLISREL81) in order to examine multiple relationships simultaneously We hypothesized identical

154 Margaret Stroebe et al

Table 1 Utrecht Homesickness Scale factor structure and reliabilities (NL sample)

Factor

(1) (2) (3) (4) (5)

Percentage of explained varianceItem

62 15 9 8 6

Factor loadings

(1) Missing family (a = 90)Missing your parents 89 12 20 06 16Missing your family 87 13 19 09 14Missing home 74 12 21 18 37Feeling missed by your family 74 15 19 15 04

(2) Loneliness (a = 85)Feeling lonely 21 78 22 25 09Feeling unloved 12 75 22 15 07Feeling isolated from the rest of the world 09 72 14 33 22Feeling uprooted 08 64 21 35 22

(3) Missing friends (a = 87)Longing for acquaintances 16 22 83 19 16Searching for familiar faces 19 18 73 23 12Missing people whom you trust and can talk with 25 31 71 07 20Missing your friends 36 14 70 15 14

(4) Adjustment dif culties (a = 88)Finding it dif cult adjusting to a new situation 23 27 15 77 16Feeling uncomfortable in a new situation 15 44 06 71 23Feeling lost in the new situation 17 46 22 69 21Having dif culties in getting used to new customs 03 14 38 68 26

(5) Ruminations about home (a = 80)Having thoughts that an old situation was better thanhere and now 16 15 18 28 81Regretting the decision to leave an old situation 10 21 08 31 79Continuously having thoughts about home 41 08 24 13 68Repeatedly thinking of the past 18 45 28 plusmn 01 52

Total homesickness (a = 94) (20 items)

Note The questions were formulated in the following manner Could you please indicate to what extentyou have experienced the following in the past 4 weeks Missing your parents notndashweakndashmoderatendashstrongndashvery strong N = 439

models for both cultures assuming that the patterns of associations would be similar Weexamined models of the relationships between homesickness and its putative antecedents andconsequences in terms of both overall scores and subscale scores Precise details of these modelsare given in the results section

Several t indices as described by Jaccard and Wan (1996) are appropriate for our investigationof path models To demonstrate the overall degree of correspondence between predicted andobserved covariance matrices the x2 statistic with p gt 05 indicates that a perfect t model doesexist in the population The goodness-of-t index (GFI gt 90) indicates goodness of t Thestandardized root mean square residual (SRMR lt 10) indicates a deviation of less than 10correlation units on average between predicted and observed correlations The root mean squareerror of approximation (RMSEA lt 08) implies an adequate model and lt 05 implies a good modelwith the p value for close t (CFit) being non-signicant (gt 05) CFI gt 90 implies a good t

Study 2 The United Kingdom (UK)

Respondents and procedureDuring the autumn term of 1996 new-intake students at Cardiff University were askedto participate in a student survey Questionnaires were administered in October and Novemberand 280 completed questionnaires were returned representing approximately a 60 responserate Seventy-two per cent of the respondents were female The mean age was 188 years(SD = 21) All respondents had moved from the parental home to lodgings on campus TheCardiff students are referred to here as the UK sample

MeasuresAn English translation of the Utrecht Homesickness Scale (UHS) was used to measure the extent ofhomesickness The reliabilities of the scales were comparable to the NLsample missing the family(a = 85) loneliness (a = 84) missing friends (a = 78) adjustment difculties (a = 84)ruminations about home (a = 86) and the total score for homesickness (a = 93) Againrespondents were also asked to indicate how often they had felt homesick during the past 4weeks The Beck Depression Inventory (BDI Beck Ward Mendelson Mock amp Erbaugh 1961)was used to give a reliable measure of current level of depression (a = 82) The Self-likingSelf-competence Scale was used to assess self-esteem (a = 91) (cf Tafarodi amp Swann 1995) Anadapted version of the Conict Tactics Scale (cf Straus Hamby Boney-McCoy ampSugarman 1996)was used to ascertain type of family background of the UK respondents The Conict Tactics Scaleconsists of six subscales with similar acceptable alpha levels (on average 72) which assess theextent to which parents are perceived as handling conicts with the respondents in certain waysThe items used here reected three factors that is verbal reasoning (three items) verbalaggression (six items) and physical aggression (three items) of either father or mother Selectionof these items from the original much longer scale was thought to be less intrusive and easier forrespondents to answer while still providing a measure of the three factors The students were alsoasked to state the name of their home town On the basis of this information distances could becalculated between their home and their university lodgings

Statistical analysesThe analyses described above for Study 1 apply here too

ResultsThe results section is structured in three parts In the rst section results on the NLsample are reported In the second section the same is done for the UK sample In thelast section a comparison is made between the two samples

Homesickness among students in two cultures 155

Study 1 NL

HomesicknessAlmost 50of the students reported having felt homesick at least some of the time afterhaving started their new education (N = 230) There were no differences betweenmales and females with respect to homesickness intensity The intensity of home-sickness as measured by the UHS subscales (range 1 (not experienced) to 5 (verystrongly)) revealed that on average the highest scores were obtained for missing friends(M = 238 SD = 102) and missing the family (M = 215 SD = 095)2 The otherdimensions scored were adjustment difculties (M = 180 SD = 088) ruminationsabout home (M = 152 SD = 071) and loneliness (M = 179 SD = 087) Theseaveraged to a total mean score of 193 (SD = 071)

Antecedents and consequences of homesicknessThe NL data set offered the possibility of comparisons between students still living athome and those who had relocated3 This provided the opportunity to test whetherrelocation (R) would indeed be associated with depression (D) which in and of itselfwe assumed would be associated with homesickness (H) We must be cautious giventhe cross-sectional design of our studies in assuming mediation between the RndashDndashHvariables Based on Baron and Kenny (1986) a requirement for mediation with this dataset would be that relocation and depression are signicantly associated in other wordsthere must be an effect to be mediated In our data we nd that relocation anddepression are not signicantly associated and therefore a mediation analysis is notappropriate Nevertheless a path analysis is possible if we regress depression on re-location and homesickness and regress homesickness on relocation (see Fig 1) Ourresults show a signicant path from relocation to homesickness (b = 21) a signicantpath from homesickness to depression controlling for relocation (b = 62) and a non-signicant direct path from relocation to depression (b = plusmn 05) Showing that thetwo component paths of the indirect effect are each signicant is not the same asshowing that the indirect effect is signicant Testing for an indirect path (Sobel 1982)revealed that the indirect effect of relocation on depression is signicant (b = 13) Itseems reasonable to conclude then that the data are consistent with an indirect effectof relocation on depression by means of the intervening variable homesickness Whilethere are good reasons to argue that there is an indirect effect of relocation ondepression through homesickness it is evident that this effect though signicant isquite small and not strong enough to produce an overall effect of R on D

The model shown in Fig 1 was also tested using the SCL-90 total score as the healthmeasure instead of using the depression subscale This showed an almost identicalsolution The analysis provides further evidence that the assumed pattern is justiedand that it is not just due to some correspondence between depression and home-sickness The pattern suggests that relocation may be an antecedent for homesicknessbut not for depressive moods or general ill-health and that homesickness may bean important antecedent for ill-health Having found evidence for this relationshipstructural equation models could be tested with respect to the effects of emotionalstability gender and duration of stay on homesickness and depression (see Fig 2) The

156 Margaret Stroebe et al

2 Controlling for age and duration of stay in order to make comparisons with the UK sample possible3 Some students who had not moved also reported being homesick (15) in the past 4 weeks The vast majority of them(90) reported that they experienced homesickness only lsquorarelyrsquo or lsquosometimesrsquo

literature suggests that emotional instability is a good predictor for both homesickness(Eurelings-Bontekoe et al 1996) and depression (cf Watson ampPennebaker 1989) Thusone could assume that emotional stability (the opposite of emotional instability) wouldexplain the variance of depression However we nd that even though reducedhomesickness retains an effect on depression This suggests that irrespective of thelevel of emotional stability homesickness is a precursor for depression Furthermorethe emotionally unstable as expected have higher levels of homesickness and depres-sion What we also nd is that the longer one is away from home the less intense thehomesickness reaction which suggests that adaptation to the new situation takes place

In order to create a more comprehensive picture a model was formulated whichincluded the subscales of the UHS instead of the total homesickness score Theassumptions were that (1) adjustment difculties would be associated with missingfamily and friends ruminations about home and loneliness (2) missing family andfriends in turn would be associated with ruminations about home and loneliness (3)ruminations would also be associated to loneliness and depression and (4) lonelinessin turn to depression Results with respect to this model revealed that in principle theassumptions were correct if two modications were incorporated These wereestablishing a link between missing family and friends and removing the path frommissing family to loneliness The choice was made to let the errors of missing friendsand family correlate because no prior hypotheses were made as to a causal relation-ship between both features and to delete the non-signicant path4 Figure 3 indicateshow the homesickness subscales are related and how these are linked to the measuredantecedents and consequences To enhance the visual impact of the model the thick-ness of paths corresponds to the size of the path coefcients

Figure 3 suggests that to become depressed one has to experience adjustment

Homesickness among students in two cultures 157

Figure 1 Model of relocation homesickness and depression for NL students

4 For the sake of comparison with the forthcoming UK model the non-signicant paths between missing friends and lonelinessgender and adjustment dif culties and duration of stay and missing family were not removed from the model

difculties and consequently loneliness and to ruminate about home If the relocatedstudents experience severe adjustment difculties these do seem to go hand in handwith missing family and friends ruminations about home and loneliness Specicallyloneliness is most strongly associated with depression Furthermore the analysessuggest that the best predictor for adjustment difculties in this study is emotionalstability (b = plusmn 41) Emotional stability also has a strong relationship to depression(b = plusmn 45) and to a lesser extent to missing friends (b = plusmn 20) and loneliness(b = plusmn 15) The slight association of duration of stay with total homesickness dis-appeared when its relationship was tested on the ve separate homesickness subscales

Study 2 UK

HomesicknessMore than 80of the students reported having felt homesick at least some of the timeafter having started their new education (N = 280) The intensity of homesicknessas measured by the UHS subscales (likewise controlling for age and the number of daysaway from home) revealed that on average the highest scores in the UK were obtainedfor missing family (M = 293 SD = 097) and missing friends (M = 272 SD = 094)The other dimensions scored were adjustment difculties (M = 205 SD = 084)ruminations about home (M = 166 SD = 086) and loneliness (M = 195SD = 092) These averaged to a total mean score of 226 (SD = 073)

Antecedents and consequences of homesicknessAlthough the hypothesis could not be tested with the UKsample it seems reasonable toassume based on the NLdata that relocation would be associated with homesicknesswhich in turn would be associated with depression and not vice versa (of course similarcautions about mediating variables to those expressed for Study 1 are in order here too)

158 Margaret Stroebe et al

Figure 2 Model of NL sample

Accepting this premise and despite the differences in the depression and personalityscales used in the UK and the NL what one would expect is comparable patterns ofcorrelations since these should be the same irrespective of the particular instrumentused We formulated a structural equation model analogous to the one used in the NLstudy examining the effects of self-esteem gender and duration of stay based onhomesickness and depression for the UK students (see Fig 4) This analysis revealedthat analogous to the NL data (Fig 2) homesickness is associated with depressionand that low self-esteem can be considered a risk factor for becoming homesick anddepressed To see whether or not homesickness is the precursor for self-esteem insteadof self-esteem being the precursor as hypothesized the alternative model was testedResults did not show a signicant effect of homesickness on self-esteem Thus low self-esteem like low emotional stability can rather be considered a vulnerability factorfor homesickness Again we see that after controlling for self-esteem the relationshipbetween homesickness and depression remains signicant This suggests that home-sickness is associated with depression irrespective of the level of self-esteem while ifone has low self-esteem the risk of becoming homesick and depressed is greater TheUK results also reveal the slight negative effect of duration of stay the longer studentslive in a new situation the lower the level of homesickness In contrast to the NLdata the UK data do reveal a gender difference Female students (M = 241 SD = 074)report higher levels of homesickness (F(278 1) = 1585 p lt 001) than male students(M = 204 SD = 060)

Analogous to the NL study we formulated a more detailed model including theUHS subscales (Fig 5) This revealed a similar pattern of correlations although thereare some obvious differences Adjustment difculties are assumed to be the crux of

Homesickness among students in two cultures 159

Figure 3 Detailed model of NL sample

the model and the data conrm this From adjustment difculties there are two pathsthat are connected indirectly to depression one via loneliness and the second viaruminations about home Self-esteem (a personality trait which was assumed to havesome commonality with emotional stability) has a similar relationship to adjustmentdifculties and depression in the UK sample as emotional stability in the NLsample Asdemonstrated previously with respect to the total homesickness score when analysedat the subscale level females in the UK study report having more adjustment difcultiesthan males This would indicate that the difference in homesickness mainly lies infemales reporting more adjustment difculties

Study 1 and Study 2 comparisonThere were a number of signicant differences between the two national samplesDutch students were older (M = 192 years) than British students (M = 188 yearst (478) = 295 p lt 01) and Dutch students (M = 1988 days) had spent a longer numberof days away from home than British students (M = 501 days t (224) = 1098 p lt 001)Furthermore there were proportionately more females in the NLsample than the UKsample (82versus 72) The implications of these differences are discussed later

HomesicknessAfter controlling for age and duration of relocation it was still found that the NLand UKstudents differed considerably with respect to homesickness with the UK studentsreporting more intense homesickness Univariate differences were found for three ofthe ve subscales (missing family missing friends adjustment difculties) and the totalhomesickness score Adjusted means for the subscales and for total homesickness forthe two samples are presented in Table 2

Furthermore when asked in the separate question about general feelings of home-sickness the UK students more frequently endorsed having felt homesick (see Table 3)

160 Margaret Stroebe et al

Figure 4 Model of UK sample

These analyses indicate not only that the UK students are more often homesick butthat they are so to a greater extent

Antecedents and consequences of homesicknessThe global models (Figs 2 and 4) demonstrate that homesickness is related to depres-sion in both cultures and that low self-esteem or emotional instability are linked tohigher levels of self-reported homesickness and depression More possibly howeverthe longer the stay in the new situation the less homesick and consequently lessdepressed one tends to be However the evidence also suggests that there may becultural differences with respect to gender differences in homesickness

The more detailed models (Figs 3 and 5) have the same basic structure althoughthere are some differences In the UK females are more likely to experience adjustmentdifculties than are males (b = 15) and the longer a UK student has experienced thenew situation the less he or she is inclined to miss the family (b = plusmn 12) effects whichare non-signicant in the NLstudy To enhance comparisons between the cultures thegeographical distance to the parental home (data not available in the NL sample) wasnot included here in the UKmodel However if included the pattern remained the sameand revealed that distance from home was only reliably associated with missing thefamily (b = 19)

Attachment style and family backgroundWe conducted further analyses to explore the association between attachment style andfamily background variables with homesickness In the NLstudy the effects of insecureattachment on relocation the homesickness features depression and emotional stability

Homesickness among students in two cultures 161

Figure 5 Detailed model of UK sample

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 3: res_6

to become homesick and characteristics of the environment or situation that areassociated with greater intensities of it Highly relevant for student populations is thequestion of gender differences as a risk factor for homesickness While lay assump-tions might lead to expectations of greater female homesickness on leaving homefor college the literature does not support this Brewin et al (1989) reported thathomesickness was similarly prevalent among male and female students a ndingconrmed by Fisher for a variety of samples (eg Fisher et al 1985 Fisher amp Hood1987 1988) Ways of coping however differed between the genders with womenseeking more social support than was the case for men Similarly Archer et al (1998)found that female students showed higher levels of intrusive thinking about home-sickness mediated by their higher scores on the attachment factor of a homesick-ness questionnaire whereas there was no gender difference in avoidant responsesto homesickness There are gender differences in coping with and consequences ofother negative life-events notablymdashin this contextmdashbereavement (eg health detri-ments are relatively more extreme for widowers than for widows see Stroebe ampStroebe 1987) Thus further examination of patterns of gender differences in home-sickness would seem timely

Van Tilburg et al (1996) have recently emphasized the need for further examinationof cultural differences in the prevalence and experience of homesickness noting thatthis had not been studied systematically Carden and Feicht (1991) studied home-sickness among Turkish and American students in their own countries but limitedinvestigation to females Hojat and Herman (1985) studied a sample of Iranian andFilipino individuals but these were all resident in the USA and the sample wascomposed of physicians Ward and Kennedy (1993) examined homesickness amongstudents in different cultures but these were all from New Zealand residing in 23different countries of the world Lu (1990) also examined students who had moved tostudy in a different culture but again the study was conned to one cultural groupnamely Chinese students all of whom had moved to a university in the UK

Particularly important too in the context of our investigation are ndings to dowith personal vulnerability factors Both Carden and Feicht (1991) and Brewin et al(1989) reported associations between dependency and homesickness In the Brewinet al (1989) study lsquodependency on othersrsquo was a predictor of homesickness in theCarden and Feicht (1991) study dependency on family and parents was found to becharacteristic of homesick but not non-homesick students Personality and dispositionalcorrelates of homesickness have also been identied among student samples Fisher(1989) found that introversion depression and obsession were related to homesicknessbut that self-esteem was no different among homesick versus non-homesick studentsIn other samples lower self-esteem has indeed been reported for homesick samples(Eurelings-Bontekoe et al 1994 Hojat amp Herman 1985) Furthermore Eurelings-Bontekoe et al (1994) also found higher levels of introversion rigidity and negativismand lower levels of dominance among homesick conscripts Rigidity proved to be thebest predictor for homesickness These patterns are not however always conrmedfor other samples (cf van Tilburg et al 1996)

Turning from personal to situational factors lsquogeographic distancersquo is also importantfor the current investigation Among students Fisher et al (1986) found homesicknessto be higher for those more distant from home (although notably distance fromhome did not make a difference among boarding school children) Needing furtherinvestigation are related factors such as psychological distance and possibilities forcommunication with home

Homesickness among students in two cultures 149

Finally with respect to the consequences or outcomes associated with homesick-ness among students studies have also found a relationship between homesicknessand a greater number of cognitive failures poor concentration handing in work latedecrements in work quality and higher scores on anxiety and depression measures(Burt 1993) This has recently been conrmed in a meta-analysis of four studiesconducted by Archer et al (1998) As Brewin et al (1989) emphasized such ndingssuggest that homesickness is a potentially important phenomenon that may exercisea considerable inuence on academic performance at least in the short term

A theoretical approachThere has been surprisingly little exploration of the theoretical underpinnings ofthese phenomena and manifestations of homesickness a notable exception being theanalysis of Fisher (eg 1989) In fact from a psychological perspective there are goodreasons to consider homesickness a lsquomini-griefrsquo experience (Stroebe Stroebe amp Schut1993) and to draw on theoretical formulations in the eld of bereavement to derivehypotheses and explain phenomena (cf Stroebe amp Stroebe 1987) It is well establishedthat bereavement has severe mental and physical consequences for some but not allindividuals It is also known that the time course and symptomatology of grief arecomplex and varied (Stroebe Stroebe amp Hansson 1993) If homesickness is understoodas the emotional reaction to (temporary) loss of signicant persons then somemdashperhaps less extrememdashparallels with reactions of grief following losses through deathwould be expected Thus insofar as homesickness is a grief experience one wouldexpect more homesickness the greater the degree of (perceived) separation from home

A recent bereavement-specic model the Dual Process Model of Coping withBereavement (DPM Stroebe amp Schut 1999) provides a framework for integratingsome of the central propositions of two very different approaches cognitive stress(Lazarus ampFolkman 1984) and attachment (Bowlby 1980) theories both of which havebeen applied to the specic stressor of bereavement The DPM identies two typesof stressor that are relevant to the prediction of outcome (in terms of variables suchas health and well-being) namely those that are loss- versus restoration-orientedLoss-orientation refers to concentration on coping with the loss experience itselfand restoration-orientation refers to efforts to adjust to the concurrent changed anddemanding situation The model postulates the necessity of lsquooscillationrsquo betweenthese two tasks identifying the need both to come to terms with loss and to adaptto the different environment

In the case of loss through death loss-orientation would include missing thedeceased person in the case of homesickness it would include missing the absentfamily and friends Elements of attachment theory are clearly relevant here One wouldexpect more homesickness among those whose attachment is less secure (cf Bowlby1980) Following the reasoning of attachment theory as applied to grief experiences(cf Parkes Stevenson-Hinde amp Marris 1991) an insecure style of attachment wouldbe a predictor of homesickness which would lead to health consequences

Restoration-orientation following a death would include developing a new identityand new roles independent of the deceased After relocation it would seem likely thatsomewhat similar adjustments are needed (becoming a lsquocollege studentrsquo member ofnew social housing and academic groups) as are necessary following a loss throughdeath In line with cognitive stress theory (Lazarus amp Folkman 1984) the DPMpredictsthat those individuals experiencing a severe stressor (relocation) who lack the resources

150 Margaret Stroebe et al

(eg social support from home) would be unable to cope with the demands of the newsituation (eg academic performance) and would suffer homesickness and possiblydetrimental health consequences

Thus the DPM predicts that attention to and oscillation between these two tasksloss- and restoration-oriented are necessary for successful adaptation Similarly wewould predict that studentsrsquo well-being is dependent not only on the fact of havingleft home but also on successful adjustment to the new situation Thus the DPM pro-vides a theoretical basis for introducing two classes of variables into a structuralequation model for empirical investigation namely those to do with missing homefriends and so on and adjustment difculties in the new environment InterestinglyArcher et al (1998) recently identied two similar factors on a questionnaire measureof homesickness representing attachment to home and adjustment to the university

Purposes of the empirical studyTaken together what are the implications of this empirical and theoretical overviewfor further investigation of homesickness among students The review of empiricalstudies identies some emergent patterns of homesickness in relationship to personaland situational factors but some conicting results There is a need for replication ofndings and extension of the variables under investigation We have emphasized theneed for further investigation of the origins manifestations and consequences ofhomesickness among vulnerable groups From the outline above it is evident thatstudents leaving home for college belong in this category of vulnerable groups Lackingstill is not only information on predictors and outcomes but also on component partsof homesickness derived from the implementation of standard validation techniquesThus the rst aim of this study was to provide a detailed examination of homesick-ness among college students To this end two studies of homesickness were conductedone in the Netherlands and one in the UK

This two-study investigation enabled some examination of cultural patterns ofhomesickness To our knowledge this research provides the rst cross-cultural exam-ination of homesickness in both male and female students in their own countries Thusthe additional factor of acculturation (ie moving abroad and adjusting to the newcountry) does not occur in the examination of homesickness here Our research isalso unique in that it focuses on two cultures similar in terms of general societal back-ground in that they are both western European but different with respect to a centralattachment theory-related variable noted above namely accessibility to home (as willbe discussed later accessibility is greater in general in the Netherlands) As the abovereview indicated it is important to clarify the relationship between gender and home-sickness among student samples

A further advantage of the two reported studies is that they used the samestandardized homesickness measure In addition in each of the studies the relationshipof homesickness to a number of potentially important variables could be exploredincluding depression self-liking competence or self-esteem and secure or conictedattachment to the family The indices were standardized and thus provide usefulwithin-culture information which was the main focus of the analyses Between-culturecomparisons are to be made cautiously since some of the measures employed differedbetween the two countries (eg one standardized depression scale was used in theUK a different one in the Netherlands) However given the paucity of research todate and the intrinsic interest in comparing these phenomena across the two cultures

Homesickness among students in two cultures 151

as well as within each culture results are presented comparatively In addition anumber of demographic variables are analysed (eg gender age living arrangements)

Following the theoretical predictions outlined above it was expected that home-sickness would be prevalent and associated with distress and poorer mental health inthe student samples of both cultures but that prevalence would be greater in theculture with lower accessibility to home (UK) Following the DPM it was predictedthat missing home and difculty adjusting to the new situation would be antecedentto distress (rather than vice versa) Also that those who missed home more would bethe ones to have more difculty adapting to the new college situation (subscales onthe homesickness questionnaire) Furthermore it was predicted that homesicknesswould be greater among those living further from home irrespective of culture sincefollowing attachment theory the possibility of meeting up with signicant othersshould reduce separation anxiety and distress Again also following attachmenttheory it was predicted that students with greater parental conict1 or insecure attach-ment patterns would feel more homesick Homesickness was expected to occur in bothmale and female students of both cultures but it was predicted that female rateswould be higher (they could feel more comfortable admitting to homesickness) Finallyit was predicted that a longer time period since leaving home (or being at university)would be associated with lower homesickness (ie an adjustment effect) but that somestudents would still suffer homesickness even after being away for a longer period

Method

Study 1 The Netherlands (NL)

Respondents and procedureAll new-intake students entering the Faculty of Social Sciences of Utrecht University were mailed aquestionnaire containing the homesickness scale This took place in January 1995 Four weekslater a reminder was mailed to non-respondents This resulted in 482 completed and returnedquestionnaires (response rate 65) Eighty per cent of the respondents were female which issimilar to the proportion of females entering this faculty The mean age was 196 years (SD= 21)Twenty-nine per cent still lived at home with their parents 489had moved from the parentalhome to lodgings 196had recently moved from one place of lodging to another) and 26hadmoved back from lodgings to the parental home Most of the data analyses were conducted on the489relocated students while those living at home were included in specic analyses for controlpurposes (see below) The Utrecht students are referred to here as the NL sample

Measures

Socio-demographic informationRespondents were asked questions about their age gender previous living situation presentliving situation and the duration of their present living situation

The Utrecht Homesickness Scale (UHS)Designed to assess the extent of homesickness the UHS was developed in the following manner(see van Vliet Stroebe ampSchut 1998) Items were initiallyderived from previous research (Carden ampFeight 1991 Eurelings-Bontekoe et al 1994 Fisher Elder ampPeacock 1990 Fisher Frazer ampMurray

152 Margaret Stroebe et al

1 This prediction was tentative given that the evidence on con icted relationships is mixed (cf Archer 1999 pp 170ndash172)

1984 1985 1986 Fisher amp Hood 1987 1988) Items derived from a pilot study in which 100respondents were asked to give a denition of homesickness supplemented the original list

This preliminary questionnaire was given to 300 social science students who were asked torate the extent to which 51 items were associated with homesickness This resulted in a list of 45items This version of the questionnaire was then given to 117 third year psychology studentsThese students were asked to rate the extent to which theyhad experienced the 45 homesickness-related aspects in the past four weeks This led to a third version of the questionnaire which wasused in the Utrecht Homesickness Project The 45 items assess the extent to which studentsexperienced these various aspects of homesickness in the past 4 weeks (using the answercategories lsquonotrsquo lsquoweakrsquo lsquomoderatersquo lsquostrongrsquo lsquovery strongrsquo)

Aprincipal component analysis (minimum eigenvalue gt 1) with a varimax rotation over the 45items was computed for students who had felt homesick in the past four weeks (N = 151)resulting in nine factors On the basis of a scree plot we used the rst ve factors which wereclearly interpretable and explained 57of the variance These ve factors correspond closely tothe features that according to most authors constitute homesickness missing family missingfriends having adjustment difculties ruminations about home and feeling lonely

Based on these ve factors a reduced set of items (20) was selected to ensure that thedeterminant of the correlation matrix would be gt 0 This is necessary to avoid co-linearity of setsof items Asecond principal component analysis with a varimax rotation over these 20 items andthe restriction that ve factors should be extracted revealed that the items loaded as expected onthe relevant ve factors This last procedure was repeated with all the students (N = 439) andrevealed the same factor loading pattern Within this set of 20 variables the ve factors explained73 of the variance On the basis of the ve-factor structure scales were computed resultingin very acceptable Cronbachrsquos alphas (see Table 1) This set of subscales will be referred to as theUtrecht Homesickness Scale (UHS) Note that if there were missing values lsquolist-wise deletionrsquo wasused

Additional measuresStudents were also asked how often they had experienced homesickness in the past 4 weeks(never rarely sometimes often very often) Adiscriminant analysis whereby the ve UHS scalesare used to predict whether or not students endorsed feeling homesick (never versus rarelysometimes often very often) in the past 4 weeks revealed that 83 were correctly classiedFurthermore there was a strong correlation between the total homesickness score and thefrequency of homesickness in the past 4 weeks (r = 71) which indicates that the UHS scale isclosely related to a direct measure of homesickness even though the scale does not mentionhomesickness specically The Dutch version of the Symptoms Check List (SCL-90 Arrindell ampEttema 1975) was used as an indicator of mental health status of the NL sample Only thedepressive mood subscale (a = 92) and the total health score (a = 97) were used for this part ofthe investigation

Personality factors were measured by means of a Dutch translation of the Golberg Bi-polar Big5 (De Jong van Eck ampvan den Bos 1994 eg Goldberg 1992) Only emotional stability (a = 85)was used for this investigation

A Dutch translation of the Attachment Style Measure (Hazan amp Shaver 1987) was used toassess insecure (that is avoidant and anxiousambivalent see Hazan ampShaver 1987) versus secureattachment This particular scale was chosen because of its conciseness and applicability for usewith these student samples

Statistical analysesComparable to what was done in the pre-test described previously we rst conducted factoranalysis on the homesickness scale which identied the ve subscales A principal componentanalysis (minimum eigenvalue gt 1) with a varimax rotation over the 45 items resulted in ninefactors On the basis of a scree plot we used the rst ve factors which were clearly interpretable

Homesickness among students in two cultures 153

and explained 57of the variance These ve factors correspond closely to the factors identiedin the pre-test on the UHS namely missing family missing friends having adjustment difcultiesruminations about home and feeling lonely

Then we carried out a multivariate analysis of variance (MANOVA) to test differences inhomesickness scores between the UK and NL samples We then computed path analyses (usingLISREL81) in order to examine multiple relationships simultaneously We hypothesized identical

154 Margaret Stroebe et al

Table 1 Utrecht Homesickness Scale factor structure and reliabilities (NL sample)

Factor

(1) (2) (3) (4) (5)

Percentage of explained varianceItem

62 15 9 8 6

Factor loadings

(1) Missing family (a = 90)Missing your parents 89 12 20 06 16Missing your family 87 13 19 09 14Missing home 74 12 21 18 37Feeling missed by your family 74 15 19 15 04

(2) Loneliness (a = 85)Feeling lonely 21 78 22 25 09Feeling unloved 12 75 22 15 07Feeling isolated from the rest of the world 09 72 14 33 22Feeling uprooted 08 64 21 35 22

(3) Missing friends (a = 87)Longing for acquaintances 16 22 83 19 16Searching for familiar faces 19 18 73 23 12Missing people whom you trust and can talk with 25 31 71 07 20Missing your friends 36 14 70 15 14

(4) Adjustment dif culties (a = 88)Finding it dif cult adjusting to a new situation 23 27 15 77 16Feeling uncomfortable in a new situation 15 44 06 71 23Feeling lost in the new situation 17 46 22 69 21Having dif culties in getting used to new customs 03 14 38 68 26

(5) Ruminations about home (a = 80)Having thoughts that an old situation was better thanhere and now 16 15 18 28 81Regretting the decision to leave an old situation 10 21 08 31 79Continuously having thoughts about home 41 08 24 13 68Repeatedly thinking of the past 18 45 28 plusmn 01 52

Total homesickness (a = 94) (20 items)

Note The questions were formulated in the following manner Could you please indicate to what extentyou have experienced the following in the past 4 weeks Missing your parents notndashweakndashmoderatendashstrongndashvery strong N = 439

models for both cultures assuming that the patterns of associations would be similar Weexamined models of the relationships between homesickness and its putative antecedents andconsequences in terms of both overall scores and subscale scores Precise details of these modelsare given in the results section

Several t indices as described by Jaccard and Wan (1996) are appropriate for our investigationof path models To demonstrate the overall degree of correspondence between predicted andobserved covariance matrices the x2 statistic with p gt 05 indicates that a perfect t model doesexist in the population The goodness-of-t index (GFI gt 90) indicates goodness of t Thestandardized root mean square residual (SRMR lt 10) indicates a deviation of less than 10correlation units on average between predicted and observed correlations The root mean squareerror of approximation (RMSEA lt 08) implies an adequate model and lt 05 implies a good modelwith the p value for close t (CFit) being non-signicant (gt 05) CFI gt 90 implies a good t

Study 2 The United Kingdom (UK)

Respondents and procedureDuring the autumn term of 1996 new-intake students at Cardiff University were askedto participate in a student survey Questionnaires were administered in October and Novemberand 280 completed questionnaires were returned representing approximately a 60 responserate Seventy-two per cent of the respondents were female The mean age was 188 years(SD = 21) All respondents had moved from the parental home to lodgings on campus TheCardiff students are referred to here as the UK sample

MeasuresAn English translation of the Utrecht Homesickness Scale (UHS) was used to measure the extent ofhomesickness The reliabilities of the scales were comparable to the NLsample missing the family(a = 85) loneliness (a = 84) missing friends (a = 78) adjustment difculties (a = 84)ruminations about home (a = 86) and the total score for homesickness (a = 93) Againrespondents were also asked to indicate how often they had felt homesick during the past 4weeks The Beck Depression Inventory (BDI Beck Ward Mendelson Mock amp Erbaugh 1961)was used to give a reliable measure of current level of depression (a = 82) The Self-likingSelf-competence Scale was used to assess self-esteem (a = 91) (cf Tafarodi amp Swann 1995) Anadapted version of the Conict Tactics Scale (cf Straus Hamby Boney-McCoy ampSugarman 1996)was used to ascertain type of family background of the UK respondents The Conict Tactics Scaleconsists of six subscales with similar acceptable alpha levels (on average 72) which assess theextent to which parents are perceived as handling conicts with the respondents in certain waysThe items used here reected three factors that is verbal reasoning (three items) verbalaggression (six items) and physical aggression (three items) of either father or mother Selectionof these items from the original much longer scale was thought to be less intrusive and easier forrespondents to answer while still providing a measure of the three factors The students were alsoasked to state the name of their home town On the basis of this information distances could becalculated between their home and their university lodgings

Statistical analysesThe analyses described above for Study 1 apply here too

ResultsThe results section is structured in three parts In the rst section results on the NLsample are reported In the second section the same is done for the UK sample In thelast section a comparison is made between the two samples

Homesickness among students in two cultures 155

Study 1 NL

HomesicknessAlmost 50of the students reported having felt homesick at least some of the time afterhaving started their new education (N = 230) There were no differences betweenmales and females with respect to homesickness intensity The intensity of home-sickness as measured by the UHS subscales (range 1 (not experienced) to 5 (verystrongly)) revealed that on average the highest scores were obtained for missing friends(M = 238 SD = 102) and missing the family (M = 215 SD = 095)2 The otherdimensions scored were adjustment difculties (M = 180 SD = 088) ruminationsabout home (M = 152 SD = 071) and loneliness (M = 179 SD = 087) Theseaveraged to a total mean score of 193 (SD = 071)

Antecedents and consequences of homesicknessThe NL data set offered the possibility of comparisons between students still living athome and those who had relocated3 This provided the opportunity to test whetherrelocation (R) would indeed be associated with depression (D) which in and of itselfwe assumed would be associated with homesickness (H) We must be cautious giventhe cross-sectional design of our studies in assuming mediation between the RndashDndashHvariables Based on Baron and Kenny (1986) a requirement for mediation with this dataset would be that relocation and depression are signicantly associated in other wordsthere must be an effect to be mediated In our data we nd that relocation anddepression are not signicantly associated and therefore a mediation analysis is notappropriate Nevertheless a path analysis is possible if we regress depression on re-location and homesickness and regress homesickness on relocation (see Fig 1) Ourresults show a signicant path from relocation to homesickness (b = 21) a signicantpath from homesickness to depression controlling for relocation (b = 62) and a non-signicant direct path from relocation to depression (b = plusmn 05) Showing that thetwo component paths of the indirect effect are each signicant is not the same asshowing that the indirect effect is signicant Testing for an indirect path (Sobel 1982)revealed that the indirect effect of relocation on depression is signicant (b = 13) Itseems reasonable to conclude then that the data are consistent with an indirect effectof relocation on depression by means of the intervening variable homesickness Whilethere are good reasons to argue that there is an indirect effect of relocation ondepression through homesickness it is evident that this effect though signicant isquite small and not strong enough to produce an overall effect of R on D

The model shown in Fig 1 was also tested using the SCL-90 total score as the healthmeasure instead of using the depression subscale This showed an almost identicalsolution The analysis provides further evidence that the assumed pattern is justiedand that it is not just due to some correspondence between depression and home-sickness The pattern suggests that relocation may be an antecedent for homesicknessbut not for depressive moods or general ill-health and that homesickness may bean important antecedent for ill-health Having found evidence for this relationshipstructural equation models could be tested with respect to the effects of emotionalstability gender and duration of stay on homesickness and depression (see Fig 2) The

156 Margaret Stroebe et al

2 Controlling for age and duration of stay in order to make comparisons with the UK sample possible3 Some students who had not moved also reported being homesick (15) in the past 4 weeks The vast majority of them(90) reported that they experienced homesickness only lsquorarelyrsquo or lsquosometimesrsquo

literature suggests that emotional instability is a good predictor for both homesickness(Eurelings-Bontekoe et al 1996) and depression (cf Watson ampPennebaker 1989) Thusone could assume that emotional stability (the opposite of emotional instability) wouldexplain the variance of depression However we nd that even though reducedhomesickness retains an effect on depression This suggests that irrespective of thelevel of emotional stability homesickness is a precursor for depression Furthermorethe emotionally unstable as expected have higher levels of homesickness and depres-sion What we also nd is that the longer one is away from home the less intense thehomesickness reaction which suggests that adaptation to the new situation takes place

In order to create a more comprehensive picture a model was formulated whichincluded the subscales of the UHS instead of the total homesickness score Theassumptions were that (1) adjustment difculties would be associated with missingfamily and friends ruminations about home and loneliness (2) missing family andfriends in turn would be associated with ruminations about home and loneliness (3)ruminations would also be associated to loneliness and depression and (4) lonelinessin turn to depression Results with respect to this model revealed that in principle theassumptions were correct if two modications were incorporated These wereestablishing a link between missing family and friends and removing the path frommissing family to loneliness The choice was made to let the errors of missing friendsand family correlate because no prior hypotheses were made as to a causal relation-ship between both features and to delete the non-signicant path4 Figure 3 indicateshow the homesickness subscales are related and how these are linked to the measuredantecedents and consequences To enhance the visual impact of the model the thick-ness of paths corresponds to the size of the path coefcients

Figure 3 suggests that to become depressed one has to experience adjustment

Homesickness among students in two cultures 157

Figure 1 Model of relocation homesickness and depression for NL students

4 For the sake of comparison with the forthcoming UK model the non-signicant paths between missing friends and lonelinessgender and adjustment dif culties and duration of stay and missing family were not removed from the model

difculties and consequently loneliness and to ruminate about home If the relocatedstudents experience severe adjustment difculties these do seem to go hand in handwith missing family and friends ruminations about home and loneliness Specicallyloneliness is most strongly associated with depression Furthermore the analysessuggest that the best predictor for adjustment difculties in this study is emotionalstability (b = plusmn 41) Emotional stability also has a strong relationship to depression(b = plusmn 45) and to a lesser extent to missing friends (b = plusmn 20) and loneliness(b = plusmn 15) The slight association of duration of stay with total homesickness dis-appeared when its relationship was tested on the ve separate homesickness subscales

Study 2 UK

HomesicknessMore than 80of the students reported having felt homesick at least some of the timeafter having started their new education (N = 280) The intensity of homesicknessas measured by the UHS subscales (likewise controlling for age and the number of daysaway from home) revealed that on average the highest scores in the UK were obtainedfor missing family (M = 293 SD = 097) and missing friends (M = 272 SD = 094)The other dimensions scored were adjustment difculties (M = 205 SD = 084)ruminations about home (M = 166 SD = 086) and loneliness (M = 195SD = 092) These averaged to a total mean score of 226 (SD = 073)

Antecedents and consequences of homesicknessAlthough the hypothesis could not be tested with the UKsample it seems reasonable toassume based on the NLdata that relocation would be associated with homesicknesswhich in turn would be associated with depression and not vice versa (of course similarcautions about mediating variables to those expressed for Study 1 are in order here too)

158 Margaret Stroebe et al

Figure 2 Model of NL sample

Accepting this premise and despite the differences in the depression and personalityscales used in the UK and the NL what one would expect is comparable patterns ofcorrelations since these should be the same irrespective of the particular instrumentused We formulated a structural equation model analogous to the one used in the NLstudy examining the effects of self-esteem gender and duration of stay based onhomesickness and depression for the UK students (see Fig 4) This analysis revealedthat analogous to the NL data (Fig 2) homesickness is associated with depressionand that low self-esteem can be considered a risk factor for becoming homesick anddepressed To see whether or not homesickness is the precursor for self-esteem insteadof self-esteem being the precursor as hypothesized the alternative model was testedResults did not show a signicant effect of homesickness on self-esteem Thus low self-esteem like low emotional stability can rather be considered a vulnerability factorfor homesickness Again we see that after controlling for self-esteem the relationshipbetween homesickness and depression remains signicant This suggests that home-sickness is associated with depression irrespective of the level of self-esteem while ifone has low self-esteem the risk of becoming homesick and depressed is greater TheUK results also reveal the slight negative effect of duration of stay the longer studentslive in a new situation the lower the level of homesickness In contrast to the NLdata the UK data do reveal a gender difference Female students (M = 241 SD = 074)report higher levels of homesickness (F(278 1) = 1585 p lt 001) than male students(M = 204 SD = 060)

Analogous to the NL study we formulated a more detailed model including theUHS subscales (Fig 5) This revealed a similar pattern of correlations although thereare some obvious differences Adjustment difculties are assumed to be the crux of

Homesickness among students in two cultures 159

Figure 3 Detailed model of NL sample

the model and the data conrm this From adjustment difculties there are two pathsthat are connected indirectly to depression one via loneliness and the second viaruminations about home Self-esteem (a personality trait which was assumed to havesome commonality with emotional stability) has a similar relationship to adjustmentdifculties and depression in the UK sample as emotional stability in the NLsample Asdemonstrated previously with respect to the total homesickness score when analysedat the subscale level females in the UK study report having more adjustment difcultiesthan males This would indicate that the difference in homesickness mainly lies infemales reporting more adjustment difculties

Study 1 and Study 2 comparisonThere were a number of signicant differences between the two national samplesDutch students were older (M = 192 years) than British students (M = 188 yearst (478) = 295 p lt 01) and Dutch students (M = 1988 days) had spent a longer numberof days away from home than British students (M = 501 days t (224) = 1098 p lt 001)Furthermore there were proportionately more females in the NLsample than the UKsample (82versus 72) The implications of these differences are discussed later

HomesicknessAfter controlling for age and duration of relocation it was still found that the NLand UKstudents differed considerably with respect to homesickness with the UK studentsreporting more intense homesickness Univariate differences were found for three ofthe ve subscales (missing family missing friends adjustment difculties) and the totalhomesickness score Adjusted means for the subscales and for total homesickness forthe two samples are presented in Table 2

Furthermore when asked in the separate question about general feelings of home-sickness the UK students more frequently endorsed having felt homesick (see Table 3)

160 Margaret Stroebe et al

Figure 4 Model of UK sample

These analyses indicate not only that the UK students are more often homesick butthat they are so to a greater extent

Antecedents and consequences of homesicknessThe global models (Figs 2 and 4) demonstrate that homesickness is related to depres-sion in both cultures and that low self-esteem or emotional instability are linked tohigher levels of self-reported homesickness and depression More possibly howeverthe longer the stay in the new situation the less homesick and consequently lessdepressed one tends to be However the evidence also suggests that there may becultural differences with respect to gender differences in homesickness

The more detailed models (Figs 3 and 5) have the same basic structure althoughthere are some differences In the UK females are more likely to experience adjustmentdifculties than are males (b = 15) and the longer a UK student has experienced thenew situation the less he or she is inclined to miss the family (b = plusmn 12) effects whichare non-signicant in the NLstudy To enhance comparisons between the cultures thegeographical distance to the parental home (data not available in the NL sample) wasnot included here in the UKmodel However if included the pattern remained the sameand revealed that distance from home was only reliably associated with missing thefamily (b = 19)

Attachment style and family backgroundWe conducted further analyses to explore the association between attachment style andfamily background variables with homesickness In the NLstudy the effects of insecureattachment on relocation the homesickness features depression and emotional stability

Homesickness among students in two cultures 161

Figure 5 Detailed model of UK sample

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 4: res_6

Finally with respect to the consequences or outcomes associated with homesick-ness among students studies have also found a relationship between homesicknessand a greater number of cognitive failures poor concentration handing in work latedecrements in work quality and higher scores on anxiety and depression measures(Burt 1993) This has recently been conrmed in a meta-analysis of four studiesconducted by Archer et al (1998) As Brewin et al (1989) emphasized such ndingssuggest that homesickness is a potentially important phenomenon that may exercisea considerable inuence on academic performance at least in the short term

A theoretical approachThere has been surprisingly little exploration of the theoretical underpinnings ofthese phenomena and manifestations of homesickness a notable exception being theanalysis of Fisher (eg 1989) In fact from a psychological perspective there are goodreasons to consider homesickness a lsquomini-griefrsquo experience (Stroebe Stroebe amp Schut1993) and to draw on theoretical formulations in the eld of bereavement to derivehypotheses and explain phenomena (cf Stroebe amp Stroebe 1987) It is well establishedthat bereavement has severe mental and physical consequences for some but not allindividuals It is also known that the time course and symptomatology of grief arecomplex and varied (Stroebe Stroebe amp Hansson 1993) If homesickness is understoodas the emotional reaction to (temporary) loss of signicant persons then somemdashperhaps less extrememdashparallels with reactions of grief following losses through deathwould be expected Thus insofar as homesickness is a grief experience one wouldexpect more homesickness the greater the degree of (perceived) separation from home

A recent bereavement-specic model the Dual Process Model of Coping withBereavement (DPM Stroebe amp Schut 1999) provides a framework for integratingsome of the central propositions of two very different approaches cognitive stress(Lazarus ampFolkman 1984) and attachment (Bowlby 1980) theories both of which havebeen applied to the specic stressor of bereavement The DPM identies two typesof stressor that are relevant to the prediction of outcome (in terms of variables suchas health and well-being) namely those that are loss- versus restoration-orientedLoss-orientation refers to concentration on coping with the loss experience itselfand restoration-orientation refers to efforts to adjust to the concurrent changed anddemanding situation The model postulates the necessity of lsquooscillationrsquo betweenthese two tasks identifying the need both to come to terms with loss and to adaptto the different environment

In the case of loss through death loss-orientation would include missing thedeceased person in the case of homesickness it would include missing the absentfamily and friends Elements of attachment theory are clearly relevant here One wouldexpect more homesickness among those whose attachment is less secure (cf Bowlby1980) Following the reasoning of attachment theory as applied to grief experiences(cf Parkes Stevenson-Hinde amp Marris 1991) an insecure style of attachment wouldbe a predictor of homesickness which would lead to health consequences

Restoration-orientation following a death would include developing a new identityand new roles independent of the deceased After relocation it would seem likely thatsomewhat similar adjustments are needed (becoming a lsquocollege studentrsquo member ofnew social housing and academic groups) as are necessary following a loss throughdeath In line with cognitive stress theory (Lazarus amp Folkman 1984) the DPMpredictsthat those individuals experiencing a severe stressor (relocation) who lack the resources

150 Margaret Stroebe et al

(eg social support from home) would be unable to cope with the demands of the newsituation (eg academic performance) and would suffer homesickness and possiblydetrimental health consequences

Thus the DPM predicts that attention to and oscillation between these two tasksloss- and restoration-oriented are necessary for successful adaptation Similarly wewould predict that studentsrsquo well-being is dependent not only on the fact of havingleft home but also on successful adjustment to the new situation Thus the DPM pro-vides a theoretical basis for introducing two classes of variables into a structuralequation model for empirical investigation namely those to do with missing homefriends and so on and adjustment difculties in the new environment InterestinglyArcher et al (1998) recently identied two similar factors on a questionnaire measureof homesickness representing attachment to home and adjustment to the university

Purposes of the empirical studyTaken together what are the implications of this empirical and theoretical overviewfor further investigation of homesickness among students The review of empiricalstudies identies some emergent patterns of homesickness in relationship to personaland situational factors but some conicting results There is a need for replication ofndings and extension of the variables under investigation We have emphasized theneed for further investigation of the origins manifestations and consequences ofhomesickness among vulnerable groups From the outline above it is evident thatstudents leaving home for college belong in this category of vulnerable groups Lackingstill is not only information on predictors and outcomes but also on component partsof homesickness derived from the implementation of standard validation techniquesThus the rst aim of this study was to provide a detailed examination of homesick-ness among college students To this end two studies of homesickness were conductedone in the Netherlands and one in the UK

This two-study investigation enabled some examination of cultural patterns ofhomesickness To our knowledge this research provides the rst cross-cultural exam-ination of homesickness in both male and female students in their own countries Thusthe additional factor of acculturation (ie moving abroad and adjusting to the newcountry) does not occur in the examination of homesickness here Our research isalso unique in that it focuses on two cultures similar in terms of general societal back-ground in that they are both western European but different with respect to a centralattachment theory-related variable noted above namely accessibility to home (as willbe discussed later accessibility is greater in general in the Netherlands) As the abovereview indicated it is important to clarify the relationship between gender and home-sickness among student samples

A further advantage of the two reported studies is that they used the samestandardized homesickness measure In addition in each of the studies the relationshipof homesickness to a number of potentially important variables could be exploredincluding depression self-liking competence or self-esteem and secure or conictedattachment to the family The indices were standardized and thus provide usefulwithin-culture information which was the main focus of the analyses Between-culturecomparisons are to be made cautiously since some of the measures employed differedbetween the two countries (eg one standardized depression scale was used in theUK a different one in the Netherlands) However given the paucity of research todate and the intrinsic interest in comparing these phenomena across the two cultures

Homesickness among students in two cultures 151

as well as within each culture results are presented comparatively In addition anumber of demographic variables are analysed (eg gender age living arrangements)

Following the theoretical predictions outlined above it was expected that home-sickness would be prevalent and associated with distress and poorer mental health inthe student samples of both cultures but that prevalence would be greater in theculture with lower accessibility to home (UK) Following the DPM it was predictedthat missing home and difculty adjusting to the new situation would be antecedentto distress (rather than vice versa) Also that those who missed home more would bethe ones to have more difculty adapting to the new college situation (subscales onthe homesickness questionnaire) Furthermore it was predicted that homesicknesswould be greater among those living further from home irrespective of culture sincefollowing attachment theory the possibility of meeting up with signicant othersshould reduce separation anxiety and distress Again also following attachmenttheory it was predicted that students with greater parental conict1 or insecure attach-ment patterns would feel more homesick Homesickness was expected to occur in bothmale and female students of both cultures but it was predicted that female rateswould be higher (they could feel more comfortable admitting to homesickness) Finallyit was predicted that a longer time period since leaving home (or being at university)would be associated with lower homesickness (ie an adjustment effect) but that somestudents would still suffer homesickness even after being away for a longer period

Method

Study 1 The Netherlands (NL)

Respondents and procedureAll new-intake students entering the Faculty of Social Sciences of Utrecht University were mailed aquestionnaire containing the homesickness scale This took place in January 1995 Four weekslater a reminder was mailed to non-respondents This resulted in 482 completed and returnedquestionnaires (response rate 65) Eighty per cent of the respondents were female which issimilar to the proportion of females entering this faculty The mean age was 196 years (SD= 21)Twenty-nine per cent still lived at home with their parents 489had moved from the parentalhome to lodgings 196had recently moved from one place of lodging to another) and 26hadmoved back from lodgings to the parental home Most of the data analyses were conducted on the489relocated students while those living at home were included in specic analyses for controlpurposes (see below) The Utrecht students are referred to here as the NL sample

Measures

Socio-demographic informationRespondents were asked questions about their age gender previous living situation presentliving situation and the duration of their present living situation

The Utrecht Homesickness Scale (UHS)Designed to assess the extent of homesickness the UHS was developed in the following manner(see van Vliet Stroebe ampSchut 1998) Items were initiallyderived from previous research (Carden ampFeight 1991 Eurelings-Bontekoe et al 1994 Fisher Elder ampPeacock 1990 Fisher Frazer ampMurray

152 Margaret Stroebe et al

1 This prediction was tentative given that the evidence on con icted relationships is mixed (cf Archer 1999 pp 170ndash172)

1984 1985 1986 Fisher amp Hood 1987 1988) Items derived from a pilot study in which 100respondents were asked to give a denition of homesickness supplemented the original list

This preliminary questionnaire was given to 300 social science students who were asked torate the extent to which 51 items were associated with homesickness This resulted in a list of 45items This version of the questionnaire was then given to 117 third year psychology studentsThese students were asked to rate the extent to which theyhad experienced the 45 homesickness-related aspects in the past four weeks This led to a third version of the questionnaire which wasused in the Utrecht Homesickness Project The 45 items assess the extent to which studentsexperienced these various aspects of homesickness in the past 4 weeks (using the answercategories lsquonotrsquo lsquoweakrsquo lsquomoderatersquo lsquostrongrsquo lsquovery strongrsquo)

Aprincipal component analysis (minimum eigenvalue gt 1) with a varimax rotation over the 45items was computed for students who had felt homesick in the past four weeks (N = 151)resulting in nine factors On the basis of a scree plot we used the rst ve factors which wereclearly interpretable and explained 57of the variance These ve factors correspond closely tothe features that according to most authors constitute homesickness missing family missingfriends having adjustment difculties ruminations about home and feeling lonely

Based on these ve factors a reduced set of items (20) was selected to ensure that thedeterminant of the correlation matrix would be gt 0 This is necessary to avoid co-linearity of setsof items Asecond principal component analysis with a varimax rotation over these 20 items andthe restriction that ve factors should be extracted revealed that the items loaded as expected onthe relevant ve factors This last procedure was repeated with all the students (N = 439) andrevealed the same factor loading pattern Within this set of 20 variables the ve factors explained73 of the variance On the basis of the ve-factor structure scales were computed resultingin very acceptable Cronbachrsquos alphas (see Table 1) This set of subscales will be referred to as theUtrecht Homesickness Scale (UHS) Note that if there were missing values lsquolist-wise deletionrsquo wasused

Additional measuresStudents were also asked how often they had experienced homesickness in the past 4 weeks(never rarely sometimes often very often) Adiscriminant analysis whereby the ve UHS scalesare used to predict whether or not students endorsed feeling homesick (never versus rarelysometimes often very often) in the past 4 weeks revealed that 83 were correctly classiedFurthermore there was a strong correlation between the total homesickness score and thefrequency of homesickness in the past 4 weeks (r = 71) which indicates that the UHS scale isclosely related to a direct measure of homesickness even though the scale does not mentionhomesickness specically The Dutch version of the Symptoms Check List (SCL-90 Arrindell ampEttema 1975) was used as an indicator of mental health status of the NL sample Only thedepressive mood subscale (a = 92) and the total health score (a = 97) were used for this part ofthe investigation

Personality factors were measured by means of a Dutch translation of the Golberg Bi-polar Big5 (De Jong van Eck ampvan den Bos 1994 eg Goldberg 1992) Only emotional stability (a = 85)was used for this investigation

A Dutch translation of the Attachment Style Measure (Hazan amp Shaver 1987) was used toassess insecure (that is avoidant and anxiousambivalent see Hazan ampShaver 1987) versus secureattachment This particular scale was chosen because of its conciseness and applicability for usewith these student samples

Statistical analysesComparable to what was done in the pre-test described previously we rst conducted factoranalysis on the homesickness scale which identied the ve subscales A principal componentanalysis (minimum eigenvalue gt 1) with a varimax rotation over the 45 items resulted in ninefactors On the basis of a scree plot we used the rst ve factors which were clearly interpretable

Homesickness among students in two cultures 153

and explained 57of the variance These ve factors correspond closely to the factors identiedin the pre-test on the UHS namely missing family missing friends having adjustment difcultiesruminations about home and feeling lonely

Then we carried out a multivariate analysis of variance (MANOVA) to test differences inhomesickness scores between the UK and NL samples We then computed path analyses (usingLISREL81) in order to examine multiple relationships simultaneously We hypothesized identical

154 Margaret Stroebe et al

Table 1 Utrecht Homesickness Scale factor structure and reliabilities (NL sample)

Factor

(1) (2) (3) (4) (5)

Percentage of explained varianceItem

62 15 9 8 6

Factor loadings

(1) Missing family (a = 90)Missing your parents 89 12 20 06 16Missing your family 87 13 19 09 14Missing home 74 12 21 18 37Feeling missed by your family 74 15 19 15 04

(2) Loneliness (a = 85)Feeling lonely 21 78 22 25 09Feeling unloved 12 75 22 15 07Feeling isolated from the rest of the world 09 72 14 33 22Feeling uprooted 08 64 21 35 22

(3) Missing friends (a = 87)Longing for acquaintances 16 22 83 19 16Searching for familiar faces 19 18 73 23 12Missing people whom you trust and can talk with 25 31 71 07 20Missing your friends 36 14 70 15 14

(4) Adjustment dif culties (a = 88)Finding it dif cult adjusting to a new situation 23 27 15 77 16Feeling uncomfortable in a new situation 15 44 06 71 23Feeling lost in the new situation 17 46 22 69 21Having dif culties in getting used to new customs 03 14 38 68 26

(5) Ruminations about home (a = 80)Having thoughts that an old situation was better thanhere and now 16 15 18 28 81Regretting the decision to leave an old situation 10 21 08 31 79Continuously having thoughts about home 41 08 24 13 68Repeatedly thinking of the past 18 45 28 plusmn 01 52

Total homesickness (a = 94) (20 items)

Note The questions were formulated in the following manner Could you please indicate to what extentyou have experienced the following in the past 4 weeks Missing your parents notndashweakndashmoderatendashstrongndashvery strong N = 439

models for both cultures assuming that the patterns of associations would be similar Weexamined models of the relationships between homesickness and its putative antecedents andconsequences in terms of both overall scores and subscale scores Precise details of these modelsare given in the results section

Several t indices as described by Jaccard and Wan (1996) are appropriate for our investigationof path models To demonstrate the overall degree of correspondence between predicted andobserved covariance matrices the x2 statistic with p gt 05 indicates that a perfect t model doesexist in the population The goodness-of-t index (GFI gt 90) indicates goodness of t Thestandardized root mean square residual (SRMR lt 10) indicates a deviation of less than 10correlation units on average between predicted and observed correlations The root mean squareerror of approximation (RMSEA lt 08) implies an adequate model and lt 05 implies a good modelwith the p value for close t (CFit) being non-signicant (gt 05) CFI gt 90 implies a good t

Study 2 The United Kingdom (UK)

Respondents and procedureDuring the autumn term of 1996 new-intake students at Cardiff University were askedto participate in a student survey Questionnaires were administered in October and Novemberand 280 completed questionnaires were returned representing approximately a 60 responserate Seventy-two per cent of the respondents were female The mean age was 188 years(SD = 21) All respondents had moved from the parental home to lodgings on campus TheCardiff students are referred to here as the UK sample

MeasuresAn English translation of the Utrecht Homesickness Scale (UHS) was used to measure the extent ofhomesickness The reliabilities of the scales were comparable to the NLsample missing the family(a = 85) loneliness (a = 84) missing friends (a = 78) adjustment difculties (a = 84)ruminations about home (a = 86) and the total score for homesickness (a = 93) Againrespondents were also asked to indicate how often they had felt homesick during the past 4weeks The Beck Depression Inventory (BDI Beck Ward Mendelson Mock amp Erbaugh 1961)was used to give a reliable measure of current level of depression (a = 82) The Self-likingSelf-competence Scale was used to assess self-esteem (a = 91) (cf Tafarodi amp Swann 1995) Anadapted version of the Conict Tactics Scale (cf Straus Hamby Boney-McCoy ampSugarman 1996)was used to ascertain type of family background of the UK respondents The Conict Tactics Scaleconsists of six subscales with similar acceptable alpha levels (on average 72) which assess theextent to which parents are perceived as handling conicts with the respondents in certain waysThe items used here reected three factors that is verbal reasoning (three items) verbalaggression (six items) and physical aggression (three items) of either father or mother Selectionof these items from the original much longer scale was thought to be less intrusive and easier forrespondents to answer while still providing a measure of the three factors The students were alsoasked to state the name of their home town On the basis of this information distances could becalculated between their home and their university lodgings

Statistical analysesThe analyses described above for Study 1 apply here too

ResultsThe results section is structured in three parts In the rst section results on the NLsample are reported In the second section the same is done for the UK sample In thelast section a comparison is made between the two samples

Homesickness among students in two cultures 155

Study 1 NL

HomesicknessAlmost 50of the students reported having felt homesick at least some of the time afterhaving started their new education (N = 230) There were no differences betweenmales and females with respect to homesickness intensity The intensity of home-sickness as measured by the UHS subscales (range 1 (not experienced) to 5 (verystrongly)) revealed that on average the highest scores were obtained for missing friends(M = 238 SD = 102) and missing the family (M = 215 SD = 095)2 The otherdimensions scored were adjustment difculties (M = 180 SD = 088) ruminationsabout home (M = 152 SD = 071) and loneliness (M = 179 SD = 087) Theseaveraged to a total mean score of 193 (SD = 071)

Antecedents and consequences of homesicknessThe NL data set offered the possibility of comparisons between students still living athome and those who had relocated3 This provided the opportunity to test whetherrelocation (R) would indeed be associated with depression (D) which in and of itselfwe assumed would be associated with homesickness (H) We must be cautious giventhe cross-sectional design of our studies in assuming mediation between the RndashDndashHvariables Based on Baron and Kenny (1986) a requirement for mediation with this dataset would be that relocation and depression are signicantly associated in other wordsthere must be an effect to be mediated In our data we nd that relocation anddepression are not signicantly associated and therefore a mediation analysis is notappropriate Nevertheless a path analysis is possible if we regress depression on re-location and homesickness and regress homesickness on relocation (see Fig 1) Ourresults show a signicant path from relocation to homesickness (b = 21) a signicantpath from homesickness to depression controlling for relocation (b = 62) and a non-signicant direct path from relocation to depression (b = plusmn 05) Showing that thetwo component paths of the indirect effect are each signicant is not the same asshowing that the indirect effect is signicant Testing for an indirect path (Sobel 1982)revealed that the indirect effect of relocation on depression is signicant (b = 13) Itseems reasonable to conclude then that the data are consistent with an indirect effectof relocation on depression by means of the intervening variable homesickness Whilethere are good reasons to argue that there is an indirect effect of relocation ondepression through homesickness it is evident that this effect though signicant isquite small and not strong enough to produce an overall effect of R on D

The model shown in Fig 1 was also tested using the SCL-90 total score as the healthmeasure instead of using the depression subscale This showed an almost identicalsolution The analysis provides further evidence that the assumed pattern is justiedand that it is not just due to some correspondence between depression and home-sickness The pattern suggests that relocation may be an antecedent for homesicknessbut not for depressive moods or general ill-health and that homesickness may bean important antecedent for ill-health Having found evidence for this relationshipstructural equation models could be tested with respect to the effects of emotionalstability gender and duration of stay on homesickness and depression (see Fig 2) The

156 Margaret Stroebe et al

2 Controlling for age and duration of stay in order to make comparisons with the UK sample possible3 Some students who had not moved also reported being homesick (15) in the past 4 weeks The vast majority of them(90) reported that they experienced homesickness only lsquorarelyrsquo or lsquosometimesrsquo

literature suggests that emotional instability is a good predictor for both homesickness(Eurelings-Bontekoe et al 1996) and depression (cf Watson ampPennebaker 1989) Thusone could assume that emotional stability (the opposite of emotional instability) wouldexplain the variance of depression However we nd that even though reducedhomesickness retains an effect on depression This suggests that irrespective of thelevel of emotional stability homesickness is a precursor for depression Furthermorethe emotionally unstable as expected have higher levels of homesickness and depres-sion What we also nd is that the longer one is away from home the less intense thehomesickness reaction which suggests that adaptation to the new situation takes place

In order to create a more comprehensive picture a model was formulated whichincluded the subscales of the UHS instead of the total homesickness score Theassumptions were that (1) adjustment difculties would be associated with missingfamily and friends ruminations about home and loneliness (2) missing family andfriends in turn would be associated with ruminations about home and loneliness (3)ruminations would also be associated to loneliness and depression and (4) lonelinessin turn to depression Results with respect to this model revealed that in principle theassumptions were correct if two modications were incorporated These wereestablishing a link between missing family and friends and removing the path frommissing family to loneliness The choice was made to let the errors of missing friendsand family correlate because no prior hypotheses were made as to a causal relation-ship between both features and to delete the non-signicant path4 Figure 3 indicateshow the homesickness subscales are related and how these are linked to the measuredantecedents and consequences To enhance the visual impact of the model the thick-ness of paths corresponds to the size of the path coefcients

Figure 3 suggests that to become depressed one has to experience adjustment

Homesickness among students in two cultures 157

Figure 1 Model of relocation homesickness and depression for NL students

4 For the sake of comparison with the forthcoming UK model the non-signicant paths between missing friends and lonelinessgender and adjustment dif culties and duration of stay and missing family were not removed from the model

difculties and consequently loneliness and to ruminate about home If the relocatedstudents experience severe adjustment difculties these do seem to go hand in handwith missing family and friends ruminations about home and loneliness Specicallyloneliness is most strongly associated with depression Furthermore the analysessuggest that the best predictor for adjustment difculties in this study is emotionalstability (b = plusmn 41) Emotional stability also has a strong relationship to depression(b = plusmn 45) and to a lesser extent to missing friends (b = plusmn 20) and loneliness(b = plusmn 15) The slight association of duration of stay with total homesickness dis-appeared when its relationship was tested on the ve separate homesickness subscales

Study 2 UK

HomesicknessMore than 80of the students reported having felt homesick at least some of the timeafter having started their new education (N = 280) The intensity of homesicknessas measured by the UHS subscales (likewise controlling for age and the number of daysaway from home) revealed that on average the highest scores in the UK were obtainedfor missing family (M = 293 SD = 097) and missing friends (M = 272 SD = 094)The other dimensions scored were adjustment difculties (M = 205 SD = 084)ruminations about home (M = 166 SD = 086) and loneliness (M = 195SD = 092) These averaged to a total mean score of 226 (SD = 073)

Antecedents and consequences of homesicknessAlthough the hypothesis could not be tested with the UKsample it seems reasonable toassume based on the NLdata that relocation would be associated with homesicknesswhich in turn would be associated with depression and not vice versa (of course similarcautions about mediating variables to those expressed for Study 1 are in order here too)

158 Margaret Stroebe et al

Figure 2 Model of NL sample

Accepting this premise and despite the differences in the depression and personalityscales used in the UK and the NL what one would expect is comparable patterns ofcorrelations since these should be the same irrespective of the particular instrumentused We formulated a structural equation model analogous to the one used in the NLstudy examining the effects of self-esteem gender and duration of stay based onhomesickness and depression for the UK students (see Fig 4) This analysis revealedthat analogous to the NL data (Fig 2) homesickness is associated with depressionand that low self-esteem can be considered a risk factor for becoming homesick anddepressed To see whether or not homesickness is the precursor for self-esteem insteadof self-esteem being the precursor as hypothesized the alternative model was testedResults did not show a signicant effect of homesickness on self-esteem Thus low self-esteem like low emotional stability can rather be considered a vulnerability factorfor homesickness Again we see that after controlling for self-esteem the relationshipbetween homesickness and depression remains signicant This suggests that home-sickness is associated with depression irrespective of the level of self-esteem while ifone has low self-esteem the risk of becoming homesick and depressed is greater TheUK results also reveal the slight negative effect of duration of stay the longer studentslive in a new situation the lower the level of homesickness In contrast to the NLdata the UK data do reveal a gender difference Female students (M = 241 SD = 074)report higher levels of homesickness (F(278 1) = 1585 p lt 001) than male students(M = 204 SD = 060)

Analogous to the NL study we formulated a more detailed model including theUHS subscales (Fig 5) This revealed a similar pattern of correlations although thereare some obvious differences Adjustment difculties are assumed to be the crux of

Homesickness among students in two cultures 159

Figure 3 Detailed model of NL sample

the model and the data conrm this From adjustment difculties there are two pathsthat are connected indirectly to depression one via loneliness and the second viaruminations about home Self-esteem (a personality trait which was assumed to havesome commonality with emotional stability) has a similar relationship to adjustmentdifculties and depression in the UK sample as emotional stability in the NLsample Asdemonstrated previously with respect to the total homesickness score when analysedat the subscale level females in the UK study report having more adjustment difcultiesthan males This would indicate that the difference in homesickness mainly lies infemales reporting more adjustment difculties

Study 1 and Study 2 comparisonThere were a number of signicant differences between the two national samplesDutch students were older (M = 192 years) than British students (M = 188 yearst (478) = 295 p lt 01) and Dutch students (M = 1988 days) had spent a longer numberof days away from home than British students (M = 501 days t (224) = 1098 p lt 001)Furthermore there were proportionately more females in the NLsample than the UKsample (82versus 72) The implications of these differences are discussed later

HomesicknessAfter controlling for age and duration of relocation it was still found that the NLand UKstudents differed considerably with respect to homesickness with the UK studentsreporting more intense homesickness Univariate differences were found for three ofthe ve subscales (missing family missing friends adjustment difculties) and the totalhomesickness score Adjusted means for the subscales and for total homesickness forthe two samples are presented in Table 2

Furthermore when asked in the separate question about general feelings of home-sickness the UK students more frequently endorsed having felt homesick (see Table 3)

160 Margaret Stroebe et al

Figure 4 Model of UK sample

These analyses indicate not only that the UK students are more often homesick butthat they are so to a greater extent

Antecedents and consequences of homesicknessThe global models (Figs 2 and 4) demonstrate that homesickness is related to depres-sion in both cultures and that low self-esteem or emotional instability are linked tohigher levels of self-reported homesickness and depression More possibly howeverthe longer the stay in the new situation the less homesick and consequently lessdepressed one tends to be However the evidence also suggests that there may becultural differences with respect to gender differences in homesickness

The more detailed models (Figs 3 and 5) have the same basic structure althoughthere are some differences In the UK females are more likely to experience adjustmentdifculties than are males (b = 15) and the longer a UK student has experienced thenew situation the less he or she is inclined to miss the family (b = plusmn 12) effects whichare non-signicant in the NLstudy To enhance comparisons between the cultures thegeographical distance to the parental home (data not available in the NL sample) wasnot included here in the UKmodel However if included the pattern remained the sameand revealed that distance from home was only reliably associated with missing thefamily (b = 19)

Attachment style and family backgroundWe conducted further analyses to explore the association between attachment style andfamily background variables with homesickness In the NLstudy the effects of insecureattachment on relocation the homesickness features depression and emotional stability

Homesickness among students in two cultures 161

Figure 5 Detailed model of UK sample

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 5: res_6

(eg social support from home) would be unable to cope with the demands of the newsituation (eg academic performance) and would suffer homesickness and possiblydetrimental health consequences

Thus the DPM predicts that attention to and oscillation between these two tasksloss- and restoration-oriented are necessary for successful adaptation Similarly wewould predict that studentsrsquo well-being is dependent not only on the fact of havingleft home but also on successful adjustment to the new situation Thus the DPM pro-vides a theoretical basis for introducing two classes of variables into a structuralequation model for empirical investigation namely those to do with missing homefriends and so on and adjustment difculties in the new environment InterestinglyArcher et al (1998) recently identied two similar factors on a questionnaire measureof homesickness representing attachment to home and adjustment to the university

Purposes of the empirical studyTaken together what are the implications of this empirical and theoretical overviewfor further investigation of homesickness among students The review of empiricalstudies identies some emergent patterns of homesickness in relationship to personaland situational factors but some conicting results There is a need for replication ofndings and extension of the variables under investigation We have emphasized theneed for further investigation of the origins manifestations and consequences ofhomesickness among vulnerable groups From the outline above it is evident thatstudents leaving home for college belong in this category of vulnerable groups Lackingstill is not only information on predictors and outcomes but also on component partsof homesickness derived from the implementation of standard validation techniquesThus the rst aim of this study was to provide a detailed examination of homesick-ness among college students To this end two studies of homesickness were conductedone in the Netherlands and one in the UK

This two-study investigation enabled some examination of cultural patterns ofhomesickness To our knowledge this research provides the rst cross-cultural exam-ination of homesickness in both male and female students in their own countries Thusthe additional factor of acculturation (ie moving abroad and adjusting to the newcountry) does not occur in the examination of homesickness here Our research isalso unique in that it focuses on two cultures similar in terms of general societal back-ground in that they are both western European but different with respect to a centralattachment theory-related variable noted above namely accessibility to home (as willbe discussed later accessibility is greater in general in the Netherlands) As the abovereview indicated it is important to clarify the relationship between gender and home-sickness among student samples

A further advantage of the two reported studies is that they used the samestandardized homesickness measure In addition in each of the studies the relationshipof homesickness to a number of potentially important variables could be exploredincluding depression self-liking competence or self-esteem and secure or conictedattachment to the family The indices were standardized and thus provide usefulwithin-culture information which was the main focus of the analyses Between-culturecomparisons are to be made cautiously since some of the measures employed differedbetween the two countries (eg one standardized depression scale was used in theUK a different one in the Netherlands) However given the paucity of research todate and the intrinsic interest in comparing these phenomena across the two cultures

Homesickness among students in two cultures 151

as well as within each culture results are presented comparatively In addition anumber of demographic variables are analysed (eg gender age living arrangements)

Following the theoretical predictions outlined above it was expected that home-sickness would be prevalent and associated with distress and poorer mental health inthe student samples of both cultures but that prevalence would be greater in theculture with lower accessibility to home (UK) Following the DPM it was predictedthat missing home and difculty adjusting to the new situation would be antecedentto distress (rather than vice versa) Also that those who missed home more would bethe ones to have more difculty adapting to the new college situation (subscales onthe homesickness questionnaire) Furthermore it was predicted that homesicknesswould be greater among those living further from home irrespective of culture sincefollowing attachment theory the possibility of meeting up with signicant othersshould reduce separation anxiety and distress Again also following attachmenttheory it was predicted that students with greater parental conict1 or insecure attach-ment patterns would feel more homesick Homesickness was expected to occur in bothmale and female students of both cultures but it was predicted that female rateswould be higher (they could feel more comfortable admitting to homesickness) Finallyit was predicted that a longer time period since leaving home (or being at university)would be associated with lower homesickness (ie an adjustment effect) but that somestudents would still suffer homesickness even after being away for a longer period

Method

Study 1 The Netherlands (NL)

Respondents and procedureAll new-intake students entering the Faculty of Social Sciences of Utrecht University were mailed aquestionnaire containing the homesickness scale This took place in January 1995 Four weekslater a reminder was mailed to non-respondents This resulted in 482 completed and returnedquestionnaires (response rate 65) Eighty per cent of the respondents were female which issimilar to the proportion of females entering this faculty The mean age was 196 years (SD= 21)Twenty-nine per cent still lived at home with their parents 489had moved from the parentalhome to lodgings 196had recently moved from one place of lodging to another) and 26hadmoved back from lodgings to the parental home Most of the data analyses were conducted on the489relocated students while those living at home were included in specic analyses for controlpurposes (see below) The Utrecht students are referred to here as the NL sample

Measures

Socio-demographic informationRespondents were asked questions about their age gender previous living situation presentliving situation and the duration of their present living situation

The Utrecht Homesickness Scale (UHS)Designed to assess the extent of homesickness the UHS was developed in the following manner(see van Vliet Stroebe ampSchut 1998) Items were initiallyderived from previous research (Carden ampFeight 1991 Eurelings-Bontekoe et al 1994 Fisher Elder ampPeacock 1990 Fisher Frazer ampMurray

152 Margaret Stroebe et al

1 This prediction was tentative given that the evidence on con icted relationships is mixed (cf Archer 1999 pp 170ndash172)

1984 1985 1986 Fisher amp Hood 1987 1988) Items derived from a pilot study in which 100respondents were asked to give a denition of homesickness supplemented the original list

This preliminary questionnaire was given to 300 social science students who were asked torate the extent to which 51 items were associated with homesickness This resulted in a list of 45items This version of the questionnaire was then given to 117 third year psychology studentsThese students were asked to rate the extent to which theyhad experienced the 45 homesickness-related aspects in the past four weeks This led to a third version of the questionnaire which wasused in the Utrecht Homesickness Project The 45 items assess the extent to which studentsexperienced these various aspects of homesickness in the past 4 weeks (using the answercategories lsquonotrsquo lsquoweakrsquo lsquomoderatersquo lsquostrongrsquo lsquovery strongrsquo)

Aprincipal component analysis (minimum eigenvalue gt 1) with a varimax rotation over the 45items was computed for students who had felt homesick in the past four weeks (N = 151)resulting in nine factors On the basis of a scree plot we used the rst ve factors which wereclearly interpretable and explained 57of the variance These ve factors correspond closely tothe features that according to most authors constitute homesickness missing family missingfriends having adjustment difculties ruminations about home and feeling lonely

Based on these ve factors a reduced set of items (20) was selected to ensure that thedeterminant of the correlation matrix would be gt 0 This is necessary to avoid co-linearity of setsof items Asecond principal component analysis with a varimax rotation over these 20 items andthe restriction that ve factors should be extracted revealed that the items loaded as expected onthe relevant ve factors This last procedure was repeated with all the students (N = 439) andrevealed the same factor loading pattern Within this set of 20 variables the ve factors explained73 of the variance On the basis of the ve-factor structure scales were computed resultingin very acceptable Cronbachrsquos alphas (see Table 1) This set of subscales will be referred to as theUtrecht Homesickness Scale (UHS) Note that if there were missing values lsquolist-wise deletionrsquo wasused

Additional measuresStudents were also asked how often they had experienced homesickness in the past 4 weeks(never rarely sometimes often very often) Adiscriminant analysis whereby the ve UHS scalesare used to predict whether or not students endorsed feeling homesick (never versus rarelysometimes often very often) in the past 4 weeks revealed that 83 were correctly classiedFurthermore there was a strong correlation between the total homesickness score and thefrequency of homesickness in the past 4 weeks (r = 71) which indicates that the UHS scale isclosely related to a direct measure of homesickness even though the scale does not mentionhomesickness specically The Dutch version of the Symptoms Check List (SCL-90 Arrindell ampEttema 1975) was used as an indicator of mental health status of the NL sample Only thedepressive mood subscale (a = 92) and the total health score (a = 97) were used for this part ofthe investigation

Personality factors were measured by means of a Dutch translation of the Golberg Bi-polar Big5 (De Jong van Eck ampvan den Bos 1994 eg Goldberg 1992) Only emotional stability (a = 85)was used for this investigation

A Dutch translation of the Attachment Style Measure (Hazan amp Shaver 1987) was used toassess insecure (that is avoidant and anxiousambivalent see Hazan ampShaver 1987) versus secureattachment This particular scale was chosen because of its conciseness and applicability for usewith these student samples

Statistical analysesComparable to what was done in the pre-test described previously we rst conducted factoranalysis on the homesickness scale which identied the ve subscales A principal componentanalysis (minimum eigenvalue gt 1) with a varimax rotation over the 45 items resulted in ninefactors On the basis of a scree plot we used the rst ve factors which were clearly interpretable

Homesickness among students in two cultures 153

and explained 57of the variance These ve factors correspond closely to the factors identiedin the pre-test on the UHS namely missing family missing friends having adjustment difcultiesruminations about home and feeling lonely

Then we carried out a multivariate analysis of variance (MANOVA) to test differences inhomesickness scores between the UK and NL samples We then computed path analyses (usingLISREL81) in order to examine multiple relationships simultaneously We hypothesized identical

154 Margaret Stroebe et al

Table 1 Utrecht Homesickness Scale factor structure and reliabilities (NL sample)

Factor

(1) (2) (3) (4) (5)

Percentage of explained varianceItem

62 15 9 8 6

Factor loadings

(1) Missing family (a = 90)Missing your parents 89 12 20 06 16Missing your family 87 13 19 09 14Missing home 74 12 21 18 37Feeling missed by your family 74 15 19 15 04

(2) Loneliness (a = 85)Feeling lonely 21 78 22 25 09Feeling unloved 12 75 22 15 07Feeling isolated from the rest of the world 09 72 14 33 22Feeling uprooted 08 64 21 35 22

(3) Missing friends (a = 87)Longing for acquaintances 16 22 83 19 16Searching for familiar faces 19 18 73 23 12Missing people whom you trust and can talk with 25 31 71 07 20Missing your friends 36 14 70 15 14

(4) Adjustment dif culties (a = 88)Finding it dif cult adjusting to a new situation 23 27 15 77 16Feeling uncomfortable in a new situation 15 44 06 71 23Feeling lost in the new situation 17 46 22 69 21Having dif culties in getting used to new customs 03 14 38 68 26

(5) Ruminations about home (a = 80)Having thoughts that an old situation was better thanhere and now 16 15 18 28 81Regretting the decision to leave an old situation 10 21 08 31 79Continuously having thoughts about home 41 08 24 13 68Repeatedly thinking of the past 18 45 28 plusmn 01 52

Total homesickness (a = 94) (20 items)

Note The questions were formulated in the following manner Could you please indicate to what extentyou have experienced the following in the past 4 weeks Missing your parents notndashweakndashmoderatendashstrongndashvery strong N = 439

models for both cultures assuming that the patterns of associations would be similar Weexamined models of the relationships between homesickness and its putative antecedents andconsequences in terms of both overall scores and subscale scores Precise details of these modelsare given in the results section

Several t indices as described by Jaccard and Wan (1996) are appropriate for our investigationof path models To demonstrate the overall degree of correspondence between predicted andobserved covariance matrices the x2 statistic with p gt 05 indicates that a perfect t model doesexist in the population The goodness-of-t index (GFI gt 90) indicates goodness of t Thestandardized root mean square residual (SRMR lt 10) indicates a deviation of less than 10correlation units on average between predicted and observed correlations The root mean squareerror of approximation (RMSEA lt 08) implies an adequate model and lt 05 implies a good modelwith the p value for close t (CFit) being non-signicant (gt 05) CFI gt 90 implies a good t

Study 2 The United Kingdom (UK)

Respondents and procedureDuring the autumn term of 1996 new-intake students at Cardiff University were askedto participate in a student survey Questionnaires were administered in October and Novemberand 280 completed questionnaires were returned representing approximately a 60 responserate Seventy-two per cent of the respondents were female The mean age was 188 years(SD = 21) All respondents had moved from the parental home to lodgings on campus TheCardiff students are referred to here as the UK sample

MeasuresAn English translation of the Utrecht Homesickness Scale (UHS) was used to measure the extent ofhomesickness The reliabilities of the scales were comparable to the NLsample missing the family(a = 85) loneliness (a = 84) missing friends (a = 78) adjustment difculties (a = 84)ruminations about home (a = 86) and the total score for homesickness (a = 93) Againrespondents were also asked to indicate how often they had felt homesick during the past 4weeks The Beck Depression Inventory (BDI Beck Ward Mendelson Mock amp Erbaugh 1961)was used to give a reliable measure of current level of depression (a = 82) The Self-likingSelf-competence Scale was used to assess self-esteem (a = 91) (cf Tafarodi amp Swann 1995) Anadapted version of the Conict Tactics Scale (cf Straus Hamby Boney-McCoy ampSugarman 1996)was used to ascertain type of family background of the UK respondents The Conict Tactics Scaleconsists of six subscales with similar acceptable alpha levels (on average 72) which assess theextent to which parents are perceived as handling conicts with the respondents in certain waysThe items used here reected three factors that is verbal reasoning (three items) verbalaggression (six items) and physical aggression (three items) of either father or mother Selectionof these items from the original much longer scale was thought to be less intrusive and easier forrespondents to answer while still providing a measure of the three factors The students were alsoasked to state the name of their home town On the basis of this information distances could becalculated between their home and their university lodgings

Statistical analysesThe analyses described above for Study 1 apply here too

ResultsThe results section is structured in three parts In the rst section results on the NLsample are reported In the second section the same is done for the UK sample In thelast section a comparison is made between the two samples

Homesickness among students in two cultures 155

Study 1 NL

HomesicknessAlmost 50of the students reported having felt homesick at least some of the time afterhaving started their new education (N = 230) There were no differences betweenmales and females with respect to homesickness intensity The intensity of home-sickness as measured by the UHS subscales (range 1 (not experienced) to 5 (verystrongly)) revealed that on average the highest scores were obtained for missing friends(M = 238 SD = 102) and missing the family (M = 215 SD = 095)2 The otherdimensions scored were adjustment difculties (M = 180 SD = 088) ruminationsabout home (M = 152 SD = 071) and loneliness (M = 179 SD = 087) Theseaveraged to a total mean score of 193 (SD = 071)

Antecedents and consequences of homesicknessThe NL data set offered the possibility of comparisons between students still living athome and those who had relocated3 This provided the opportunity to test whetherrelocation (R) would indeed be associated with depression (D) which in and of itselfwe assumed would be associated with homesickness (H) We must be cautious giventhe cross-sectional design of our studies in assuming mediation between the RndashDndashHvariables Based on Baron and Kenny (1986) a requirement for mediation with this dataset would be that relocation and depression are signicantly associated in other wordsthere must be an effect to be mediated In our data we nd that relocation anddepression are not signicantly associated and therefore a mediation analysis is notappropriate Nevertheless a path analysis is possible if we regress depression on re-location and homesickness and regress homesickness on relocation (see Fig 1) Ourresults show a signicant path from relocation to homesickness (b = 21) a signicantpath from homesickness to depression controlling for relocation (b = 62) and a non-signicant direct path from relocation to depression (b = plusmn 05) Showing that thetwo component paths of the indirect effect are each signicant is not the same asshowing that the indirect effect is signicant Testing for an indirect path (Sobel 1982)revealed that the indirect effect of relocation on depression is signicant (b = 13) Itseems reasonable to conclude then that the data are consistent with an indirect effectof relocation on depression by means of the intervening variable homesickness Whilethere are good reasons to argue that there is an indirect effect of relocation ondepression through homesickness it is evident that this effect though signicant isquite small and not strong enough to produce an overall effect of R on D

The model shown in Fig 1 was also tested using the SCL-90 total score as the healthmeasure instead of using the depression subscale This showed an almost identicalsolution The analysis provides further evidence that the assumed pattern is justiedand that it is not just due to some correspondence between depression and home-sickness The pattern suggests that relocation may be an antecedent for homesicknessbut not for depressive moods or general ill-health and that homesickness may bean important antecedent for ill-health Having found evidence for this relationshipstructural equation models could be tested with respect to the effects of emotionalstability gender and duration of stay on homesickness and depression (see Fig 2) The

156 Margaret Stroebe et al

2 Controlling for age and duration of stay in order to make comparisons with the UK sample possible3 Some students who had not moved also reported being homesick (15) in the past 4 weeks The vast majority of them(90) reported that they experienced homesickness only lsquorarelyrsquo or lsquosometimesrsquo

literature suggests that emotional instability is a good predictor for both homesickness(Eurelings-Bontekoe et al 1996) and depression (cf Watson ampPennebaker 1989) Thusone could assume that emotional stability (the opposite of emotional instability) wouldexplain the variance of depression However we nd that even though reducedhomesickness retains an effect on depression This suggests that irrespective of thelevel of emotional stability homesickness is a precursor for depression Furthermorethe emotionally unstable as expected have higher levels of homesickness and depres-sion What we also nd is that the longer one is away from home the less intense thehomesickness reaction which suggests that adaptation to the new situation takes place

In order to create a more comprehensive picture a model was formulated whichincluded the subscales of the UHS instead of the total homesickness score Theassumptions were that (1) adjustment difculties would be associated with missingfamily and friends ruminations about home and loneliness (2) missing family andfriends in turn would be associated with ruminations about home and loneliness (3)ruminations would also be associated to loneliness and depression and (4) lonelinessin turn to depression Results with respect to this model revealed that in principle theassumptions were correct if two modications were incorporated These wereestablishing a link between missing family and friends and removing the path frommissing family to loneliness The choice was made to let the errors of missing friendsand family correlate because no prior hypotheses were made as to a causal relation-ship between both features and to delete the non-signicant path4 Figure 3 indicateshow the homesickness subscales are related and how these are linked to the measuredantecedents and consequences To enhance the visual impact of the model the thick-ness of paths corresponds to the size of the path coefcients

Figure 3 suggests that to become depressed one has to experience adjustment

Homesickness among students in two cultures 157

Figure 1 Model of relocation homesickness and depression for NL students

4 For the sake of comparison with the forthcoming UK model the non-signicant paths between missing friends and lonelinessgender and adjustment dif culties and duration of stay and missing family were not removed from the model

difculties and consequently loneliness and to ruminate about home If the relocatedstudents experience severe adjustment difculties these do seem to go hand in handwith missing family and friends ruminations about home and loneliness Specicallyloneliness is most strongly associated with depression Furthermore the analysessuggest that the best predictor for adjustment difculties in this study is emotionalstability (b = plusmn 41) Emotional stability also has a strong relationship to depression(b = plusmn 45) and to a lesser extent to missing friends (b = plusmn 20) and loneliness(b = plusmn 15) The slight association of duration of stay with total homesickness dis-appeared when its relationship was tested on the ve separate homesickness subscales

Study 2 UK

HomesicknessMore than 80of the students reported having felt homesick at least some of the timeafter having started their new education (N = 280) The intensity of homesicknessas measured by the UHS subscales (likewise controlling for age and the number of daysaway from home) revealed that on average the highest scores in the UK were obtainedfor missing family (M = 293 SD = 097) and missing friends (M = 272 SD = 094)The other dimensions scored were adjustment difculties (M = 205 SD = 084)ruminations about home (M = 166 SD = 086) and loneliness (M = 195SD = 092) These averaged to a total mean score of 226 (SD = 073)

Antecedents and consequences of homesicknessAlthough the hypothesis could not be tested with the UKsample it seems reasonable toassume based on the NLdata that relocation would be associated with homesicknesswhich in turn would be associated with depression and not vice versa (of course similarcautions about mediating variables to those expressed for Study 1 are in order here too)

158 Margaret Stroebe et al

Figure 2 Model of NL sample

Accepting this premise and despite the differences in the depression and personalityscales used in the UK and the NL what one would expect is comparable patterns ofcorrelations since these should be the same irrespective of the particular instrumentused We formulated a structural equation model analogous to the one used in the NLstudy examining the effects of self-esteem gender and duration of stay based onhomesickness and depression for the UK students (see Fig 4) This analysis revealedthat analogous to the NL data (Fig 2) homesickness is associated with depressionand that low self-esteem can be considered a risk factor for becoming homesick anddepressed To see whether or not homesickness is the precursor for self-esteem insteadof self-esteem being the precursor as hypothesized the alternative model was testedResults did not show a signicant effect of homesickness on self-esteem Thus low self-esteem like low emotional stability can rather be considered a vulnerability factorfor homesickness Again we see that after controlling for self-esteem the relationshipbetween homesickness and depression remains signicant This suggests that home-sickness is associated with depression irrespective of the level of self-esteem while ifone has low self-esteem the risk of becoming homesick and depressed is greater TheUK results also reveal the slight negative effect of duration of stay the longer studentslive in a new situation the lower the level of homesickness In contrast to the NLdata the UK data do reveal a gender difference Female students (M = 241 SD = 074)report higher levels of homesickness (F(278 1) = 1585 p lt 001) than male students(M = 204 SD = 060)

Analogous to the NL study we formulated a more detailed model including theUHS subscales (Fig 5) This revealed a similar pattern of correlations although thereare some obvious differences Adjustment difculties are assumed to be the crux of

Homesickness among students in two cultures 159

Figure 3 Detailed model of NL sample

the model and the data conrm this From adjustment difculties there are two pathsthat are connected indirectly to depression one via loneliness and the second viaruminations about home Self-esteem (a personality trait which was assumed to havesome commonality with emotional stability) has a similar relationship to adjustmentdifculties and depression in the UK sample as emotional stability in the NLsample Asdemonstrated previously with respect to the total homesickness score when analysedat the subscale level females in the UK study report having more adjustment difcultiesthan males This would indicate that the difference in homesickness mainly lies infemales reporting more adjustment difculties

Study 1 and Study 2 comparisonThere were a number of signicant differences between the two national samplesDutch students were older (M = 192 years) than British students (M = 188 yearst (478) = 295 p lt 01) and Dutch students (M = 1988 days) had spent a longer numberof days away from home than British students (M = 501 days t (224) = 1098 p lt 001)Furthermore there were proportionately more females in the NLsample than the UKsample (82versus 72) The implications of these differences are discussed later

HomesicknessAfter controlling for age and duration of relocation it was still found that the NLand UKstudents differed considerably with respect to homesickness with the UK studentsreporting more intense homesickness Univariate differences were found for three ofthe ve subscales (missing family missing friends adjustment difculties) and the totalhomesickness score Adjusted means for the subscales and for total homesickness forthe two samples are presented in Table 2

Furthermore when asked in the separate question about general feelings of home-sickness the UK students more frequently endorsed having felt homesick (see Table 3)

160 Margaret Stroebe et al

Figure 4 Model of UK sample

These analyses indicate not only that the UK students are more often homesick butthat they are so to a greater extent

Antecedents and consequences of homesicknessThe global models (Figs 2 and 4) demonstrate that homesickness is related to depres-sion in both cultures and that low self-esteem or emotional instability are linked tohigher levels of self-reported homesickness and depression More possibly howeverthe longer the stay in the new situation the less homesick and consequently lessdepressed one tends to be However the evidence also suggests that there may becultural differences with respect to gender differences in homesickness

The more detailed models (Figs 3 and 5) have the same basic structure althoughthere are some differences In the UK females are more likely to experience adjustmentdifculties than are males (b = 15) and the longer a UK student has experienced thenew situation the less he or she is inclined to miss the family (b = plusmn 12) effects whichare non-signicant in the NLstudy To enhance comparisons between the cultures thegeographical distance to the parental home (data not available in the NL sample) wasnot included here in the UKmodel However if included the pattern remained the sameand revealed that distance from home was only reliably associated with missing thefamily (b = 19)

Attachment style and family backgroundWe conducted further analyses to explore the association between attachment style andfamily background variables with homesickness In the NLstudy the effects of insecureattachment on relocation the homesickness features depression and emotional stability

Homesickness among students in two cultures 161

Figure 5 Detailed model of UK sample

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 6: res_6

as well as within each culture results are presented comparatively In addition anumber of demographic variables are analysed (eg gender age living arrangements)

Following the theoretical predictions outlined above it was expected that home-sickness would be prevalent and associated with distress and poorer mental health inthe student samples of both cultures but that prevalence would be greater in theculture with lower accessibility to home (UK) Following the DPM it was predictedthat missing home and difculty adjusting to the new situation would be antecedentto distress (rather than vice versa) Also that those who missed home more would bethe ones to have more difculty adapting to the new college situation (subscales onthe homesickness questionnaire) Furthermore it was predicted that homesicknesswould be greater among those living further from home irrespective of culture sincefollowing attachment theory the possibility of meeting up with signicant othersshould reduce separation anxiety and distress Again also following attachmenttheory it was predicted that students with greater parental conict1 or insecure attach-ment patterns would feel more homesick Homesickness was expected to occur in bothmale and female students of both cultures but it was predicted that female rateswould be higher (they could feel more comfortable admitting to homesickness) Finallyit was predicted that a longer time period since leaving home (or being at university)would be associated with lower homesickness (ie an adjustment effect) but that somestudents would still suffer homesickness even after being away for a longer period

Method

Study 1 The Netherlands (NL)

Respondents and procedureAll new-intake students entering the Faculty of Social Sciences of Utrecht University were mailed aquestionnaire containing the homesickness scale This took place in January 1995 Four weekslater a reminder was mailed to non-respondents This resulted in 482 completed and returnedquestionnaires (response rate 65) Eighty per cent of the respondents were female which issimilar to the proportion of females entering this faculty The mean age was 196 years (SD= 21)Twenty-nine per cent still lived at home with their parents 489had moved from the parentalhome to lodgings 196had recently moved from one place of lodging to another) and 26hadmoved back from lodgings to the parental home Most of the data analyses were conducted on the489relocated students while those living at home were included in specic analyses for controlpurposes (see below) The Utrecht students are referred to here as the NL sample

Measures

Socio-demographic informationRespondents were asked questions about their age gender previous living situation presentliving situation and the duration of their present living situation

The Utrecht Homesickness Scale (UHS)Designed to assess the extent of homesickness the UHS was developed in the following manner(see van Vliet Stroebe ampSchut 1998) Items were initiallyderived from previous research (Carden ampFeight 1991 Eurelings-Bontekoe et al 1994 Fisher Elder ampPeacock 1990 Fisher Frazer ampMurray

152 Margaret Stroebe et al

1 This prediction was tentative given that the evidence on con icted relationships is mixed (cf Archer 1999 pp 170ndash172)

1984 1985 1986 Fisher amp Hood 1987 1988) Items derived from a pilot study in which 100respondents were asked to give a denition of homesickness supplemented the original list

This preliminary questionnaire was given to 300 social science students who were asked torate the extent to which 51 items were associated with homesickness This resulted in a list of 45items This version of the questionnaire was then given to 117 third year psychology studentsThese students were asked to rate the extent to which theyhad experienced the 45 homesickness-related aspects in the past four weeks This led to a third version of the questionnaire which wasused in the Utrecht Homesickness Project The 45 items assess the extent to which studentsexperienced these various aspects of homesickness in the past 4 weeks (using the answercategories lsquonotrsquo lsquoweakrsquo lsquomoderatersquo lsquostrongrsquo lsquovery strongrsquo)

Aprincipal component analysis (minimum eigenvalue gt 1) with a varimax rotation over the 45items was computed for students who had felt homesick in the past four weeks (N = 151)resulting in nine factors On the basis of a scree plot we used the rst ve factors which wereclearly interpretable and explained 57of the variance These ve factors correspond closely tothe features that according to most authors constitute homesickness missing family missingfriends having adjustment difculties ruminations about home and feeling lonely

Based on these ve factors a reduced set of items (20) was selected to ensure that thedeterminant of the correlation matrix would be gt 0 This is necessary to avoid co-linearity of setsof items Asecond principal component analysis with a varimax rotation over these 20 items andthe restriction that ve factors should be extracted revealed that the items loaded as expected onthe relevant ve factors This last procedure was repeated with all the students (N = 439) andrevealed the same factor loading pattern Within this set of 20 variables the ve factors explained73 of the variance On the basis of the ve-factor structure scales were computed resultingin very acceptable Cronbachrsquos alphas (see Table 1) This set of subscales will be referred to as theUtrecht Homesickness Scale (UHS) Note that if there were missing values lsquolist-wise deletionrsquo wasused

Additional measuresStudents were also asked how often they had experienced homesickness in the past 4 weeks(never rarely sometimes often very often) Adiscriminant analysis whereby the ve UHS scalesare used to predict whether or not students endorsed feeling homesick (never versus rarelysometimes often very often) in the past 4 weeks revealed that 83 were correctly classiedFurthermore there was a strong correlation between the total homesickness score and thefrequency of homesickness in the past 4 weeks (r = 71) which indicates that the UHS scale isclosely related to a direct measure of homesickness even though the scale does not mentionhomesickness specically The Dutch version of the Symptoms Check List (SCL-90 Arrindell ampEttema 1975) was used as an indicator of mental health status of the NL sample Only thedepressive mood subscale (a = 92) and the total health score (a = 97) were used for this part ofthe investigation

Personality factors were measured by means of a Dutch translation of the Golberg Bi-polar Big5 (De Jong van Eck ampvan den Bos 1994 eg Goldberg 1992) Only emotional stability (a = 85)was used for this investigation

A Dutch translation of the Attachment Style Measure (Hazan amp Shaver 1987) was used toassess insecure (that is avoidant and anxiousambivalent see Hazan ampShaver 1987) versus secureattachment This particular scale was chosen because of its conciseness and applicability for usewith these student samples

Statistical analysesComparable to what was done in the pre-test described previously we rst conducted factoranalysis on the homesickness scale which identied the ve subscales A principal componentanalysis (minimum eigenvalue gt 1) with a varimax rotation over the 45 items resulted in ninefactors On the basis of a scree plot we used the rst ve factors which were clearly interpretable

Homesickness among students in two cultures 153

and explained 57of the variance These ve factors correspond closely to the factors identiedin the pre-test on the UHS namely missing family missing friends having adjustment difcultiesruminations about home and feeling lonely

Then we carried out a multivariate analysis of variance (MANOVA) to test differences inhomesickness scores between the UK and NL samples We then computed path analyses (usingLISREL81) in order to examine multiple relationships simultaneously We hypothesized identical

154 Margaret Stroebe et al

Table 1 Utrecht Homesickness Scale factor structure and reliabilities (NL sample)

Factor

(1) (2) (3) (4) (5)

Percentage of explained varianceItem

62 15 9 8 6

Factor loadings

(1) Missing family (a = 90)Missing your parents 89 12 20 06 16Missing your family 87 13 19 09 14Missing home 74 12 21 18 37Feeling missed by your family 74 15 19 15 04

(2) Loneliness (a = 85)Feeling lonely 21 78 22 25 09Feeling unloved 12 75 22 15 07Feeling isolated from the rest of the world 09 72 14 33 22Feeling uprooted 08 64 21 35 22

(3) Missing friends (a = 87)Longing for acquaintances 16 22 83 19 16Searching for familiar faces 19 18 73 23 12Missing people whom you trust and can talk with 25 31 71 07 20Missing your friends 36 14 70 15 14

(4) Adjustment dif culties (a = 88)Finding it dif cult adjusting to a new situation 23 27 15 77 16Feeling uncomfortable in a new situation 15 44 06 71 23Feeling lost in the new situation 17 46 22 69 21Having dif culties in getting used to new customs 03 14 38 68 26

(5) Ruminations about home (a = 80)Having thoughts that an old situation was better thanhere and now 16 15 18 28 81Regretting the decision to leave an old situation 10 21 08 31 79Continuously having thoughts about home 41 08 24 13 68Repeatedly thinking of the past 18 45 28 plusmn 01 52

Total homesickness (a = 94) (20 items)

Note The questions were formulated in the following manner Could you please indicate to what extentyou have experienced the following in the past 4 weeks Missing your parents notndashweakndashmoderatendashstrongndashvery strong N = 439

models for both cultures assuming that the patterns of associations would be similar Weexamined models of the relationships between homesickness and its putative antecedents andconsequences in terms of both overall scores and subscale scores Precise details of these modelsare given in the results section

Several t indices as described by Jaccard and Wan (1996) are appropriate for our investigationof path models To demonstrate the overall degree of correspondence between predicted andobserved covariance matrices the x2 statistic with p gt 05 indicates that a perfect t model doesexist in the population The goodness-of-t index (GFI gt 90) indicates goodness of t Thestandardized root mean square residual (SRMR lt 10) indicates a deviation of less than 10correlation units on average between predicted and observed correlations The root mean squareerror of approximation (RMSEA lt 08) implies an adequate model and lt 05 implies a good modelwith the p value for close t (CFit) being non-signicant (gt 05) CFI gt 90 implies a good t

Study 2 The United Kingdom (UK)

Respondents and procedureDuring the autumn term of 1996 new-intake students at Cardiff University were askedto participate in a student survey Questionnaires were administered in October and Novemberand 280 completed questionnaires were returned representing approximately a 60 responserate Seventy-two per cent of the respondents were female The mean age was 188 years(SD = 21) All respondents had moved from the parental home to lodgings on campus TheCardiff students are referred to here as the UK sample

MeasuresAn English translation of the Utrecht Homesickness Scale (UHS) was used to measure the extent ofhomesickness The reliabilities of the scales were comparable to the NLsample missing the family(a = 85) loneliness (a = 84) missing friends (a = 78) adjustment difculties (a = 84)ruminations about home (a = 86) and the total score for homesickness (a = 93) Againrespondents were also asked to indicate how often they had felt homesick during the past 4weeks The Beck Depression Inventory (BDI Beck Ward Mendelson Mock amp Erbaugh 1961)was used to give a reliable measure of current level of depression (a = 82) The Self-likingSelf-competence Scale was used to assess self-esteem (a = 91) (cf Tafarodi amp Swann 1995) Anadapted version of the Conict Tactics Scale (cf Straus Hamby Boney-McCoy ampSugarman 1996)was used to ascertain type of family background of the UK respondents The Conict Tactics Scaleconsists of six subscales with similar acceptable alpha levels (on average 72) which assess theextent to which parents are perceived as handling conicts with the respondents in certain waysThe items used here reected three factors that is verbal reasoning (three items) verbalaggression (six items) and physical aggression (three items) of either father or mother Selectionof these items from the original much longer scale was thought to be less intrusive and easier forrespondents to answer while still providing a measure of the three factors The students were alsoasked to state the name of their home town On the basis of this information distances could becalculated between their home and their university lodgings

Statistical analysesThe analyses described above for Study 1 apply here too

ResultsThe results section is structured in three parts In the rst section results on the NLsample are reported In the second section the same is done for the UK sample In thelast section a comparison is made between the two samples

Homesickness among students in two cultures 155

Study 1 NL

HomesicknessAlmost 50of the students reported having felt homesick at least some of the time afterhaving started their new education (N = 230) There were no differences betweenmales and females with respect to homesickness intensity The intensity of home-sickness as measured by the UHS subscales (range 1 (not experienced) to 5 (verystrongly)) revealed that on average the highest scores were obtained for missing friends(M = 238 SD = 102) and missing the family (M = 215 SD = 095)2 The otherdimensions scored were adjustment difculties (M = 180 SD = 088) ruminationsabout home (M = 152 SD = 071) and loneliness (M = 179 SD = 087) Theseaveraged to a total mean score of 193 (SD = 071)

Antecedents and consequences of homesicknessThe NL data set offered the possibility of comparisons between students still living athome and those who had relocated3 This provided the opportunity to test whetherrelocation (R) would indeed be associated with depression (D) which in and of itselfwe assumed would be associated with homesickness (H) We must be cautious giventhe cross-sectional design of our studies in assuming mediation between the RndashDndashHvariables Based on Baron and Kenny (1986) a requirement for mediation with this dataset would be that relocation and depression are signicantly associated in other wordsthere must be an effect to be mediated In our data we nd that relocation anddepression are not signicantly associated and therefore a mediation analysis is notappropriate Nevertheless a path analysis is possible if we regress depression on re-location and homesickness and regress homesickness on relocation (see Fig 1) Ourresults show a signicant path from relocation to homesickness (b = 21) a signicantpath from homesickness to depression controlling for relocation (b = 62) and a non-signicant direct path from relocation to depression (b = plusmn 05) Showing that thetwo component paths of the indirect effect are each signicant is not the same asshowing that the indirect effect is signicant Testing for an indirect path (Sobel 1982)revealed that the indirect effect of relocation on depression is signicant (b = 13) Itseems reasonable to conclude then that the data are consistent with an indirect effectof relocation on depression by means of the intervening variable homesickness Whilethere are good reasons to argue that there is an indirect effect of relocation ondepression through homesickness it is evident that this effect though signicant isquite small and not strong enough to produce an overall effect of R on D

The model shown in Fig 1 was also tested using the SCL-90 total score as the healthmeasure instead of using the depression subscale This showed an almost identicalsolution The analysis provides further evidence that the assumed pattern is justiedand that it is not just due to some correspondence between depression and home-sickness The pattern suggests that relocation may be an antecedent for homesicknessbut not for depressive moods or general ill-health and that homesickness may bean important antecedent for ill-health Having found evidence for this relationshipstructural equation models could be tested with respect to the effects of emotionalstability gender and duration of stay on homesickness and depression (see Fig 2) The

156 Margaret Stroebe et al

2 Controlling for age and duration of stay in order to make comparisons with the UK sample possible3 Some students who had not moved also reported being homesick (15) in the past 4 weeks The vast majority of them(90) reported that they experienced homesickness only lsquorarelyrsquo or lsquosometimesrsquo

literature suggests that emotional instability is a good predictor for both homesickness(Eurelings-Bontekoe et al 1996) and depression (cf Watson ampPennebaker 1989) Thusone could assume that emotional stability (the opposite of emotional instability) wouldexplain the variance of depression However we nd that even though reducedhomesickness retains an effect on depression This suggests that irrespective of thelevel of emotional stability homesickness is a precursor for depression Furthermorethe emotionally unstable as expected have higher levels of homesickness and depres-sion What we also nd is that the longer one is away from home the less intense thehomesickness reaction which suggests that adaptation to the new situation takes place

In order to create a more comprehensive picture a model was formulated whichincluded the subscales of the UHS instead of the total homesickness score Theassumptions were that (1) adjustment difculties would be associated with missingfamily and friends ruminations about home and loneliness (2) missing family andfriends in turn would be associated with ruminations about home and loneliness (3)ruminations would also be associated to loneliness and depression and (4) lonelinessin turn to depression Results with respect to this model revealed that in principle theassumptions were correct if two modications were incorporated These wereestablishing a link between missing family and friends and removing the path frommissing family to loneliness The choice was made to let the errors of missing friendsand family correlate because no prior hypotheses were made as to a causal relation-ship between both features and to delete the non-signicant path4 Figure 3 indicateshow the homesickness subscales are related and how these are linked to the measuredantecedents and consequences To enhance the visual impact of the model the thick-ness of paths corresponds to the size of the path coefcients

Figure 3 suggests that to become depressed one has to experience adjustment

Homesickness among students in two cultures 157

Figure 1 Model of relocation homesickness and depression for NL students

4 For the sake of comparison with the forthcoming UK model the non-signicant paths between missing friends and lonelinessgender and adjustment dif culties and duration of stay and missing family were not removed from the model

difculties and consequently loneliness and to ruminate about home If the relocatedstudents experience severe adjustment difculties these do seem to go hand in handwith missing family and friends ruminations about home and loneliness Specicallyloneliness is most strongly associated with depression Furthermore the analysessuggest that the best predictor for adjustment difculties in this study is emotionalstability (b = plusmn 41) Emotional stability also has a strong relationship to depression(b = plusmn 45) and to a lesser extent to missing friends (b = plusmn 20) and loneliness(b = plusmn 15) The slight association of duration of stay with total homesickness dis-appeared when its relationship was tested on the ve separate homesickness subscales

Study 2 UK

HomesicknessMore than 80of the students reported having felt homesick at least some of the timeafter having started their new education (N = 280) The intensity of homesicknessas measured by the UHS subscales (likewise controlling for age and the number of daysaway from home) revealed that on average the highest scores in the UK were obtainedfor missing family (M = 293 SD = 097) and missing friends (M = 272 SD = 094)The other dimensions scored were adjustment difculties (M = 205 SD = 084)ruminations about home (M = 166 SD = 086) and loneliness (M = 195SD = 092) These averaged to a total mean score of 226 (SD = 073)

Antecedents and consequences of homesicknessAlthough the hypothesis could not be tested with the UKsample it seems reasonable toassume based on the NLdata that relocation would be associated with homesicknesswhich in turn would be associated with depression and not vice versa (of course similarcautions about mediating variables to those expressed for Study 1 are in order here too)

158 Margaret Stroebe et al

Figure 2 Model of NL sample

Accepting this premise and despite the differences in the depression and personalityscales used in the UK and the NL what one would expect is comparable patterns ofcorrelations since these should be the same irrespective of the particular instrumentused We formulated a structural equation model analogous to the one used in the NLstudy examining the effects of self-esteem gender and duration of stay based onhomesickness and depression for the UK students (see Fig 4) This analysis revealedthat analogous to the NL data (Fig 2) homesickness is associated with depressionand that low self-esteem can be considered a risk factor for becoming homesick anddepressed To see whether or not homesickness is the precursor for self-esteem insteadof self-esteem being the precursor as hypothesized the alternative model was testedResults did not show a signicant effect of homesickness on self-esteem Thus low self-esteem like low emotional stability can rather be considered a vulnerability factorfor homesickness Again we see that after controlling for self-esteem the relationshipbetween homesickness and depression remains signicant This suggests that home-sickness is associated with depression irrespective of the level of self-esteem while ifone has low self-esteem the risk of becoming homesick and depressed is greater TheUK results also reveal the slight negative effect of duration of stay the longer studentslive in a new situation the lower the level of homesickness In contrast to the NLdata the UK data do reveal a gender difference Female students (M = 241 SD = 074)report higher levels of homesickness (F(278 1) = 1585 p lt 001) than male students(M = 204 SD = 060)

Analogous to the NL study we formulated a more detailed model including theUHS subscales (Fig 5) This revealed a similar pattern of correlations although thereare some obvious differences Adjustment difculties are assumed to be the crux of

Homesickness among students in two cultures 159

Figure 3 Detailed model of NL sample

the model and the data conrm this From adjustment difculties there are two pathsthat are connected indirectly to depression one via loneliness and the second viaruminations about home Self-esteem (a personality trait which was assumed to havesome commonality with emotional stability) has a similar relationship to adjustmentdifculties and depression in the UK sample as emotional stability in the NLsample Asdemonstrated previously with respect to the total homesickness score when analysedat the subscale level females in the UK study report having more adjustment difcultiesthan males This would indicate that the difference in homesickness mainly lies infemales reporting more adjustment difculties

Study 1 and Study 2 comparisonThere were a number of signicant differences between the two national samplesDutch students were older (M = 192 years) than British students (M = 188 yearst (478) = 295 p lt 01) and Dutch students (M = 1988 days) had spent a longer numberof days away from home than British students (M = 501 days t (224) = 1098 p lt 001)Furthermore there were proportionately more females in the NLsample than the UKsample (82versus 72) The implications of these differences are discussed later

HomesicknessAfter controlling for age and duration of relocation it was still found that the NLand UKstudents differed considerably with respect to homesickness with the UK studentsreporting more intense homesickness Univariate differences were found for three ofthe ve subscales (missing family missing friends adjustment difculties) and the totalhomesickness score Adjusted means for the subscales and for total homesickness forthe two samples are presented in Table 2

Furthermore when asked in the separate question about general feelings of home-sickness the UK students more frequently endorsed having felt homesick (see Table 3)

160 Margaret Stroebe et al

Figure 4 Model of UK sample

These analyses indicate not only that the UK students are more often homesick butthat they are so to a greater extent

Antecedents and consequences of homesicknessThe global models (Figs 2 and 4) demonstrate that homesickness is related to depres-sion in both cultures and that low self-esteem or emotional instability are linked tohigher levels of self-reported homesickness and depression More possibly howeverthe longer the stay in the new situation the less homesick and consequently lessdepressed one tends to be However the evidence also suggests that there may becultural differences with respect to gender differences in homesickness

The more detailed models (Figs 3 and 5) have the same basic structure althoughthere are some differences In the UK females are more likely to experience adjustmentdifculties than are males (b = 15) and the longer a UK student has experienced thenew situation the less he or she is inclined to miss the family (b = plusmn 12) effects whichare non-signicant in the NLstudy To enhance comparisons between the cultures thegeographical distance to the parental home (data not available in the NL sample) wasnot included here in the UKmodel However if included the pattern remained the sameand revealed that distance from home was only reliably associated with missing thefamily (b = 19)

Attachment style and family backgroundWe conducted further analyses to explore the association between attachment style andfamily background variables with homesickness In the NLstudy the effects of insecureattachment on relocation the homesickness features depression and emotional stability

Homesickness among students in two cultures 161

Figure 5 Detailed model of UK sample

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 7: res_6

1984 1985 1986 Fisher amp Hood 1987 1988) Items derived from a pilot study in which 100respondents were asked to give a denition of homesickness supplemented the original list

This preliminary questionnaire was given to 300 social science students who were asked torate the extent to which 51 items were associated with homesickness This resulted in a list of 45items This version of the questionnaire was then given to 117 third year psychology studentsThese students were asked to rate the extent to which theyhad experienced the 45 homesickness-related aspects in the past four weeks This led to a third version of the questionnaire which wasused in the Utrecht Homesickness Project The 45 items assess the extent to which studentsexperienced these various aspects of homesickness in the past 4 weeks (using the answercategories lsquonotrsquo lsquoweakrsquo lsquomoderatersquo lsquostrongrsquo lsquovery strongrsquo)

Aprincipal component analysis (minimum eigenvalue gt 1) with a varimax rotation over the 45items was computed for students who had felt homesick in the past four weeks (N = 151)resulting in nine factors On the basis of a scree plot we used the rst ve factors which wereclearly interpretable and explained 57of the variance These ve factors correspond closely tothe features that according to most authors constitute homesickness missing family missingfriends having adjustment difculties ruminations about home and feeling lonely

Based on these ve factors a reduced set of items (20) was selected to ensure that thedeterminant of the correlation matrix would be gt 0 This is necessary to avoid co-linearity of setsof items Asecond principal component analysis with a varimax rotation over these 20 items andthe restriction that ve factors should be extracted revealed that the items loaded as expected onthe relevant ve factors This last procedure was repeated with all the students (N = 439) andrevealed the same factor loading pattern Within this set of 20 variables the ve factors explained73 of the variance On the basis of the ve-factor structure scales were computed resultingin very acceptable Cronbachrsquos alphas (see Table 1) This set of subscales will be referred to as theUtrecht Homesickness Scale (UHS) Note that if there were missing values lsquolist-wise deletionrsquo wasused

Additional measuresStudents were also asked how often they had experienced homesickness in the past 4 weeks(never rarely sometimes often very often) Adiscriminant analysis whereby the ve UHS scalesare used to predict whether or not students endorsed feeling homesick (never versus rarelysometimes often very often) in the past 4 weeks revealed that 83 were correctly classiedFurthermore there was a strong correlation between the total homesickness score and thefrequency of homesickness in the past 4 weeks (r = 71) which indicates that the UHS scale isclosely related to a direct measure of homesickness even though the scale does not mentionhomesickness specically The Dutch version of the Symptoms Check List (SCL-90 Arrindell ampEttema 1975) was used as an indicator of mental health status of the NL sample Only thedepressive mood subscale (a = 92) and the total health score (a = 97) were used for this part ofthe investigation

Personality factors were measured by means of a Dutch translation of the Golberg Bi-polar Big5 (De Jong van Eck ampvan den Bos 1994 eg Goldberg 1992) Only emotional stability (a = 85)was used for this investigation

A Dutch translation of the Attachment Style Measure (Hazan amp Shaver 1987) was used toassess insecure (that is avoidant and anxiousambivalent see Hazan ampShaver 1987) versus secureattachment This particular scale was chosen because of its conciseness and applicability for usewith these student samples

Statistical analysesComparable to what was done in the pre-test described previously we rst conducted factoranalysis on the homesickness scale which identied the ve subscales A principal componentanalysis (minimum eigenvalue gt 1) with a varimax rotation over the 45 items resulted in ninefactors On the basis of a scree plot we used the rst ve factors which were clearly interpretable

Homesickness among students in two cultures 153

and explained 57of the variance These ve factors correspond closely to the factors identiedin the pre-test on the UHS namely missing family missing friends having adjustment difcultiesruminations about home and feeling lonely

Then we carried out a multivariate analysis of variance (MANOVA) to test differences inhomesickness scores between the UK and NL samples We then computed path analyses (usingLISREL81) in order to examine multiple relationships simultaneously We hypothesized identical

154 Margaret Stroebe et al

Table 1 Utrecht Homesickness Scale factor structure and reliabilities (NL sample)

Factor

(1) (2) (3) (4) (5)

Percentage of explained varianceItem

62 15 9 8 6

Factor loadings

(1) Missing family (a = 90)Missing your parents 89 12 20 06 16Missing your family 87 13 19 09 14Missing home 74 12 21 18 37Feeling missed by your family 74 15 19 15 04

(2) Loneliness (a = 85)Feeling lonely 21 78 22 25 09Feeling unloved 12 75 22 15 07Feeling isolated from the rest of the world 09 72 14 33 22Feeling uprooted 08 64 21 35 22

(3) Missing friends (a = 87)Longing for acquaintances 16 22 83 19 16Searching for familiar faces 19 18 73 23 12Missing people whom you trust and can talk with 25 31 71 07 20Missing your friends 36 14 70 15 14

(4) Adjustment dif culties (a = 88)Finding it dif cult adjusting to a new situation 23 27 15 77 16Feeling uncomfortable in a new situation 15 44 06 71 23Feeling lost in the new situation 17 46 22 69 21Having dif culties in getting used to new customs 03 14 38 68 26

(5) Ruminations about home (a = 80)Having thoughts that an old situation was better thanhere and now 16 15 18 28 81Regretting the decision to leave an old situation 10 21 08 31 79Continuously having thoughts about home 41 08 24 13 68Repeatedly thinking of the past 18 45 28 plusmn 01 52

Total homesickness (a = 94) (20 items)

Note The questions were formulated in the following manner Could you please indicate to what extentyou have experienced the following in the past 4 weeks Missing your parents notndashweakndashmoderatendashstrongndashvery strong N = 439

models for both cultures assuming that the patterns of associations would be similar Weexamined models of the relationships between homesickness and its putative antecedents andconsequences in terms of both overall scores and subscale scores Precise details of these modelsare given in the results section

Several t indices as described by Jaccard and Wan (1996) are appropriate for our investigationof path models To demonstrate the overall degree of correspondence between predicted andobserved covariance matrices the x2 statistic with p gt 05 indicates that a perfect t model doesexist in the population The goodness-of-t index (GFI gt 90) indicates goodness of t Thestandardized root mean square residual (SRMR lt 10) indicates a deviation of less than 10correlation units on average between predicted and observed correlations The root mean squareerror of approximation (RMSEA lt 08) implies an adequate model and lt 05 implies a good modelwith the p value for close t (CFit) being non-signicant (gt 05) CFI gt 90 implies a good t

Study 2 The United Kingdom (UK)

Respondents and procedureDuring the autumn term of 1996 new-intake students at Cardiff University were askedto participate in a student survey Questionnaires were administered in October and Novemberand 280 completed questionnaires were returned representing approximately a 60 responserate Seventy-two per cent of the respondents were female The mean age was 188 years(SD = 21) All respondents had moved from the parental home to lodgings on campus TheCardiff students are referred to here as the UK sample

MeasuresAn English translation of the Utrecht Homesickness Scale (UHS) was used to measure the extent ofhomesickness The reliabilities of the scales were comparable to the NLsample missing the family(a = 85) loneliness (a = 84) missing friends (a = 78) adjustment difculties (a = 84)ruminations about home (a = 86) and the total score for homesickness (a = 93) Againrespondents were also asked to indicate how often they had felt homesick during the past 4weeks The Beck Depression Inventory (BDI Beck Ward Mendelson Mock amp Erbaugh 1961)was used to give a reliable measure of current level of depression (a = 82) The Self-likingSelf-competence Scale was used to assess self-esteem (a = 91) (cf Tafarodi amp Swann 1995) Anadapted version of the Conict Tactics Scale (cf Straus Hamby Boney-McCoy ampSugarman 1996)was used to ascertain type of family background of the UK respondents The Conict Tactics Scaleconsists of six subscales with similar acceptable alpha levels (on average 72) which assess theextent to which parents are perceived as handling conicts with the respondents in certain waysThe items used here reected three factors that is verbal reasoning (three items) verbalaggression (six items) and physical aggression (three items) of either father or mother Selectionof these items from the original much longer scale was thought to be less intrusive and easier forrespondents to answer while still providing a measure of the three factors The students were alsoasked to state the name of their home town On the basis of this information distances could becalculated between their home and their university lodgings

Statistical analysesThe analyses described above for Study 1 apply here too

ResultsThe results section is structured in three parts In the rst section results on the NLsample are reported In the second section the same is done for the UK sample In thelast section a comparison is made between the two samples

Homesickness among students in two cultures 155

Study 1 NL

HomesicknessAlmost 50of the students reported having felt homesick at least some of the time afterhaving started their new education (N = 230) There were no differences betweenmales and females with respect to homesickness intensity The intensity of home-sickness as measured by the UHS subscales (range 1 (not experienced) to 5 (verystrongly)) revealed that on average the highest scores were obtained for missing friends(M = 238 SD = 102) and missing the family (M = 215 SD = 095)2 The otherdimensions scored were adjustment difculties (M = 180 SD = 088) ruminationsabout home (M = 152 SD = 071) and loneliness (M = 179 SD = 087) Theseaveraged to a total mean score of 193 (SD = 071)

Antecedents and consequences of homesicknessThe NL data set offered the possibility of comparisons between students still living athome and those who had relocated3 This provided the opportunity to test whetherrelocation (R) would indeed be associated with depression (D) which in and of itselfwe assumed would be associated with homesickness (H) We must be cautious giventhe cross-sectional design of our studies in assuming mediation between the RndashDndashHvariables Based on Baron and Kenny (1986) a requirement for mediation with this dataset would be that relocation and depression are signicantly associated in other wordsthere must be an effect to be mediated In our data we nd that relocation anddepression are not signicantly associated and therefore a mediation analysis is notappropriate Nevertheless a path analysis is possible if we regress depression on re-location and homesickness and regress homesickness on relocation (see Fig 1) Ourresults show a signicant path from relocation to homesickness (b = 21) a signicantpath from homesickness to depression controlling for relocation (b = 62) and a non-signicant direct path from relocation to depression (b = plusmn 05) Showing that thetwo component paths of the indirect effect are each signicant is not the same asshowing that the indirect effect is signicant Testing for an indirect path (Sobel 1982)revealed that the indirect effect of relocation on depression is signicant (b = 13) Itseems reasonable to conclude then that the data are consistent with an indirect effectof relocation on depression by means of the intervening variable homesickness Whilethere are good reasons to argue that there is an indirect effect of relocation ondepression through homesickness it is evident that this effect though signicant isquite small and not strong enough to produce an overall effect of R on D

The model shown in Fig 1 was also tested using the SCL-90 total score as the healthmeasure instead of using the depression subscale This showed an almost identicalsolution The analysis provides further evidence that the assumed pattern is justiedand that it is not just due to some correspondence between depression and home-sickness The pattern suggests that relocation may be an antecedent for homesicknessbut not for depressive moods or general ill-health and that homesickness may bean important antecedent for ill-health Having found evidence for this relationshipstructural equation models could be tested with respect to the effects of emotionalstability gender and duration of stay on homesickness and depression (see Fig 2) The

156 Margaret Stroebe et al

2 Controlling for age and duration of stay in order to make comparisons with the UK sample possible3 Some students who had not moved also reported being homesick (15) in the past 4 weeks The vast majority of them(90) reported that they experienced homesickness only lsquorarelyrsquo or lsquosometimesrsquo

literature suggests that emotional instability is a good predictor for both homesickness(Eurelings-Bontekoe et al 1996) and depression (cf Watson ampPennebaker 1989) Thusone could assume that emotional stability (the opposite of emotional instability) wouldexplain the variance of depression However we nd that even though reducedhomesickness retains an effect on depression This suggests that irrespective of thelevel of emotional stability homesickness is a precursor for depression Furthermorethe emotionally unstable as expected have higher levels of homesickness and depres-sion What we also nd is that the longer one is away from home the less intense thehomesickness reaction which suggests that adaptation to the new situation takes place

In order to create a more comprehensive picture a model was formulated whichincluded the subscales of the UHS instead of the total homesickness score Theassumptions were that (1) adjustment difculties would be associated with missingfamily and friends ruminations about home and loneliness (2) missing family andfriends in turn would be associated with ruminations about home and loneliness (3)ruminations would also be associated to loneliness and depression and (4) lonelinessin turn to depression Results with respect to this model revealed that in principle theassumptions were correct if two modications were incorporated These wereestablishing a link between missing family and friends and removing the path frommissing family to loneliness The choice was made to let the errors of missing friendsand family correlate because no prior hypotheses were made as to a causal relation-ship between both features and to delete the non-signicant path4 Figure 3 indicateshow the homesickness subscales are related and how these are linked to the measuredantecedents and consequences To enhance the visual impact of the model the thick-ness of paths corresponds to the size of the path coefcients

Figure 3 suggests that to become depressed one has to experience adjustment

Homesickness among students in two cultures 157

Figure 1 Model of relocation homesickness and depression for NL students

4 For the sake of comparison with the forthcoming UK model the non-signicant paths between missing friends and lonelinessgender and adjustment dif culties and duration of stay and missing family were not removed from the model

difculties and consequently loneliness and to ruminate about home If the relocatedstudents experience severe adjustment difculties these do seem to go hand in handwith missing family and friends ruminations about home and loneliness Specicallyloneliness is most strongly associated with depression Furthermore the analysessuggest that the best predictor for adjustment difculties in this study is emotionalstability (b = plusmn 41) Emotional stability also has a strong relationship to depression(b = plusmn 45) and to a lesser extent to missing friends (b = plusmn 20) and loneliness(b = plusmn 15) The slight association of duration of stay with total homesickness dis-appeared when its relationship was tested on the ve separate homesickness subscales

Study 2 UK

HomesicknessMore than 80of the students reported having felt homesick at least some of the timeafter having started their new education (N = 280) The intensity of homesicknessas measured by the UHS subscales (likewise controlling for age and the number of daysaway from home) revealed that on average the highest scores in the UK were obtainedfor missing family (M = 293 SD = 097) and missing friends (M = 272 SD = 094)The other dimensions scored were adjustment difculties (M = 205 SD = 084)ruminations about home (M = 166 SD = 086) and loneliness (M = 195SD = 092) These averaged to a total mean score of 226 (SD = 073)

Antecedents and consequences of homesicknessAlthough the hypothesis could not be tested with the UKsample it seems reasonable toassume based on the NLdata that relocation would be associated with homesicknesswhich in turn would be associated with depression and not vice versa (of course similarcautions about mediating variables to those expressed for Study 1 are in order here too)

158 Margaret Stroebe et al

Figure 2 Model of NL sample

Accepting this premise and despite the differences in the depression and personalityscales used in the UK and the NL what one would expect is comparable patterns ofcorrelations since these should be the same irrespective of the particular instrumentused We formulated a structural equation model analogous to the one used in the NLstudy examining the effects of self-esteem gender and duration of stay based onhomesickness and depression for the UK students (see Fig 4) This analysis revealedthat analogous to the NL data (Fig 2) homesickness is associated with depressionand that low self-esteem can be considered a risk factor for becoming homesick anddepressed To see whether or not homesickness is the precursor for self-esteem insteadof self-esteem being the precursor as hypothesized the alternative model was testedResults did not show a signicant effect of homesickness on self-esteem Thus low self-esteem like low emotional stability can rather be considered a vulnerability factorfor homesickness Again we see that after controlling for self-esteem the relationshipbetween homesickness and depression remains signicant This suggests that home-sickness is associated with depression irrespective of the level of self-esteem while ifone has low self-esteem the risk of becoming homesick and depressed is greater TheUK results also reveal the slight negative effect of duration of stay the longer studentslive in a new situation the lower the level of homesickness In contrast to the NLdata the UK data do reveal a gender difference Female students (M = 241 SD = 074)report higher levels of homesickness (F(278 1) = 1585 p lt 001) than male students(M = 204 SD = 060)

Analogous to the NL study we formulated a more detailed model including theUHS subscales (Fig 5) This revealed a similar pattern of correlations although thereare some obvious differences Adjustment difculties are assumed to be the crux of

Homesickness among students in two cultures 159

Figure 3 Detailed model of NL sample

the model and the data conrm this From adjustment difculties there are two pathsthat are connected indirectly to depression one via loneliness and the second viaruminations about home Self-esteem (a personality trait which was assumed to havesome commonality with emotional stability) has a similar relationship to adjustmentdifculties and depression in the UK sample as emotional stability in the NLsample Asdemonstrated previously with respect to the total homesickness score when analysedat the subscale level females in the UK study report having more adjustment difcultiesthan males This would indicate that the difference in homesickness mainly lies infemales reporting more adjustment difculties

Study 1 and Study 2 comparisonThere were a number of signicant differences between the two national samplesDutch students were older (M = 192 years) than British students (M = 188 yearst (478) = 295 p lt 01) and Dutch students (M = 1988 days) had spent a longer numberof days away from home than British students (M = 501 days t (224) = 1098 p lt 001)Furthermore there were proportionately more females in the NLsample than the UKsample (82versus 72) The implications of these differences are discussed later

HomesicknessAfter controlling for age and duration of relocation it was still found that the NLand UKstudents differed considerably with respect to homesickness with the UK studentsreporting more intense homesickness Univariate differences were found for three ofthe ve subscales (missing family missing friends adjustment difculties) and the totalhomesickness score Adjusted means for the subscales and for total homesickness forthe two samples are presented in Table 2

Furthermore when asked in the separate question about general feelings of home-sickness the UK students more frequently endorsed having felt homesick (see Table 3)

160 Margaret Stroebe et al

Figure 4 Model of UK sample

These analyses indicate not only that the UK students are more often homesick butthat they are so to a greater extent

Antecedents and consequences of homesicknessThe global models (Figs 2 and 4) demonstrate that homesickness is related to depres-sion in both cultures and that low self-esteem or emotional instability are linked tohigher levels of self-reported homesickness and depression More possibly howeverthe longer the stay in the new situation the less homesick and consequently lessdepressed one tends to be However the evidence also suggests that there may becultural differences with respect to gender differences in homesickness

The more detailed models (Figs 3 and 5) have the same basic structure althoughthere are some differences In the UK females are more likely to experience adjustmentdifculties than are males (b = 15) and the longer a UK student has experienced thenew situation the less he or she is inclined to miss the family (b = plusmn 12) effects whichare non-signicant in the NLstudy To enhance comparisons between the cultures thegeographical distance to the parental home (data not available in the NL sample) wasnot included here in the UKmodel However if included the pattern remained the sameand revealed that distance from home was only reliably associated with missing thefamily (b = 19)

Attachment style and family backgroundWe conducted further analyses to explore the association between attachment style andfamily background variables with homesickness In the NLstudy the effects of insecureattachment on relocation the homesickness features depression and emotional stability

Homesickness among students in two cultures 161

Figure 5 Detailed model of UK sample

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 8: res_6

and explained 57of the variance These ve factors correspond closely to the factors identiedin the pre-test on the UHS namely missing family missing friends having adjustment difcultiesruminations about home and feeling lonely

Then we carried out a multivariate analysis of variance (MANOVA) to test differences inhomesickness scores between the UK and NL samples We then computed path analyses (usingLISREL81) in order to examine multiple relationships simultaneously We hypothesized identical

154 Margaret Stroebe et al

Table 1 Utrecht Homesickness Scale factor structure and reliabilities (NL sample)

Factor

(1) (2) (3) (4) (5)

Percentage of explained varianceItem

62 15 9 8 6

Factor loadings

(1) Missing family (a = 90)Missing your parents 89 12 20 06 16Missing your family 87 13 19 09 14Missing home 74 12 21 18 37Feeling missed by your family 74 15 19 15 04

(2) Loneliness (a = 85)Feeling lonely 21 78 22 25 09Feeling unloved 12 75 22 15 07Feeling isolated from the rest of the world 09 72 14 33 22Feeling uprooted 08 64 21 35 22

(3) Missing friends (a = 87)Longing for acquaintances 16 22 83 19 16Searching for familiar faces 19 18 73 23 12Missing people whom you trust and can talk with 25 31 71 07 20Missing your friends 36 14 70 15 14

(4) Adjustment dif culties (a = 88)Finding it dif cult adjusting to a new situation 23 27 15 77 16Feeling uncomfortable in a new situation 15 44 06 71 23Feeling lost in the new situation 17 46 22 69 21Having dif culties in getting used to new customs 03 14 38 68 26

(5) Ruminations about home (a = 80)Having thoughts that an old situation was better thanhere and now 16 15 18 28 81Regretting the decision to leave an old situation 10 21 08 31 79Continuously having thoughts about home 41 08 24 13 68Repeatedly thinking of the past 18 45 28 plusmn 01 52

Total homesickness (a = 94) (20 items)

Note The questions were formulated in the following manner Could you please indicate to what extentyou have experienced the following in the past 4 weeks Missing your parents notndashweakndashmoderatendashstrongndashvery strong N = 439

models for both cultures assuming that the patterns of associations would be similar Weexamined models of the relationships between homesickness and its putative antecedents andconsequences in terms of both overall scores and subscale scores Precise details of these modelsare given in the results section

Several t indices as described by Jaccard and Wan (1996) are appropriate for our investigationof path models To demonstrate the overall degree of correspondence between predicted andobserved covariance matrices the x2 statistic with p gt 05 indicates that a perfect t model doesexist in the population The goodness-of-t index (GFI gt 90) indicates goodness of t Thestandardized root mean square residual (SRMR lt 10) indicates a deviation of less than 10correlation units on average between predicted and observed correlations The root mean squareerror of approximation (RMSEA lt 08) implies an adequate model and lt 05 implies a good modelwith the p value for close t (CFit) being non-signicant (gt 05) CFI gt 90 implies a good t

Study 2 The United Kingdom (UK)

Respondents and procedureDuring the autumn term of 1996 new-intake students at Cardiff University were askedto participate in a student survey Questionnaires were administered in October and Novemberand 280 completed questionnaires were returned representing approximately a 60 responserate Seventy-two per cent of the respondents were female The mean age was 188 years(SD = 21) All respondents had moved from the parental home to lodgings on campus TheCardiff students are referred to here as the UK sample

MeasuresAn English translation of the Utrecht Homesickness Scale (UHS) was used to measure the extent ofhomesickness The reliabilities of the scales were comparable to the NLsample missing the family(a = 85) loneliness (a = 84) missing friends (a = 78) adjustment difculties (a = 84)ruminations about home (a = 86) and the total score for homesickness (a = 93) Againrespondents were also asked to indicate how often they had felt homesick during the past 4weeks The Beck Depression Inventory (BDI Beck Ward Mendelson Mock amp Erbaugh 1961)was used to give a reliable measure of current level of depression (a = 82) The Self-likingSelf-competence Scale was used to assess self-esteem (a = 91) (cf Tafarodi amp Swann 1995) Anadapted version of the Conict Tactics Scale (cf Straus Hamby Boney-McCoy ampSugarman 1996)was used to ascertain type of family background of the UK respondents The Conict Tactics Scaleconsists of six subscales with similar acceptable alpha levels (on average 72) which assess theextent to which parents are perceived as handling conicts with the respondents in certain waysThe items used here reected three factors that is verbal reasoning (three items) verbalaggression (six items) and physical aggression (three items) of either father or mother Selectionof these items from the original much longer scale was thought to be less intrusive and easier forrespondents to answer while still providing a measure of the three factors The students were alsoasked to state the name of their home town On the basis of this information distances could becalculated between their home and their university lodgings

Statistical analysesThe analyses described above for Study 1 apply here too

ResultsThe results section is structured in three parts In the rst section results on the NLsample are reported In the second section the same is done for the UK sample In thelast section a comparison is made between the two samples

Homesickness among students in two cultures 155

Study 1 NL

HomesicknessAlmost 50of the students reported having felt homesick at least some of the time afterhaving started their new education (N = 230) There were no differences betweenmales and females with respect to homesickness intensity The intensity of home-sickness as measured by the UHS subscales (range 1 (not experienced) to 5 (verystrongly)) revealed that on average the highest scores were obtained for missing friends(M = 238 SD = 102) and missing the family (M = 215 SD = 095)2 The otherdimensions scored were adjustment difculties (M = 180 SD = 088) ruminationsabout home (M = 152 SD = 071) and loneliness (M = 179 SD = 087) Theseaveraged to a total mean score of 193 (SD = 071)

Antecedents and consequences of homesicknessThe NL data set offered the possibility of comparisons between students still living athome and those who had relocated3 This provided the opportunity to test whetherrelocation (R) would indeed be associated with depression (D) which in and of itselfwe assumed would be associated with homesickness (H) We must be cautious giventhe cross-sectional design of our studies in assuming mediation between the RndashDndashHvariables Based on Baron and Kenny (1986) a requirement for mediation with this dataset would be that relocation and depression are signicantly associated in other wordsthere must be an effect to be mediated In our data we nd that relocation anddepression are not signicantly associated and therefore a mediation analysis is notappropriate Nevertheless a path analysis is possible if we regress depression on re-location and homesickness and regress homesickness on relocation (see Fig 1) Ourresults show a signicant path from relocation to homesickness (b = 21) a signicantpath from homesickness to depression controlling for relocation (b = 62) and a non-signicant direct path from relocation to depression (b = plusmn 05) Showing that thetwo component paths of the indirect effect are each signicant is not the same asshowing that the indirect effect is signicant Testing for an indirect path (Sobel 1982)revealed that the indirect effect of relocation on depression is signicant (b = 13) Itseems reasonable to conclude then that the data are consistent with an indirect effectof relocation on depression by means of the intervening variable homesickness Whilethere are good reasons to argue that there is an indirect effect of relocation ondepression through homesickness it is evident that this effect though signicant isquite small and not strong enough to produce an overall effect of R on D

The model shown in Fig 1 was also tested using the SCL-90 total score as the healthmeasure instead of using the depression subscale This showed an almost identicalsolution The analysis provides further evidence that the assumed pattern is justiedand that it is not just due to some correspondence between depression and home-sickness The pattern suggests that relocation may be an antecedent for homesicknessbut not for depressive moods or general ill-health and that homesickness may bean important antecedent for ill-health Having found evidence for this relationshipstructural equation models could be tested with respect to the effects of emotionalstability gender and duration of stay on homesickness and depression (see Fig 2) The

156 Margaret Stroebe et al

2 Controlling for age and duration of stay in order to make comparisons with the UK sample possible3 Some students who had not moved also reported being homesick (15) in the past 4 weeks The vast majority of them(90) reported that they experienced homesickness only lsquorarelyrsquo or lsquosometimesrsquo

literature suggests that emotional instability is a good predictor for both homesickness(Eurelings-Bontekoe et al 1996) and depression (cf Watson ampPennebaker 1989) Thusone could assume that emotional stability (the opposite of emotional instability) wouldexplain the variance of depression However we nd that even though reducedhomesickness retains an effect on depression This suggests that irrespective of thelevel of emotional stability homesickness is a precursor for depression Furthermorethe emotionally unstable as expected have higher levels of homesickness and depres-sion What we also nd is that the longer one is away from home the less intense thehomesickness reaction which suggests that adaptation to the new situation takes place

In order to create a more comprehensive picture a model was formulated whichincluded the subscales of the UHS instead of the total homesickness score Theassumptions were that (1) adjustment difculties would be associated with missingfamily and friends ruminations about home and loneliness (2) missing family andfriends in turn would be associated with ruminations about home and loneliness (3)ruminations would also be associated to loneliness and depression and (4) lonelinessin turn to depression Results with respect to this model revealed that in principle theassumptions were correct if two modications were incorporated These wereestablishing a link between missing family and friends and removing the path frommissing family to loneliness The choice was made to let the errors of missing friendsand family correlate because no prior hypotheses were made as to a causal relation-ship between both features and to delete the non-signicant path4 Figure 3 indicateshow the homesickness subscales are related and how these are linked to the measuredantecedents and consequences To enhance the visual impact of the model the thick-ness of paths corresponds to the size of the path coefcients

Figure 3 suggests that to become depressed one has to experience adjustment

Homesickness among students in two cultures 157

Figure 1 Model of relocation homesickness and depression for NL students

4 For the sake of comparison with the forthcoming UK model the non-signicant paths between missing friends and lonelinessgender and adjustment dif culties and duration of stay and missing family were not removed from the model

difculties and consequently loneliness and to ruminate about home If the relocatedstudents experience severe adjustment difculties these do seem to go hand in handwith missing family and friends ruminations about home and loneliness Specicallyloneliness is most strongly associated with depression Furthermore the analysessuggest that the best predictor for adjustment difculties in this study is emotionalstability (b = plusmn 41) Emotional stability also has a strong relationship to depression(b = plusmn 45) and to a lesser extent to missing friends (b = plusmn 20) and loneliness(b = plusmn 15) The slight association of duration of stay with total homesickness dis-appeared when its relationship was tested on the ve separate homesickness subscales

Study 2 UK

HomesicknessMore than 80of the students reported having felt homesick at least some of the timeafter having started their new education (N = 280) The intensity of homesicknessas measured by the UHS subscales (likewise controlling for age and the number of daysaway from home) revealed that on average the highest scores in the UK were obtainedfor missing family (M = 293 SD = 097) and missing friends (M = 272 SD = 094)The other dimensions scored were adjustment difculties (M = 205 SD = 084)ruminations about home (M = 166 SD = 086) and loneliness (M = 195SD = 092) These averaged to a total mean score of 226 (SD = 073)

Antecedents and consequences of homesicknessAlthough the hypothesis could not be tested with the UKsample it seems reasonable toassume based on the NLdata that relocation would be associated with homesicknesswhich in turn would be associated with depression and not vice versa (of course similarcautions about mediating variables to those expressed for Study 1 are in order here too)

158 Margaret Stroebe et al

Figure 2 Model of NL sample

Accepting this premise and despite the differences in the depression and personalityscales used in the UK and the NL what one would expect is comparable patterns ofcorrelations since these should be the same irrespective of the particular instrumentused We formulated a structural equation model analogous to the one used in the NLstudy examining the effects of self-esteem gender and duration of stay based onhomesickness and depression for the UK students (see Fig 4) This analysis revealedthat analogous to the NL data (Fig 2) homesickness is associated with depressionand that low self-esteem can be considered a risk factor for becoming homesick anddepressed To see whether or not homesickness is the precursor for self-esteem insteadof self-esteem being the precursor as hypothesized the alternative model was testedResults did not show a signicant effect of homesickness on self-esteem Thus low self-esteem like low emotional stability can rather be considered a vulnerability factorfor homesickness Again we see that after controlling for self-esteem the relationshipbetween homesickness and depression remains signicant This suggests that home-sickness is associated with depression irrespective of the level of self-esteem while ifone has low self-esteem the risk of becoming homesick and depressed is greater TheUK results also reveal the slight negative effect of duration of stay the longer studentslive in a new situation the lower the level of homesickness In contrast to the NLdata the UK data do reveal a gender difference Female students (M = 241 SD = 074)report higher levels of homesickness (F(278 1) = 1585 p lt 001) than male students(M = 204 SD = 060)

Analogous to the NL study we formulated a more detailed model including theUHS subscales (Fig 5) This revealed a similar pattern of correlations although thereare some obvious differences Adjustment difculties are assumed to be the crux of

Homesickness among students in two cultures 159

Figure 3 Detailed model of NL sample

the model and the data conrm this From adjustment difculties there are two pathsthat are connected indirectly to depression one via loneliness and the second viaruminations about home Self-esteem (a personality trait which was assumed to havesome commonality with emotional stability) has a similar relationship to adjustmentdifculties and depression in the UK sample as emotional stability in the NLsample Asdemonstrated previously with respect to the total homesickness score when analysedat the subscale level females in the UK study report having more adjustment difcultiesthan males This would indicate that the difference in homesickness mainly lies infemales reporting more adjustment difculties

Study 1 and Study 2 comparisonThere were a number of signicant differences between the two national samplesDutch students were older (M = 192 years) than British students (M = 188 yearst (478) = 295 p lt 01) and Dutch students (M = 1988 days) had spent a longer numberof days away from home than British students (M = 501 days t (224) = 1098 p lt 001)Furthermore there were proportionately more females in the NLsample than the UKsample (82versus 72) The implications of these differences are discussed later

HomesicknessAfter controlling for age and duration of relocation it was still found that the NLand UKstudents differed considerably with respect to homesickness with the UK studentsreporting more intense homesickness Univariate differences were found for three ofthe ve subscales (missing family missing friends adjustment difculties) and the totalhomesickness score Adjusted means for the subscales and for total homesickness forthe two samples are presented in Table 2

Furthermore when asked in the separate question about general feelings of home-sickness the UK students more frequently endorsed having felt homesick (see Table 3)

160 Margaret Stroebe et al

Figure 4 Model of UK sample

These analyses indicate not only that the UK students are more often homesick butthat they are so to a greater extent

Antecedents and consequences of homesicknessThe global models (Figs 2 and 4) demonstrate that homesickness is related to depres-sion in both cultures and that low self-esteem or emotional instability are linked tohigher levels of self-reported homesickness and depression More possibly howeverthe longer the stay in the new situation the less homesick and consequently lessdepressed one tends to be However the evidence also suggests that there may becultural differences with respect to gender differences in homesickness

The more detailed models (Figs 3 and 5) have the same basic structure althoughthere are some differences In the UK females are more likely to experience adjustmentdifculties than are males (b = 15) and the longer a UK student has experienced thenew situation the less he or she is inclined to miss the family (b = plusmn 12) effects whichare non-signicant in the NLstudy To enhance comparisons between the cultures thegeographical distance to the parental home (data not available in the NL sample) wasnot included here in the UKmodel However if included the pattern remained the sameand revealed that distance from home was only reliably associated with missing thefamily (b = 19)

Attachment style and family backgroundWe conducted further analyses to explore the association between attachment style andfamily background variables with homesickness In the NLstudy the effects of insecureattachment on relocation the homesickness features depression and emotional stability

Homesickness among students in two cultures 161

Figure 5 Detailed model of UK sample

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 9: res_6

models for both cultures assuming that the patterns of associations would be similar Weexamined models of the relationships between homesickness and its putative antecedents andconsequences in terms of both overall scores and subscale scores Precise details of these modelsare given in the results section

Several t indices as described by Jaccard and Wan (1996) are appropriate for our investigationof path models To demonstrate the overall degree of correspondence between predicted andobserved covariance matrices the x2 statistic with p gt 05 indicates that a perfect t model doesexist in the population The goodness-of-t index (GFI gt 90) indicates goodness of t Thestandardized root mean square residual (SRMR lt 10) indicates a deviation of less than 10correlation units on average between predicted and observed correlations The root mean squareerror of approximation (RMSEA lt 08) implies an adequate model and lt 05 implies a good modelwith the p value for close t (CFit) being non-signicant (gt 05) CFI gt 90 implies a good t

Study 2 The United Kingdom (UK)

Respondents and procedureDuring the autumn term of 1996 new-intake students at Cardiff University were askedto participate in a student survey Questionnaires were administered in October and Novemberand 280 completed questionnaires were returned representing approximately a 60 responserate Seventy-two per cent of the respondents were female The mean age was 188 years(SD = 21) All respondents had moved from the parental home to lodgings on campus TheCardiff students are referred to here as the UK sample

MeasuresAn English translation of the Utrecht Homesickness Scale (UHS) was used to measure the extent ofhomesickness The reliabilities of the scales were comparable to the NLsample missing the family(a = 85) loneliness (a = 84) missing friends (a = 78) adjustment difculties (a = 84)ruminations about home (a = 86) and the total score for homesickness (a = 93) Againrespondents were also asked to indicate how often they had felt homesick during the past 4weeks The Beck Depression Inventory (BDI Beck Ward Mendelson Mock amp Erbaugh 1961)was used to give a reliable measure of current level of depression (a = 82) The Self-likingSelf-competence Scale was used to assess self-esteem (a = 91) (cf Tafarodi amp Swann 1995) Anadapted version of the Conict Tactics Scale (cf Straus Hamby Boney-McCoy ampSugarman 1996)was used to ascertain type of family background of the UK respondents The Conict Tactics Scaleconsists of six subscales with similar acceptable alpha levels (on average 72) which assess theextent to which parents are perceived as handling conicts with the respondents in certain waysThe items used here reected three factors that is verbal reasoning (three items) verbalaggression (six items) and physical aggression (three items) of either father or mother Selectionof these items from the original much longer scale was thought to be less intrusive and easier forrespondents to answer while still providing a measure of the three factors The students were alsoasked to state the name of their home town On the basis of this information distances could becalculated between their home and their university lodgings

Statistical analysesThe analyses described above for Study 1 apply here too

ResultsThe results section is structured in three parts In the rst section results on the NLsample are reported In the second section the same is done for the UK sample In thelast section a comparison is made between the two samples

Homesickness among students in two cultures 155

Study 1 NL

HomesicknessAlmost 50of the students reported having felt homesick at least some of the time afterhaving started their new education (N = 230) There were no differences betweenmales and females with respect to homesickness intensity The intensity of home-sickness as measured by the UHS subscales (range 1 (not experienced) to 5 (verystrongly)) revealed that on average the highest scores were obtained for missing friends(M = 238 SD = 102) and missing the family (M = 215 SD = 095)2 The otherdimensions scored were adjustment difculties (M = 180 SD = 088) ruminationsabout home (M = 152 SD = 071) and loneliness (M = 179 SD = 087) Theseaveraged to a total mean score of 193 (SD = 071)

Antecedents and consequences of homesicknessThe NL data set offered the possibility of comparisons between students still living athome and those who had relocated3 This provided the opportunity to test whetherrelocation (R) would indeed be associated with depression (D) which in and of itselfwe assumed would be associated with homesickness (H) We must be cautious giventhe cross-sectional design of our studies in assuming mediation between the RndashDndashHvariables Based on Baron and Kenny (1986) a requirement for mediation with this dataset would be that relocation and depression are signicantly associated in other wordsthere must be an effect to be mediated In our data we nd that relocation anddepression are not signicantly associated and therefore a mediation analysis is notappropriate Nevertheless a path analysis is possible if we regress depression on re-location and homesickness and regress homesickness on relocation (see Fig 1) Ourresults show a signicant path from relocation to homesickness (b = 21) a signicantpath from homesickness to depression controlling for relocation (b = 62) and a non-signicant direct path from relocation to depression (b = plusmn 05) Showing that thetwo component paths of the indirect effect are each signicant is not the same asshowing that the indirect effect is signicant Testing for an indirect path (Sobel 1982)revealed that the indirect effect of relocation on depression is signicant (b = 13) Itseems reasonable to conclude then that the data are consistent with an indirect effectof relocation on depression by means of the intervening variable homesickness Whilethere are good reasons to argue that there is an indirect effect of relocation ondepression through homesickness it is evident that this effect though signicant isquite small and not strong enough to produce an overall effect of R on D

The model shown in Fig 1 was also tested using the SCL-90 total score as the healthmeasure instead of using the depression subscale This showed an almost identicalsolution The analysis provides further evidence that the assumed pattern is justiedand that it is not just due to some correspondence between depression and home-sickness The pattern suggests that relocation may be an antecedent for homesicknessbut not for depressive moods or general ill-health and that homesickness may bean important antecedent for ill-health Having found evidence for this relationshipstructural equation models could be tested with respect to the effects of emotionalstability gender and duration of stay on homesickness and depression (see Fig 2) The

156 Margaret Stroebe et al

2 Controlling for age and duration of stay in order to make comparisons with the UK sample possible3 Some students who had not moved also reported being homesick (15) in the past 4 weeks The vast majority of them(90) reported that they experienced homesickness only lsquorarelyrsquo or lsquosometimesrsquo

literature suggests that emotional instability is a good predictor for both homesickness(Eurelings-Bontekoe et al 1996) and depression (cf Watson ampPennebaker 1989) Thusone could assume that emotional stability (the opposite of emotional instability) wouldexplain the variance of depression However we nd that even though reducedhomesickness retains an effect on depression This suggests that irrespective of thelevel of emotional stability homesickness is a precursor for depression Furthermorethe emotionally unstable as expected have higher levels of homesickness and depres-sion What we also nd is that the longer one is away from home the less intense thehomesickness reaction which suggests that adaptation to the new situation takes place

In order to create a more comprehensive picture a model was formulated whichincluded the subscales of the UHS instead of the total homesickness score Theassumptions were that (1) adjustment difculties would be associated with missingfamily and friends ruminations about home and loneliness (2) missing family andfriends in turn would be associated with ruminations about home and loneliness (3)ruminations would also be associated to loneliness and depression and (4) lonelinessin turn to depression Results with respect to this model revealed that in principle theassumptions were correct if two modications were incorporated These wereestablishing a link between missing family and friends and removing the path frommissing family to loneliness The choice was made to let the errors of missing friendsand family correlate because no prior hypotheses were made as to a causal relation-ship between both features and to delete the non-signicant path4 Figure 3 indicateshow the homesickness subscales are related and how these are linked to the measuredantecedents and consequences To enhance the visual impact of the model the thick-ness of paths corresponds to the size of the path coefcients

Figure 3 suggests that to become depressed one has to experience adjustment

Homesickness among students in two cultures 157

Figure 1 Model of relocation homesickness and depression for NL students

4 For the sake of comparison with the forthcoming UK model the non-signicant paths between missing friends and lonelinessgender and adjustment dif culties and duration of stay and missing family were not removed from the model

difculties and consequently loneliness and to ruminate about home If the relocatedstudents experience severe adjustment difculties these do seem to go hand in handwith missing family and friends ruminations about home and loneliness Specicallyloneliness is most strongly associated with depression Furthermore the analysessuggest that the best predictor for adjustment difculties in this study is emotionalstability (b = plusmn 41) Emotional stability also has a strong relationship to depression(b = plusmn 45) and to a lesser extent to missing friends (b = plusmn 20) and loneliness(b = plusmn 15) The slight association of duration of stay with total homesickness dis-appeared when its relationship was tested on the ve separate homesickness subscales

Study 2 UK

HomesicknessMore than 80of the students reported having felt homesick at least some of the timeafter having started their new education (N = 280) The intensity of homesicknessas measured by the UHS subscales (likewise controlling for age and the number of daysaway from home) revealed that on average the highest scores in the UK were obtainedfor missing family (M = 293 SD = 097) and missing friends (M = 272 SD = 094)The other dimensions scored were adjustment difculties (M = 205 SD = 084)ruminations about home (M = 166 SD = 086) and loneliness (M = 195SD = 092) These averaged to a total mean score of 226 (SD = 073)

Antecedents and consequences of homesicknessAlthough the hypothesis could not be tested with the UKsample it seems reasonable toassume based on the NLdata that relocation would be associated with homesicknesswhich in turn would be associated with depression and not vice versa (of course similarcautions about mediating variables to those expressed for Study 1 are in order here too)

158 Margaret Stroebe et al

Figure 2 Model of NL sample

Accepting this premise and despite the differences in the depression and personalityscales used in the UK and the NL what one would expect is comparable patterns ofcorrelations since these should be the same irrespective of the particular instrumentused We formulated a structural equation model analogous to the one used in the NLstudy examining the effects of self-esteem gender and duration of stay based onhomesickness and depression for the UK students (see Fig 4) This analysis revealedthat analogous to the NL data (Fig 2) homesickness is associated with depressionand that low self-esteem can be considered a risk factor for becoming homesick anddepressed To see whether or not homesickness is the precursor for self-esteem insteadof self-esteem being the precursor as hypothesized the alternative model was testedResults did not show a signicant effect of homesickness on self-esteem Thus low self-esteem like low emotional stability can rather be considered a vulnerability factorfor homesickness Again we see that after controlling for self-esteem the relationshipbetween homesickness and depression remains signicant This suggests that home-sickness is associated with depression irrespective of the level of self-esteem while ifone has low self-esteem the risk of becoming homesick and depressed is greater TheUK results also reveal the slight negative effect of duration of stay the longer studentslive in a new situation the lower the level of homesickness In contrast to the NLdata the UK data do reveal a gender difference Female students (M = 241 SD = 074)report higher levels of homesickness (F(278 1) = 1585 p lt 001) than male students(M = 204 SD = 060)

Analogous to the NL study we formulated a more detailed model including theUHS subscales (Fig 5) This revealed a similar pattern of correlations although thereare some obvious differences Adjustment difculties are assumed to be the crux of

Homesickness among students in two cultures 159

Figure 3 Detailed model of NL sample

the model and the data conrm this From adjustment difculties there are two pathsthat are connected indirectly to depression one via loneliness and the second viaruminations about home Self-esteem (a personality trait which was assumed to havesome commonality with emotional stability) has a similar relationship to adjustmentdifculties and depression in the UK sample as emotional stability in the NLsample Asdemonstrated previously with respect to the total homesickness score when analysedat the subscale level females in the UK study report having more adjustment difcultiesthan males This would indicate that the difference in homesickness mainly lies infemales reporting more adjustment difculties

Study 1 and Study 2 comparisonThere were a number of signicant differences between the two national samplesDutch students were older (M = 192 years) than British students (M = 188 yearst (478) = 295 p lt 01) and Dutch students (M = 1988 days) had spent a longer numberof days away from home than British students (M = 501 days t (224) = 1098 p lt 001)Furthermore there were proportionately more females in the NLsample than the UKsample (82versus 72) The implications of these differences are discussed later

HomesicknessAfter controlling for age and duration of relocation it was still found that the NLand UKstudents differed considerably with respect to homesickness with the UK studentsreporting more intense homesickness Univariate differences were found for three ofthe ve subscales (missing family missing friends adjustment difculties) and the totalhomesickness score Adjusted means for the subscales and for total homesickness forthe two samples are presented in Table 2

Furthermore when asked in the separate question about general feelings of home-sickness the UK students more frequently endorsed having felt homesick (see Table 3)

160 Margaret Stroebe et al

Figure 4 Model of UK sample

These analyses indicate not only that the UK students are more often homesick butthat they are so to a greater extent

Antecedents and consequences of homesicknessThe global models (Figs 2 and 4) demonstrate that homesickness is related to depres-sion in both cultures and that low self-esteem or emotional instability are linked tohigher levels of self-reported homesickness and depression More possibly howeverthe longer the stay in the new situation the less homesick and consequently lessdepressed one tends to be However the evidence also suggests that there may becultural differences with respect to gender differences in homesickness

The more detailed models (Figs 3 and 5) have the same basic structure althoughthere are some differences In the UK females are more likely to experience adjustmentdifculties than are males (b = 15) and the longer a UK student has experienced thenew situation the less he or she is inclined to miss the family (b = plusmn 12) effects whichare non-signicant in the NLstudy To enhance comparisons between the cultures thegeographical distance to the parental home (data not available in the NL sample) wasnot included here in the UKmodel However if included the pattern remained the sameand revealed that distance from home was only reliably associated with missing thefamily (b = 19)

Attachment style and family backgroundWe conducted further analyses to explore the association between attachment style andfamily background variables with homesickness In the NLstudy the effects of insecureattachment on relocation the homesickness features depression and emotional stability

Homesickness among students in two cultures 161

Figure 5 Detailed model of UK sample

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 10: res_6

Study 1 NL

HomesicknessAlmost 50of the students reported having felt homesick at least some of the time afterhaving started their new education (N = 230) There were no differences betweenmales and females with respect to homesickness intensity The intensity of home-sickness as measured by the UHS subscales (range 1 (not experienced) to 5 (verystrongly)) revealed that on average the highest scores were obtained for missing friends(M = 238 SD = 102) and missing the family (M = 215 SD = 095)2 The otherdimensions scored were adjustment difculties (M = 180 SD = 088) ruminationsabout home (M = 152 SD = 071) and loneliness (M = 179 SD = 087) Theseaveraged to a total mean score of 193 (SD = 071)

Antecedents and consequences of homesicknessThe NL data set offered the possibility of comparisons between students still living athome and those who had relocated3 This provided the opportunity to test whetherrelocation (R) would indeed be associated with depression (D) which in and of itselfwe assumed would be associated with homesickness (H) We must be cautious giventhe cross-sectional design of our studies in assuming mediation between the RndashDndashHvariables Based on Baron and Kenny (1986) a requirement for mediation with this dataset would be that relocation and depression are signicantly associated in other wordsthere must be an effect to be mediated In our data we nd that relocation anddepression are not signicantly associated and therefore a mediation analysis is notappropriate Nevertheless a path analysis is possible if we regress depression on re-location and homesickness and regress homesickness on relocation (see Fig 1) Ourresults show a signicant path from relocation to homesickness (b = 21) a signicantpath from homesickness to depression controlling for relocation (b = 62) and a non-signicant direct path from relocation to depression (b = plusmn 05) Showing that thetwo component paths of the indirect effect are each signicant is not the same asshowing that the indirect effect is signicant Testing for an indirect path (Sobel 1982)revealed that the indirect effect of relocation on depression is signicant (b = 13) Itseems reasonable to conclude then that the data are consistent with an indirect effectof relocation on depression by means of the intervening variable homesickness Whilethere are good reasons to argue that there is an indirect effect of relocation ondepression through homesickness it is evident that this effect though signicant isquite small and not strong enough to produce an overall effect of R on D

The model shown in Fig 1 was also tested using the SCL-90 total score as the healthmeasure instead of using the depression subscale This showed an almost identicalsolution The analysis provides further evidence that the assumed pattern is justiedand that it is not just due to some correspondence between depression and home-sickness The pattern suggests that relocation may be an antecedent for homesicknessbut not for depressive moods or general ill-health and that homesickness may bean important antecedent for ill-health Having found evidence for this relationshipstructural equation models could be tested with respect to the effects of emotionalstability gender and duration of stay on homesickness and depression (see Fig 2) The

156 Margaret Stroebe et al

2 Controlling for age and duration of stay in order to make comparisons with the UK sample possible3 Some students who had not moved also reported being homesick (15) in the past 4 weeks The vast majority of them(90) reported that they experienced homesickness only lsquorarelyrsquo or lsquosometimesrsquo

literature suggests that emotional instability is a good predictor for both homesickness(Eurelings-Bontekoe et al 1996) and depression (cf Watson ampPennebaker 1989) Thusone could assume that emotional stability (the opposite of emotional instability) wouldexplain the variance of depression However we nd that even though reducedhomesickness retains an effect on depression This suggests that irrespective of thelevel of emotional stability homesickness is a precursor for depression Furthermorethe emotionally unstable as expected have higher levels of homesickness and depres-sion What we also nd is that the longer one is away from home the less intense thehomesickness reaction which suggests that adaptation to the new situation takes place

In order to create a more comprehensive picture a model was formulated whichincluded the subscales of the UHS instead of the total homesickness score Theassumptions were that (1) adjustment difculties would be associated with missingfamily and friends ruminations about home and loneliness (2) missing family andfriends in turn would be associated with ruminations about home and loneliness (3)ruminations would also be associated to loneliness and depression and (4) lonelinessin turn to depression Results with respect to this model revealed that in principle theassumptions were correct if two modications were incorporated These wereestablishing a link between missing family and friends and removing the path frommissing family to loneliness The choice was made to let the errors of missing friendsand family correlate because no prior hypotheses were made as to a causal relation-ship between both features and to delete the non-signicant path4 Figure 3 indicateshow the homesickness subscales are related and how these are linked to the measuredantecedents and consequences To enhance the visual impact of the model the thick-ness of paths corresponds to the size of the path coefcients

Figure 3 suggests that to become depressed one has to experience adjustment

Homesickness among students in two cultures 157

Figure 1 Model of relocation homesickness and depression for NL students

4 For the sake of comparison with the forthcoming UK model the non-signicant paths between missing friends and lonelinessgender and adjustment dif culties and duration of stay and missing family were not removed from the model

difculties and consequently loneliness and to ruminate about home If the relocatedstudents experience severe adjustment difculties these do seem to go hand in handwith missing family and friends ruminations about home and loneliness Specicallyloneliness is most strongly associated with depression Furthermore the analysessuggest that the best predictor for adjustment difculties in this study is emotionalstability (b = plusmn 41) Emotional stability also has a strong relationship to depression(b = plusmn 45) and to a lesser extent to missing friends (b = plusmn 20) and loneliness(b = plusmn 15) The slight association of duration of stay with total homesickness dis-appeared when its relationship was tested on the ve separate homesickness subscales

Study 2 UK

HomesicknessMore than 80of the students reported having felt homesick at least some of the timeafter having started their new education (N = 280) The intensity of homesicknessas measured by the UHS subscales (likewise controlling for age and the number of daysaway from home) revealed that on average the highest scores in the UK were obtainedfor missing family (M = 293 SD = 097) and missing friends (M = 272 SD = 094)The other dimensions scored were adjustment difculties (M = 205 SD = 084)ruminations about home (M = 166 SD = 086) and loneliness (M = 195SD = 092) These averaged to a total mean score of 226 (SD = 073)

Antecedents and consequences of homesicknessAlthough the hypothesis could not be tested with the UKsample it seems reasonable toassume based on the NLdata that relocation would be associated with homesicknesswhich in turn would be associated with depression and not vice versa (of course similarcautions about mediating variables to those expressed for Study 1 are in order here too)

158 Margaret Stroebe et al

Figure 2 Model of NL sample

Accepting this premise and despite the differences in the depression and personalityscales used in the UK and the NL what one would expect is comparable patterns ofcorrelations since these should be the same irrespective of the particular instrumentused We formulated a structural equation model analogous to the one used in the NLstudy examining the effects of self-esteem gender and duration of stay based onhomesickness and depression for the UK students (see Fig 4) This analysis revealedthat analogous to the NL data (Fig 2) homesickness is associated with depressionand that low self-esteem can be considered a risk factor for becoming homesick anddepressed To see whether or not homesickness is the precursor for self-esteem insteadof self-esteem being the precursor as hypothesized the alternative model was testedResults did not show a signicant effect of homesickness on self-esteem Thus low self-esteem like low emotional stability can rather be considered a vulnerability factorfor homesickness Again we see that after controlling for self-esteem the relationshipbetween homesickness and depression remains signicant This suggests that home-sickness is associated with depression irrespective of the level of self-esteem while ifone has low self-esteem the risk of becoming homesick and depressed is greater TheUK results also reveal the slight negative effect of duration of stay the longer studentslive in a new situation the lower the level of homesickness In contrast to the NLdata the UK data do reveal a gender difference Female students (M = 241 SD = 074)report higher levels of homesickness (F(278 1) = 1585 p lt 001) than male students(M = 204 SD = 060)

Analogous to the NL study we formulated a more detailed model including theUHS subscales (Fig 5) This revealed a similar pattern of correlations although thereare some obvious differences Adjustment difculties are assumed to be the crux of

Homesickness among students in two cultures 159

Figure 3 Detailed model of NL sample

the model and the data conrm this From adjustment difculties there are two pathsthat are connected indirectly to depression one via loneliness and the second viaruminations about home Self-esteem (a personality trait which was assumed to havesome commonality with emotional stability) has a similar relationship to adjustmentdifculties and depression in the UK sample as emotional stability in the NLsample Asdemonstrated previously with respect to the total homesickness score when analysedat the subscale level females in the UK study report having more adjustment difcultiesthan males This would indicate that the difference in homesickness mainly lies infemales reporting more adjustment difculties

Study 1 and Study 2 comparisonThere were a number of signicant differences between the two national samplesDutch students were older (M = 192 years) than British students (M = 188 yearst (478) = 295 p lt 01) and Dutch students (M = 1988 days) had spent a longer numberof days away from home than British students (M = 501 days t (224) = 1098 p lt 001)Furthermore there were proportionately more females in the NLsample than the UKsample (82versus 72) The implications of these differences are discussed later

HomesicknessAfter controlling for age and duration of relocation it was still found that the NLand UKstudents differed considerably with respect to homesickness with the UK studentsreporting more intense homesickness Univariate differences were found for three ofthe ve subscales (missing family missing friends adjustment difculties) and the totalhomesickness score Adjusted means for the subscales and for total homesickness forthe two samples are presented in Table 2

Furthermore when asked in the separate question about general feelings of home-sickness the UK students more frequently endorsed having felt homesick (see Table 3)

160 Margaret Stroebe et al

Figure 4 Model of UK sample

These analyses indicate not only that the UK students are more often homesick butthat they are so to a greater extent

Antecedents and consequences of homesicknessThe global models (Figs 2 and 4) demonstrate that homesickness is related to depres-sion in both cultures and that low self-esteem or emotional instability are linked tohigher levels of self-reported homesickness and depression More possibly howeverthe longer the stay in the new situation the less homesick and consequently lessdepressed one tends to be However the evidence also suggests that there may becultural differences with respect to gender differences in homesickness

The more detailed models (Figs 3 and 5) have the same basic structure althoughthere are some differences In the UK females are more likely to experience adjustmentdifculties than are males (b = 15) and the longer a UK student has experienced thenew situation the less he or she is inclined to miss the family (b = plusmn 12) effects whichare non-signicant in the NLstudy To enhance comparisons between the cultures thegeographical distance to the parental home (data not available in the NL sample) wasnot included here in the UKmodel However if included the pattern remained the sameand revealed that distance from home was only reliably associated with missing thefamily (b = 19)

Attachment style and family backgroundWe conducted further analyses to explore the association between attachment style andfamily background variables with homesickness In the NLstudy the effects of insecureattachment on relocation the homesickness features depression and emotional stability

Homesickness among students in two cultures 161

Figure 5 Detailed model of UK sample

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 11: res_6

literature suggests that emotional instability is a good predictor for both homesickness(Eurelings-Bontekoe et al 1996) and depression (cf Watson ampPennebaker 1989) Thusone could assume that emotional stability (the opposite of emotional instability) wouldexplain the variance of depression However we nd that even though reducedhomesickness retains an effect on depression This suggests that irrespective of thelevel of emotional stability homesickness is a precursor for depression Furthermorethe emotionally unstable as expected have higher levels of homesickness and depres-sion What we also nd is that the longer one is away from home the less intense thehomesickness reaction which suggests that adaptation to the new situation takes place

In order to create a more comprehensive picture a model was formulated whichincluded the subscales of the UHS instead of the total homesickness score Theassumptions were that (1) adjustment difculties would be associated with missingfamily and friends ruminations about home and loneliness (2) missing family andfriends in turn would be associated with ruminations about home and loneliness (3)ruminations would also be associated to loneliness and depression and (4) lonelinessin turn to depression Results with respect to this model revealed that in principle theassumptions were correct if two modications were incorporated These wereestablishing a link between missing family and friends and removing the path frommissing family to loneliness The choice was made to let the errors of missing friendsand family correlate because no prior hypotheses were made as to a causal relation-ship between both features and to delete the non-signicant path4 Figure 3 indicateshow the homesickness subscales are related and how these are linked to the measuredantecedents and consequences To enhance the visual impact of the model the thick-ness of paths corresponds to the size of the path coefcients

Figure 3 suggests that to become depressed one has to experience adjustment

Homesickness among students in two cultures 157

Figure 1 Model of relocation homesickness and depression for NL students

4 For the sake of comparison with the forthcoming UK model the non-signicant paths between missing friends and lonelinessgender and adjustment dif culties and duration of stay and missing family were not removed from the model

difculties and consequently loneliness and to ruminate about home If the relocatedstudents experience severe adjustment difculties these do seem to go hand in handwith missing family and friends ruminations about home and loneliness Specicallyloneliness is most strongly associated with depression Furthermore the analysessuggest that the best predictor for adjustment difculties in this study is emotionalstability (b = plusmn 41) Emotional stability also has a strong relationship to depression(b = plusmn 45) and to a lesser extent to missing friends (b = plusmn 20) and loneliness(b = plusmn 15) The slight association of duration of stay with total homesickness dis-appeared when its relationship was tested on the ve separate homesickness subscales

Study 2 UK

HomesicknessMore than 80of the students reported having felt homesick at least some of the timeafter having started their new education (N = 280) The intensity of homesicknessas measured by the UHS subscales (likewise controlling for age and the number of daysaway from home) revealed that on average the highest scores in the UK were obtainedfor missing family (M = 293 SD = 097) and missing friends (M = 272 SD = 094)The other dimensions scored were adjustment difculties (M = 205 SD = 084)ruminations about home (M = 166 SD = 086) and loneliness (M = 195SD = 092) These averaged to a total mean score of 226 (SD = 073)

Antecedents and consequences of homesicknessAlthough the hypothesis could not be tested with the UKsample it seems reasonable toassume based on the NLdata that relocation would be associated with homesicknesswhich in turn would be associated with depression and not vice versa (of course similarcautions about mediating variables to those expressed for Study 1 are in order here too)

158 Margaret Stroebe et al

Figure 2 Model of NL sample

Accepting this premise and despite the differences in the depression and personalityscales used in the UK and the NL what one would expect is comparable patterns ofcorrelations since these should be the same irrespective of the particular instrumentused We formulated a structural equation model analogous to the one used in the NLstudy examining the effects of self-esteem gender and duration of stay based onhomesickness and depression for the UK students (see Fig 4) This analysis revealedthat analogous to the NL data (Fig 2) homesickness is associated with depressionand that low self-esteem can be considered a risk factor for becoming homesick anddepressed To see whether or not homesickness is the precursor for self-esteem insteadof self-esteem being the precursor as hypothesized the alternative model was testedResults did not show a signicant effect of homesickness on self-esteem Thus low self-esteem like low emotional stability can rather be considered a vulnerability factorfor homesickness Again we see that after controlling for self-esteem the relationshipbetween homesickness and depression remains signicant This suggests that home-sickness is associated with depression irrespective of the level of self-esteem while ifone has low self-esteem the risk of becoming homesick and depressed is greater TheUK results also reveal the slight negative effect of duration of stay the longer studentslive in a new situation the lower the level of homesickness In contrast to the NLdata the UK data do reveal a gender difference Female students (M = 241 SD = 074)report higher levels of homesickness (F(278 1) = 1585 p lt 001) than male students(M = 204 SD = 060)

Analogous to the NL study we formulated a more detailed model including theUHS subscales (Fig 5) This revealed a similar pattern of correlations although thereare some obvious differences Adjustment difculties are assumed to be the crux of

Homesickness among students in two cultures 159

Figure 3 Detailed model of NL sample

the model and the data conrm this From adjustment difculties there are two pathsthat are connected indirectly to depression one via loneliness and the second viaruminations about home Self-esteem (a personality trait which was assumed to havesome commonality with emotional stability) has a similar relationship to adjustmentdifculties and depression in the UK sample as emotional stability in the NLsample Asdemonstrated previously with respect to the total homesickness score when analysedat the subscale level females in the UK study report having more adjustment difcultiesthan males This would indicate that the difference in homesickness mainly lies infemales reporting more adjustment difculties

Study 1 and Study 2 comparisonThere were a number of signicant differences between the two national samplesDutch students were older (M = 192 years) than British students (M = 188 yearst (478) = 295 p lt 01) and Dutch students (M = 1988 days) had spent a longer numberof days away from home than British students (M = 501 days t (224) = 1098 p lt 001)Furthermore there were proportionately more females in the NLsample than the UKsample (82versus 72) The implications of these differences are discussed later

HomesicknessAfter controlling for age and duration of relocation it was still found that the NLand UKstudents differed considerably with respect to homesickness with the UK studentsreporting more intense homesickness Univariate differences were found for three ofthe ve subscales (missing family missing friends adjustment difculties) and the totalhomesickness score Adjusted means for the subscales and for total homesickness forthe two samples are presented in Table 2

Furthermore when asked in the separate question about general feelings of home-sickness the UK students more frequently endorsed having felt homesick (see Table 3)

160 Margaret Stroebe et al

Figure 4 Model of UK sample

These analyses indicate not only that the UK students are more often homesick butthat they are so to a greater extent

Antecedents and consequences of homesicknessThe global models (Figs 2 and 4) demonstrate that homesickness is related to depres-sion in both cultures and that low self-esteem or emotional instability are linked tohigher levels of self-reported homesickness and depression More possibly howeverthe longer the stay in the new situation the less homesick and consequently lessdepressed one tends to be However the evidence also suggests that there may becultural differences with respect to gender differences in homesickness

The more detailed models (Figs 3 and 5) have the same basic structure althoughthere are some differences In the UK females are more likely to experience adjustmentdifculties than are males (b = 15) and the longer a UK student has experienced thenew situation the less he or she is inclined to miss the family (b = plusmn 12) effects whichare non-signicant in the NLstudy To enhance comparisons between the cultures thegeographical distance to the parental home (data not available in the NL sample) wasnot included here in the UKmodel However if included the pattern remained the sameand revealed that distance from home was only reliably associated with missing thefamily (b = 19)

Attachment style and family backgroundWe conducted further analyses to explore the association between attachment style andfamily background variables with homesickness In the NLstudy the effects of insecureattachment on relocation the homesickness features depression and emotional stability

Homesickness among students in two cultures 161

Figure 5 Detailed model of UK sample

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 12: res_6

difculties and consequently loneliness and to ruminate about home If the relocatedstudents experience severe adjustment difculties these do seem to go hand in handwith missing family and friends ruminations about home and loneliness Specicallyloneliness is most strongly associated with depression Furthermore the analysessuggest that the best predictor for adjustment difculties in this study is emotionalstability (b = plusmn 41) Emotional stability also has a strong relationship to depression(b = plusmn 45) and to a lesser extent to missing friends (b = plusmn 20) and loneliness(b = plusmn 15) The slight association of duration of stay with total homesickness dis-appeared when its relationship was tested on the ve separate homesickness subscales

Study 2 UK

HomesicknessMore than 80of the students reported having felt homesick at least some of the timeafter having started their new education (N = 280) The intensity of homesicknessas measured by the UHS subscales (likewise controlling for age and the number of daysaway from home) revealed that on average the highest scores in the UK were obtainedfor missing family (M = 293 SD = 097) and missing friends (M = 272 SD = 094)The other dimensions scored were adjustment difculties (M = 205 SD = 084)ruminations about home (M = 166 SD = 086) and loneliness (M = 195SD = 092) These averaged to a total mean score of 226 (SD = 073)

Antecedents and consequences of homesicknessAlthough the hypothesis could not be tested with the UKsample it seems reasonable toassume based on the NLdata that relocation would be associated with homesicknesswhich in turn would be associated with depression and not vice versa (of course similarcautions about mediating variables to those expressed for Study 1 are in order here too)

158 Margaret Stroebe et al

Figure 2 Model of NL sample

Accepting this premise and despite the differences in the depression and personalityscales used in the UK and the NL what one would expect is comparable patterns ofcorrelations since these should be the same irrespective of the particular instrumentused We formulated a structural equation model analogous to the one used in the NLstudy examining the effects of self-esteem gender and duration of stay based onhomesickness and depression for the UK students (see Fig 4) This analysis revealedthat analogous to the NL data (Fig 2) homesickness is associated with depressionand that low self-esteem can be considered a risk factor for becoming homesick anddepressed To see whether or not homesickness is the precursor for self-esteem insteadof self-esteem being the precursor as hypothesized the alternative model was testedResults did not show a signicant effect of homesickness on self-esteem Thus low self-esteem like low emotional stability can rather be considered a vulnerability factorfor homesickness Again we see that after controlling for self-esteem the relationshipbetween homesickness and depression remains signicant This suggests that home-sickness is associated with depression irrespective of the level of self-esteem while ifone has low self-esteem the risk of becoming homesick and depressed is greater TheUK results also reveal the slight negative effect of duration of stay the longer studentslive in a new situation the lower the level of homesickness In contrast to the NLdata the UK data do reveal a gender difference Female students (M = 241 SD = 074)report higher levels of homesickness (F(278 1) = 1585 p lt 001) than male students(M = 204 SD = 060)

Analogous to the NL study we formulated a more detailed model including theUHS subscales (Fig 5) This revealed a similar pattern of correlations although thereare some obvious differences Adjustment difculties are assumed to be the crux of

Homesickness among students in two cultures 159

Figure 3 Detailed model of NL sample

the model and the data conrm this From adjustment difculties there are two pathsthat are connected indirectly to depression one via loneliness and the second viaruminations about home Self-esteem (a personality trait which was assumed to havesome commonality with emotional stability) has a similar relationship to adjustmentdifculties and depression in the UK sample as emotional stability in the NLsample Asdemonstrated previously with respect to the total homesickness score when analysedat the subscale level females in the UK study report having more adjustment difcultiesthan males This would indicate that the difference in homesickness mainly lies infemales reporting more adjustment difculties

Study 1 and Study 2 comparisonThere were a number of signicant differences between the two national samplesDutch students were older (M = 192 years) than British students (M = 188 yearst (478) = 295 p lt 01) and Dutch students (M = 1988 days) had spent a longer numberof days away from home than British students (M = 501 days t (224) = 1098 p lt 001)Furthermore there were proportionately more females in the NLsample than the UKsample (82versus 72) The implications of these differences are discussed later

HomesicknessAfter controlling for age and duration of relocation it was still found that the NLand UKstudents differed considerably with respect to homesickness with the UK studentsreporting more intense homesickness Univariate differences were found for three ofthe ve subscales (missing family missing friends adjustment difculties) and the totalhomesickness score Adjusted means for the subscales and for total homesickness forthe two samples are presented in Table 2

Furthermore when asked in the separate question about general feelings of home-sickness the UK students more frequently endorsed having felt homesick (see Table 3)

160 Margaret Stroebe et al

Figure 4 Model of UK sample

These analyses indicate not only that the UK students are more often homesick butthat they are so to a greater extent

Antecedents and consequences of homesicknessThe global models (Figs 2 and 4) demonstrate that homesickness is related to depres-sion in both cultures and that low self-esteem or emotional instability are linked tohigher levels of self-reported homesickness and depression More possibly howeverthe longer the stay in the new situation the less homesick and consequently lessdepressed one tends to be However the evidence also suggests that there may becultural differences with respect to gender differences in homesickness

The more detailed models (Figs 3 and 5) have the same basic structure althoughthere are some differences In the UK females are more likely to experience adjustmentdifculties than are males (b = 15) and the longer a UK student has experienced thenew situation the less he or she is inclined to miss the family (b = plusmn 12) effects whichare non-signicant in the NLstudy To enhance comparisons between the cultures thegeographical distance to the parental home (data not available in the NL sample) wasnot included here in the UKmodel However if included the pattern remained the sameand revealed that distance from home was only reliably associated with missing thefamily (b = 19)

Attachment style and family backgroundWe conducted further analyses to explore the association between attachment style andfamily background variables with homesickness In the NLstudy the effects of insecureattachment on relocation the homesickness features depression and emotional stability

Homesickness among students in two cultures 161

Figure 5 Detailed model of UK sample

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 13: res_6

Accepting this premise and despite the differences in the depression and personalityscales used in the UK and the NL what one would expect is comparable patterns ofcorrelations since these should be the same irrespective of the particular instrumentused We formulated a structural equation model analogous to the one used in the NLstudy examining the effects of self-esteem gender and duration of stay based onhomesickness and depression for the UK students (see Fig 4) This analysis revealedthat analogous to the NL data (Fig 2) homesickness is associated with depressionand that low self-esteem can be considered a risk factor for becoming homesick anddepressed To see whether or not homesickness is the precursor for self-esteem insteadof self-esteem being the precursor as hypothesized the alternative model was testedResults did not show a signicant effect of homesickness on self-esteem Thus low self-esteem like low emotional stability can rather be considered a vulnerability factorfor homesickness Again we see that after controlling for self-esteem the relationshipbetween homesickness and depression remains signicant This suggests that home-sickness is associated with depression irrespective of the level of self-esteem while ifone has low self-esteem the risk of becoming homesick and depressed is greater TheUK results also reveal the slight negative effect of duration of stay the longer studentslive in a new situation the lower the level of homesickness In contrast to the NLdata the UK data do reveal a gender difference Female students (M = 241 SD = 074)report higher levels of homesickness (F(278 1) = 1585 p lt 001) than male students(M = 204 SD = 060)

Analogous to the NL study we formulated a more detailed model including theUHS subscales (Fig 5) This revealed a similar pattern of correlations although thereare some obvious differences Adjustment difculties are assumed to be the crux of

Homesickness among students in two cultures 159

Figure 3 Detailed model of NL sample

the model and the data conrm this From adjustment difculties there are two pathsthat are connected indirectly to depression one via loneliness and the second viaruminations about home Self-esteem (a personality trait which was assumed to havesome commonality with emotional stability) has a similar relationship to adjustmentdifculties and depression in the UK sample as emotional stability in the NLsample Asdemonstrated previously with respect to the total homesickness score when analysedat the subscale level females in the UK study report having more adjustment difcultiesthan males This would indicate that the difference in homesickness mainly lies infemales reporting more adjustment difculties

Study 1 and Study 2 comparisonThere were a number of signicant differences between the two national samplesDutch students were older (M = 192 years) than British students (M = 188 yearst (478) = 295 p lt 01) and Dutch students (M = 1988 days) had spent a longer numberof days away from home than British students (M = 501 days t (224) = 1098 p lt 001)Furthermore there were proportionately more females in the NLsample than the UKsample (82versus 72) The implications of these differences are discussed later

HomesicknessAfter controlling for age and duration of relocation it was still found that the NLand UKstudents differed considerably with respect to homesickness with the UK studentsreporting more intense homesickness Univariate differences were found for three ofthe ve subscales (missing family missing friends adjustment difculties) and the totalhomesickness score Adjusted means for the subscales and for total homesickness forthe two samples are presented in Table 2

Furthermore when asked in the separate question about general feelings of home-sickness the UK students more frequently endorsed having felt homesick (see Table 3)

160 Margaret Stroebe et al

Figure 4 Model of UK sample

These analyses indicate not only that the UK students are more often homesick butthat they are so to a greater extent

Antecedents and consequences of homesicknessThe global models (Figs 2 and 4) demonstrate that homesickness is related to depres-sion in both cultures and that low self-esteem or emotional instability are linked tohigher levels of self-reported homesickness and depression More possibly howeverthe longer the stay in the new situation the less homesick and consequently lessdepressed one tends to be However the evidence also suggests that there may becultural differences with respect to gender differences in homesickness

The more detailed models (Figs 3 and 5) have the same basic structure althoughthere are some differences In the UK females are more likely to experience adjustmentdifculties than are males (b = 15) and the longer a UK student has experienced thenew situation the less he or she is inclined to miss the family (b = plusmn 12) effects whichare non-signicant in the NLstudy To enhance comparisons between the cultures thegeographical distance to the parental home (data not available in the NL sample) wasnot included here in the UKmodel However if included the pattern remained the sameand revealed that distance from home was only reliably associated with missing thefamily (b = 19)

Attachment style and family backgroundWe conducted further analyses to explore the association between attachment style andfamily background variables with homesickness In the NLstudy the effects of insecureattachment on relocation the homesickness features depression and emotional stability

Homesickness among students in two cultures 161

Figure 5 Detailed model of UK sample

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 14: res_6

the model and the data conrm this From adjustment difculties there are two pathsthat are connected indirectly to depression one via loneliness and the second viaruminations about home Self-esteem (a personality trait which was assumed to havesome commonality with emotional stability) has a similar relationship to adjustmentdifculties and depression in the UK sample as emotional stability in the NLsample Asdemonstrated previously with respect to the total homesickness score when analysedat the subscale level females in the UK study report having more adjustment difcultiesthan males This would indicate that the difference in homesickness mainly lies infemales reporting more adjustment difculties

Study 1 and Study 2 comparisonThere were a number of signicant differences between the two national samplesDutch students were older (M = 192 years) than British students (M = 188 yearst (478) = 295 p lt 01) and Dutch students (M = 1988 days) had spent a longer numberof days away from home than British students (M = 501 days t (224) = 1098 p lt 001)Furthermore there were proportionately more females in the NLsample than the UKsample (82versus 72) The implications of these differences are discussed later

HomesicknessAfter controlling for age and duration of relocation it was still found that the NLand UKstudents differed considerably with respect to homesickness with the UK studentsreporting more intense homesickness Univariate differences were found for three ofthe ve subscales (missing family missing friends adjustment difculties) and the totalhomesickness score Adjusted means for the subscales and for total homesickness forthe two samples are presented in Table 2

Furthermore when asked in the separate question about general feelings of home-sickness the UK students more frequently endorsed having felt homesick (see Table 3)

160 Margaret Stroebe et al

Figure 4 Model of UK sample

These analyses indicate not only that the UK students are more often homesick butthat they are so to a greater extent

Antecedents and consequences of homesicknessThe global models (Figs 2 and 4) demonstrate that homesickness is related to depres-sion in both cultures and that low self-esteem or emotional instability are linked tohigher levels of self-reported homesickness and depression More possibly howeverthe longer the stay in the new situation the less homesick and consequently lessdepressed one tends to be However the evidence also suggests that there may becultural differences with respect to gender differences in homesickness

The more detailed models (Figs 3 and 5) have the same basic structure althoughthere are some differences In the UK females are more likely to experience adjustmentdifculties than are males (b = 15) and the longer a UK student has experienced thenew situation the less he or she is inclined to miss the family (b = plusmn 12) effects whichare non-signicant in the NLstudy To enhance comparisons between the cultures thegeographical distance to the parental home (data not available in the NL sample) wasnot included here in the UKmodel However if included the pattern remained the sameand revealed that distance from home was only reliably associated with missing thefamily (b = 19)

Attachment style and family backgroundWe conducted further analyses to explore the association between attachment style andfamily background variables with homesickness In the NLstudy the effects of insecureattachment on relocation the homesickness features depression and emotional stability

Homesickness among students in two cultures 161

Figure 5 Detailed model of UK sample

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 15: res_6

These analyses indicate not only that the UK students are more often homesick butthat they are so to a greater extent

Antecedents and consequences of homesicknessThe global models (Figs 2 and 4) demonstrate that homesickness is related to depres-sion in both cultures and that low self-esteem or emotional instability are linked tohigher levels of self-reported homesickness and depression More possibly howeverthe longer the stay in the new situation the less homesick and consequently lessdepressed one tends to be However the evidence also suggests that there may becultural differences with respect to gender differences in homesickness

The more detailed models (Figs 3 and 5) have the same basic structure althoughthere are some differences In the UK females are more likely to experience adjustmentdifculties than are males (b = 15) and the longer a UK student has experienced thenew situation the less he or she is inclined to miss the family (b = plusmn 12) effects whichare non-signicant in the NLstudy To enhance comparisons between the cultures thegeographical distance to the parental home (data not available in the NL sample) wasnot included here in the UKmodel However if included the pattern remained the sameand revealed that distance from home was only reliably associated with missing thefamily (b = 19)

Attachment style and family backgroundWe conducted further analyses to explore the association between attachment style andfamily background variables with homesickness In the NLstudy the effects of insecureattachment on relocation the homesickness features depression and emotional stability

Homesickness among students in two cultures 161

Figure 5 Detailed model of UK sample

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 16: res_6

were explored The analysis revealed that insecurely attached individuals had a higherprobability of being less emotionally stable (b = plusmn 38) more lonely (b = 16) (asmeasured on a subscale of the UHS) and more depressed (b = 11) It is worth notingthat only 5 of the sample fell within the insecure attachment category lsquoanxiousambivalentrsquomdashsince one might expect this category to be most dependent on home (andtherefore homesick)mdashwhile 24could be classied as lsquoavoidantrsquo

In the same fashion the associations between perceived parental conict tactics(reasoning verbal aggression and physical threat) and the homesickness featuresdepression and self-esteem were explored in the UK sample This analysis revealedthat the perceived verbal aggression of the mother was associated with a reducedlikelihood of missing the family (b = plusmn 17) and that the perceived physical threat of thefather was associated with an increased likelihood of having adjustment difculties asmeasured on the UHS (b = 23)

It is interesting to note that there is a stronger association between missing friendsand loneliness in the UK than in the NL samplemdashpossibly due to the fact that homeand friends are more accessible in NLdue to both geographical proximity and the freeavailability of public transport While there is a weaker association between lonelinessand depression in the UK than in the NLsample caution must be observed with respectto interpreting this comparison given the differences in the severity of depressionmeasured in the two samples (see also discussion)

DiscussionThe overall picture obtained from the results of the two studies is that homesicknessis quite prevalent among new-intake college students both in the Netherlands and theUK These results support the general ndings of investigators reviewed above (cf

162 Margaret Stroebe et al

Table 2 Mean scores for the homesickness subscales for the NL and UK samples

Adjusted means per scale NL UK F(1495)

Missing family 2148 2925 646Missing friends 2378 2715 116Loneliness 1788 1951 31Adjustment dif culties 1800 2046 79Ruminations about home 1518 1658 29

Total homesickness 1926 2259 207

p lt 05 p lt 01 p lt 001

Table 3 Frequency distribution of having felt homesick by study

Never Rarely Sometimes Often Very often N

NL 513 252 178 43 13 230UK 179 425 296 79 21 280

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 17: res_6

Eurelings-Bontekoe et al 1994 Fisher 1989 van Tilburg 1998) However there aresubstantial differences between the two countries in susceptibility Students in the UKare substantially more homesick Whereas approximately half of the Dutch studentssaid they had been homesick at least to some extent since coming to college this wastrue for more than 80 of the students at the British university This gure is higherthan that reported by other investigators (eg Fisher amp Hood 1987) possibly due todifferences in the measures used Whereas 5 of the Dutch students often felt home-sick this was true for 10of the British ones Not only were more of them homesickbut they suffered more intensely from it They missed both their family and their friendsmore and they had greater adjustment difculties than the Dutch students There werealso non-signicant tendencies for the British students to be more lonely and toruminate more about home This cross-cultural comparison is to our knowledge therst that uses the same instrument and controls for confounding factors (eg durationat college and time span of homesickness age) For all other studies it is difcult tocompare the amount or the intensity of homesickness for a specic period at collegeThis result raises interesting questions about the source of the cultural differenceswhich we discuss below

To what extent have our results shown that homesickness is associated with oreven a causal factor in distress or depression In our view the association is strongand indications are that homesickness affects distress or depression The limitationsof structural equation modelling must be taken into account in making such causalinterpretations particularly in cross-sectional studies such as ours (MacCullum ampAustin2000) In our view however the results showed more support for our hypothesizedmodel (homesickness preceding depression) than feasible alternatives (eg depressionand homesickness occurring simultaneously) First while relocation per se does notshow a signicant relationship to distress or depression the path analysis suggeststhat homesickness mediates these negative health consequences This implies that itis not distress or depression that precedes homesickness but that homesickness mayactually bring about distress or depression This follows hypotheses formulated fromthe DPM relocation would be antecedent to both loss (of family and home) and change(adjustment to the new situation) which themselves are associated with distress ratherthan vice versa Homesickness seems to play a mediating role between the stressor andoutcome This tentative conclusion must be further tested empirically using a long-itudinal design So far we can say that there is some justication for viewing home-sickness as a lsquomini-griefrsquo experience with consequences not unlike (if not as extreme)as those following bereavement (cf Stroebe amp Stroebe 1987) For example the data tthis interpretation better than a conceptualization of homesickness as only an aspectof emotional instability The results shown in Fig 2 are indicative of this The rela-tionship between homesickness and depressed mood remains strong even when thevulnerability factor of emotional instability is entered into the equation

It is plausible that homesick depressed students who are maladapted to their newsurroundings would be unlikely to function well academicallymdashor in other respectsfor that matter (cf Archer et al 1998 Burt 1993) For 10of the British and 5of theDutch who suffer frequently from homesickness this seems likely but there are stilllarger proportions who suffer sometimes and who may add to the size of this risk groupA longitudinal follow-up investigation would be required to validate such postulatedcausal chains (see below) Such a study also needs to provide further investigation ofparameters of the DPM eg the oscillation principle and the relationship of insecurestyles of attachment to coping strategies

Homesickness among students in two cultures 163

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 18: res_6

As expected students who have been at college longer were found to be slightly lesshomesick although some indeed remain homesick even at longer durations Given thisnding it would seem important to target students early on to try to facilitate theiradjustment away from home How could this be approached The data suggested thatnot only is homesickness associated with and probably antecedent to depression butthat personal feelings of competence or stability mediated this relationship as measuredby an emotional stability scale (NL) or self-likingcompetence scale (UK) One sugges-tion would therefore be to work with students to raise their feelings of competenceand perhaps to develop their self-esteem and perceived self-efcacy eg by monitoringtheir successes over the rst few weeks of the semester

Patterns emerging from the path analyses on separate subscales of homesicknesssuggest further aspects that intervention might focus on namely those to do withadjustment to the college situation We noted the crucial role that adjustment difcultiesto the college situation may play in relocated students missing family and friendsruminating about home and feeling lonely and showed that loneliness is linked todepression We further learnt that emotionally unstable students are particularly pronenot only to having adjustment difculties but also that these are the ones who frequentlyscore highly on the homesickness subscales

Turning to the cultural differences why should UK students be more homesickthan those in the NLOur original hypothesis was that accessibility to home would makethe critical difference and it seems likely that this is at least a partial explanation It isnoteworthy that results on the geographic indicator (distance from home) showedthat there was only an increase in the likelihood of high scores on one homesicknesssubscale namely missing the family (which could be interpreted as supportive ofthe attachment theory predictions) Further investigation of the accessibility to homevariable would be useful particularly in relationship to geographic distance In thiscontext it is interesting to note that Fisher found no relationship with distance fromhome among boarding school children (Fisher Murray amp Frazer 1985) but did amongcollege students (Fisher Frazer amp Murray 1986) Perhaps the critical differencebetween school and college students is the fact that the latter are more free to comeand go between college and home whereas boarding school pupils are typically onlyallowed home at set holiday periods Accessibility or a perceived sense of closenessmay be crucial

Another curious nding that needs to be considered alongside the cultural differ-ence in homesickness is the fact that UK females were the ones to suffer most fromhomesickness We can only speculate about the reason for this Could it be that femalestudents in general are indeed more homesick than males as we predicted they wouldbe but that in the NL sample this was mitigated by going home at weekends whichstudents in NL as opposed to UK frequently do The accessibility of home is greaterin NL than it is in UK due to the comparative geographic proximity (the maximumdistance from home in NLwould be no more than a couple of hours) and the provisionto all Dutch students of free public transportation nationwide Perhaps this is of morebenet to females than males given indications that females are more relationship-oriented than males (cf de Ridder 2000)

Finally following attachment theory and the DPM we made the prediction thatthose students who experienced problematic relationships with their parents would bethe ones to feel more homesickmdasha somewhat counter-intuitive prediction given thatone could also imagine that problems are well left behind In fact there was someevidence that attachment insecurity (NL) or conict with parents (UK) were related

164 Margaret Stroebe et al

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 19: res_6

to certain aspects of homesickness Interestingly too a differentiated picture emerged inthe UK data while motherrsquos perceived aggression was associated with reduced home-sickness (missing family subscale) perceived physical threat on the part of a father wasassociated with increased homesickness (adjustment subscale) Thus though partiallysupportive the results do not provide unequivocal support for the predictions fromattachment theory However it is questionable whether the operationalizations wehave used are adequate indices of secure versus insecure attachment As noted abovethese aspects of the study are exploratory The results are intriguing enough to meritfurther investigation of the relationship between attachment and homesickness Thefact that Brewin et al (1989) found dependency on others to be a predictor of home-sickness adds weight to the argument that relationship to family (particularly parental)members may be one of the most critical variables and one to be explored furtherin future investigations Thus attachment theory emerges as a potentially importanttheoretical perspective for homesickness research and a necessary one for furtherintegration in the more specic DPM

In conclusion it can be stated that homesickness is a widespread problem amongstudents who have recentlymoved to college It occurs more frequentlyamong studentsin the UK than in NL This could reect a difference in accessibility to home Femalestudents in the UKare the highest risk group Perhaps females are more vulnerable whenthere is lack of access to home Homesickness is associated with distress and depres-sionmdashacting we think as their antecedentmdashand it is associated with other debilitatingpsychological states including loneliness Given the links identied in these data setsthere are good reasons to argue that homesickness is indeed a lsquomini-griefrsquo phenomenonLike bereavement the mental health consequences are negative and debilitating In ourview the patterns identied in this cross-sectional investigation of homesickness needfurther investigation and theoretical understanding of the phenomena is essential Thelsquomini-griefrsquo interpretation deriving from the DPMhas explanatorypotential Thus we arenow starting a new longitudinal study This focuses on styles of attachment and copingaiming to test predictions from the theoretical perspective outlined above

AcknowledgementsThis paper was prepared while Miles Hewstone was a Fellow at the Center for Advanced Study inthe Behavioral Sciences Stanford He gratefully acknowledges nancial support provided by theWilliam and Flora Hewlett Foundation

ReferencesArcher J (1999) The nature of grief The evolution and psychology of reactions to loss London

RoutledgeArcher J Ireland J Amos S Broad H amp Currid L (1998) Derivation of a homesickness

scale British Journal of Psychology 89 205ndash221Arrindell W A amp Ettema J H M (1975) SCL-90 Handleiding bij een multidimensionele

psychopathologie-indicator [Guide to a multidimensional indicator of psychopathology]Lisse Swets amp Zeitlinger

Baier M ampWelch M (1992) An analysis of the concept of homesickness Archives of PsychiatricNursing 6 54ndash60

Baron R M amp Kenny D A (1986) The moderatorndashmediator variable distinction in socialpsychological research Conceptual strategic and statistical considerations Journal ofPersonality and Social Psychology 51 1173ndash1182

Homesickness among students in two cultures 165

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 20: res_6

Beck A T Ward C H Mendelson M Mock J E amp Erbaugh J (1961) An inventory formeasuring depression Archives of General Psychiatry 4 561ndash571

Bowlby J (1980) Attachment and loss Vol 3 Sadness and depression London The HogarthRes

Brewin C R Furnham A amp Howes M (1989) Demographic and psychological determi-nants of homesickness and conding among students British Journal of Psychology 80 467 ndash477

Burt C D B (1993) Concentration and academic ability following transition to universityAn investigation of the effects of homesickness Journal of Environmental Psychology 13 333 ndash342

Carden A amp Feicht R (1991) Homesickness among American and Turkish college studentsJournal of Cross Cultural Psychology 22 418ndash428

De Jong R D Van Eck H C M amp Van den Bos K (1994) The big ve personalityfactors leadership and military functioning In B de Raad W K B Hostee andG L van Heck (Eds) Personality psychology in Europe (Vol 5 pp 216ndash221) TilburgTilburg University Press

de Ridder D (2000) Gender stress and coping Do women handle stressful situations differentlyfrom men In L Sherr ampJ S St Lawrence (Eds) Women health and the mind (pp 115ndash135)Chichester Wiley

Eurelings-Bontekoe E H M Tolsma A Verschuur M J amp Vingerhoets A J J M (1996)Construction of a homesickness questionnaire using a female population with two types ofself-reported homesickness Preliminary results Personality and Individual Differences 20 415 ndash421

Eurelings-Bontekoe E H M Verschuur M Koudstaal A van der Schar S amp Duijsens I J(1995) Construction of a homesickness-questionnaire Preliminary results Personality andIndividual Differences 19 319 ndash325

Eurelings-Bontekoe E H M Vingerhoets A amp Fontijn T (1994) Personality and behavioralantecedents of homesickness Personality and Individual Differences 16 229ndash235

Fisher S (1989) Homesickness cognition and health London ErlbaumFisher S Elder L amp Peacock G (1990) Homesickness in a school in the Australian Bush

Chilrenrsquos Environments Quarterly 7 15ndash22Fisher S Frazer N amp Murray K (1984) The transition from home to boarding school Adiary-

style analysis of the problems and worries of boarding school pupils Journal of Environ-mental Psychology 4 211 ndash221

Fisher S Frazer N amp Murray K (1986) Homesickness and health in boarding school childrenJournal of Environmental Psychology 6 35ndash47

Fisher S amp Hood B (1987) The stress of the transition to university A longitudinal study ofpsychological disturbance absent-mindedness and vulnerability to homesickness BritishJournal of Psychology 78 425 ndash441

Fisher S amp Hood B (1988) Vulnerability factors in the transition to university Self-reportedmobility history and sex differences as factors in psychological disturbance British Journalof Psychology 79 309 ndash320

Fisher S Murray K amp Frazer N (1985) Homesickness health and efciency in rst yearstudents Journal of Environmental Psychology 5 181ndash195

Goldberg L R (1992) The development of markers for the Big-Five factor structure Psycho-logical Assessment 4 26ndash42

Hazan C amp Shaver P (1987) Romantic love conceptualized as an attachment process Journalof Personality and Social Psychology 52 511 ndash524

Hojat M ampHerman M W (1985) Adjustment and psychosocial problems of Iranian and Filipinophysicians in the US Journal of Clinical Psychology 41 130ndash136

Jaccard J amp Wan C K (1996) LISRELapproaches to interaction effects in multiple regressionThousand Oaks CA Sage

Lazarus R S amp Folkman S (1984) Stress appraisal and coping New York Springer

166 Margaret Stroebe et al

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 21: res_6

Lu L (1990) Adaptation to British universities Homesickness and mental health of Chinesestudents Counselling Psychology Quarterly 3 225ndash232

MacCallum R C amp Austin J T (2000) Application of structural equation modeling inpsychological research Annual Review of Psychology 51 201ndash226

Parkes C M Stevenson-Hinde J amp Marris P (1991) Attachment across the life cycle Londonand New York Routledge

Sobel M (1982) Asymptotic condence intervals for indirect effects in structural equationmodels In S Leinhardt (Ed) Sociological methodology 1982 San Francisco CA JosseyBass

Straus M A Hamby S L Boney-McCoy S ampSugarman D B (1996) The Revised Conict TacticScales (CTS2) Development and preliminary psychometric data Journal of Family Issues 17 283ndash316

Stroebe M amp Schut H (1999) The Dual Process Model of Coping with Bereavement Rationaleand description Death Studies 23 197ndash224

Stroebe M S Stroebe W amp Hansson R O (Eds) (1993) Handbook of bereavement Theoryresearch and intervention New York Cambridge University Press

Stroebe W amp Stroebe M (1987) Bereavement and health The psychological and physicalconsequences of partner loss New York Cambridge University Press

Stroebe W Stroebe M ampSchut H (1993) Working through loss Does it help Paper presentedat the Society of Experimental Social Psychology Santa Barbara CA

Tafarodi R W amp Swann W B (1995) Self-liking and self-competence as dimensions ofglobal self-esteem Initial validation of a measure Journal of Personality Assessment 65 322ndash342

van Tilburg M A L (1998) When it hurts to leave home Unpublished PhD thesis TilburgUniversity The Netherlands

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1996) Homesickness A reviewof the literature Psychological Medicine 26 899ndash912

van Tilburg M A L Vingerhoets A J J M amp van Heck G L (1997) Coping with homesick-ness The construction of the Adult Homesickness Coping Questionnaire Personality andIndividual Differences 22 901ndash907

van Vliet A Stroebe W amp Schut H (1998) Verhuizen heimwee en gezondheidsklachten bijeerstejaars studenten [Relocation home sickness and health with Dutch university fresh-men] Gedrag amp Gezondheid 26 91ndash100

Ward C amp Kennedy A (1993) Psychological and socio-cultural adjustment during cross-cultural transitions Acomparison of secondary students overseas and at home InternationalJournal of Psychology 28 129ndash147

Watson D amp Pennebaker J W (1989) Health complaints stress and distress Exploring thecentral role of negative affectivity Psychological Review 96 234ndash254

Received 22 December 1999 revised version received 28 December 2000

Homesickness among students in two cultures 167

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00

Page 22: res_6

168 Margaret Stroebe et al

App

endi

x

Mea

nss

tand

ard

devi

atio

nsan

dco

rrel

atio

nco

ef

cien

tsof

the

NL

and

UK

sam

ple

s

MSD

Pear

son

corr

elat

ions

NL

sam

ple

(N=

204)

12

34

56

78

910

111

Emot

iona

lsta

bilit

y5

401

291

002

Dep

ress

ion

316

812

07

plusmn0

711

003

Secu

rendashi

nsec

ure

129

045

plusmn0

380

441

004

Gen

der

180

040

plusmn0

070

130

101

005

Dur

atio

n19

800

201

610

10plusmn

007

plusmn0

07plusmn

016

100

6A

djus

tmen

tdi

fcu

lties

173

088

plusmn0

410

540

210

00plusmn

014

100

7M

issi

ngfa

mily

209

092

plusmn0

290

390

120

10plusmn

011

049

100

8M

issi

ngfr

iend

s2

281

02plusmn

038

042

012

004

plusmn0

160

520

601

009

Lone

lines

s1

720

84plusmn

051

073

036

003

plusmn0

090

760

410

501

0010

Rum

inat

ions

abou

tho

me

149

067

plusmn0

380

520

200

01plusmn

007

060

051

056

061

100

11To

tals

core

hom

esic

knes

s1

860

70plusmn

049

064

025

005

plusmn0

150

830

760

810

810

791

00

UK

sam

ple

(N=

215)

12

34

56

78

910

111

Self-

likin

gse

lf-co

mpe

tenc

e73

73

107

81

002

BDI

787

498

plusmn0

531

003

Dis

tanc

e28

611

908

27plusmn

003

plusmn0

041

004

Gen

der

172

045

plusmn0

110

040

071

005

Dur

atio

n49

62

125

40

050

00plusmn

002

006

100

6A

djus

tmen

tdi

fcu

lties

205

082

plusmn0

370

390

090

18plusmn

007

100

7M

issi

ngfa

mily

292

096

plusmn0

020

120

140

15plusmn

016

037

100

8M

issi

ngfr

iend

s2

730

95plusmn

026

038

011

014

plusmn0

130

610

481

009

Lone

lines

s1

950

91plusmn

040

050

013

015

plusmn0

120

740

340

681

0010

Rum

inat

ions

abou

tho

me

163

083

plusmn0

310

450

240

09plusmn

010

063

037

069

065

100

11To

tals

core

hom

esic

knes

s2

260

72plusmn

033

045

018

018

plusmn0

150

820

650

870

840

821

00