F

LRa

b

c

a

ARR3A

KIFH

cFco1iisAlHrnsp

Nf

0h

Early Childhood Research Quarterly 28 (2013) 210– 217

Contents lists available at SciVerse ScienceDirect

Early Childhood Research Quarterly

riendship similarities in internalizing problems in early childhood

isanne L. Stonea,∗, Matteo Gilettab, Mara Brendgenc, Roy Ottena,utger C.M.E. Engelsa, Jan M.A.M. Janssensa

Behavioural Science Institute, Radboud University Nijmegen, The NetherlandsDepartment of Psychology, University of North Carolina at Chapel Hill, NC, United StatesDepartement of Psychology, University of Québec, Montreal, Canada

r t i c l e i n f o

rticle history:eceived 19 April 2012eceived in revised form0 November 2012ccepted 7 December 2012

eywords:nternalizing problemsriendshipomophily hypothesis

a b s t r a c t

A key factor identified in friendship formation and stability is similarity. Homophily of externalizingproblems has been reported frequently, but less attention has been directed at homophily of internalizingproblems. Whether young children who are friends resemble each other in their internalizing problemsis thus largely unknown. In order to increase understanding of the social risk factors implicated in theetiology of internalizing problems, it is important to establish whether internalizing problems clusterin friendships. The present study examines homophily of internalizing problems while controlling forexternalizing problems in a sample of children aged 4–8.Methods: A community sample of 1584 children or 792 dyads with a mean age of 5.85 (SD = 1.26) par-ticipated. Of these friendship dyads, 557 were reciprocal and 235 unilateral friends. Internalizing andexternalizing problems were reported by teachers.

Results: Intraclass correlations were calculated to test for similarities between reciprocal and unilateralfriends. Support was found for the homophily hypothesis, in that reciprocal, but not unilateral friends,were substantially similar in their internalizing problems.Conclusions: In reciprocal friendships we found substantial similarity. From these results, we might con-clude that homophily is already present in friendships of young children regarding internalizing problems.Theoretical and practical implications of these findings are discussed.

© 2012 Elsevier Inc. All rights reserved.

Internalizing problems belong to the most common forms ofhild psychopathology (Canino et al., 2004; Egger & Angold, 2006;ord, Goodman, & Meltzer, 2003; Ollendick & King, 1994) and areharacterized by disturbances in emotions and moods, which mostften result in anxiety and depressive disorders (Ollendick & King,994; Zahn-Waxler, Klimes-Dougan, & Slattery, 2000). Internaliz-

ng problems refer to those feelings of distress that are directednwards and signify a core disturbance in emotions and moods,uch as sorrow, guilt, fear, and worry (Zahn-Waxler et al., 2000).s prevalence rates of internalizing problems are highest in ado-

escence and adulthood (e.g., Cole et al., 2002; Keenan-Miller,ammen, & Brennan, 2007; Twenge & Nolen-Hoeksema, 2002),

esearch has focused primarily on adolescent and adult inter-

alizing problems (Zahn-Waxler et al., 2000). However, evidenceuggests that the foundation for the development of internalizingroblems is laid in childhood (Hankin, Badanes, Abela, & Watamura,

∗ Corresponding author at: Behavioural Science Institute, Radboud Universityijmegen, P.O. Box 9104, 6500 HE Nijmegen, The Netherlands. Tel.: +31 24 3615924;

ax: +31 24 3612776.E-mail address: L.Stone@pwo.ru.nl (L.L. Stone).

885-2006/$ – see front matter © 2012 Elsevier Inc. All rights reserved.ttp://dx.doi.org/10.1016/j.ecresq.2012.12.003

2010; Luby, Si, Belden, Tandon, & Spitznagel, 2009; Zahn-Waxleret al., 2000). So, to understand the dramatic increase in internalizingproblems during adolescence, investigating factors associated withchildhood internalizing problems is essential (Cicchetti & Cohen,2006; Luby, 2010; Ollendick & King, 1994; Zahn-Waxler et al.,2000).

Specifically, social risk factors are significantly related to ado-lescent internalizing problems in that internalizing problems tendto cluster in friendships (Brendgen, Lamarche, Wanner, & Vitaro,2010; Giletta et al., 2011; Hogue & Steinberg, 1995; Prinstein, 2007;Stevens & Prinstein, 2005; Van Zalk, Van Zalk, Kerr, & Stattin, 2011).Moreover, exposure to a close reciprocal friend with internalizingproblems has been found to reinforce adolescents’ tendenciestoward depression (Giletta et al., 2011; Stevens & Prinstein,2005), which has recently been reported in eight-year-olds also(Mercer & DeRosier, 2010). The latter requires replication in asample of young children in order to ensure robustness of thesefindings. Thus, in order to increase understanding of the social

risk factors implicated in the etiology of internalizing problems, itis important to establish whether internalizing problems clusterin friendships. To address this issue, the present study examineswhether the degree of homophily in internalizing problems differs

Resear

bc

1

d(fsN(l(pTtftnafipBL

sai1ahBfbtRafVEP2KiMi&a&haacbfoitemMtbw

L.L. Stone et al. / Early Childhood

etween reciprocated and unilateral friendships during earlyhildhood.

. Friendship and internalizing problems

Friends may serve as cognitive and social resources for chil-ren, for example by promoting self-esteem and general self-worthe.g., Bagwell, Newcomb, & Bukowski, 1998; Hartup, 1996). Also,riends provide children with a context for social skills learning anderve as models for future relationship development (Bukowski,ewcomb, & Hartup, 1996). Most children have mutual friends

Hartup & Stevens, 1999), and children with internalizing prob-ems are as likely to have reciprocal best friends as other childrenLadd & Burgess, 1999). Further, internalizing problems do notredict change in number of friends over time (Brendgen, Vitaro,urgeon, Poulin, & Wanner, 2004; Ladd & Burgess, 1999). Contraryo these findings is that children with close relations may still sufferrom internalizing problems, as internalizing problems tend to clus-er in reciprocated friendships (Mercer & DeRosier, 2010). Indeedegative peer experiences (e.g., peer rejection and victimization)re important predictors of internalizing problems; likewise, dif-culties with friendships are associated with several adjustmentroblems, such as anxiety and depressed affect (Berndt, 1996;ukowski et al., 1996; Hodges, Boivin, Vitaro, & Bukowski, 1999;add & Kochenderfer, 1996; Rubin, Coplan, & Bowker, 2009).

It is widely thought that formation and maintenance of friend-hips are to a large part driven by preferences of similarities inppearances, such as age, sex, racial and ethnic background, behav-ors, and opinions (i.e., selection processes; Aboud & Mendelson,996; Rubin, Lynch, Coplan, Rose-Krasnor, & Booth, 1994). Thisssumption is referred to in the literature as the ‘homophilyypothesis’ or ‘similarity attraction hypothesis’ (Berndt, 1982;yrne, 1971; Kandel, 1978). It has been argued that reciprocated

riendships may provide a primary context for mutual influence,ecause children have the greatest opportunity to interact ando share their personal feelings within these relationships (e.g.,ose, 2002). In accordance with this hypothesis, older children anddolescent friends have been found to resemble their reciprocalriends with regard to a host of externalizing behaviors (e.g., Engels,itaro, Den Exter Blokland, De Kemp, & Scholte, 2004; Poelen,ngels, Van der Vorst, Scholte, & Vermulst, 2007; Poulin et al., 1997;rinstein, Boergers, & Spirito, 2001; Prinstein, Meade, & Cohen,003; Reitz, Dekovic, Meijer, & Engels, 2006; Vitaro, Tremblay,err, Pagani, & Bukwoski, 1997), shyness and prosocial behav-

or (Haselager, Hartup, Van Lieshout, & Riksen-Walraven, 1998;rug, Hoza, & Bukowski, 2004), and to a lesser extent to internal-

zing problems such as depression (Brendgen et al., 2010; Hogue Steinberg, 1995; Prinstein, 2007; Stevens & Prinstein, 2005)nd anxiety (Rubin, Wojslawowicz, Rose-Krasnor, Booth-LaForce,

Burgess, 2006). In general, less attention has been directed toomophily with regard to internalizing problems in friendshipsnd peer groups (see Giletta et al., 2011; Van Zalk et al., 2011),nd specifically for young children, with one exception. Recently,hildren’s levels of social anxiety, loneliness, and depression haveeen found to become more similar to the average level of theirriends in eight-year-olds (Mercer & DeRosier, 2010). Hitherto, thusnly one study explicitly concentrated on friendship similaritiesn internalizing problems in children. However, while similari-ies regarding externalizing problems in friendships have beenxtensively reported and internalizing problems are highly co-

orbid with externalizing problems in early childhood (Costello,ustillo, Erkanli, Keeler, & Angold, 2003), this study did not con-

rol for externalizing problems. The reported findings may haveeen attributable to concurrent externalizing problems. Therefore,e will control for externalizing problems in all analyses.

ch Quarterly 28 (2013) 210– 217 211

1.1. Mechanisms of homophily in internalizing problems duringearly childhood

Why would we expect homophilous friendships regardinginternalizing problems during early childhood? Several mecha-nisms have been put forth explaining homophily of internalizingproblems in adolescence, such as co-rumination, adoption ofdepressogenic attributional styles, negative affect induction, inter-personal stressor generation, and operation of network processes(Coyne, 1976; Rose, 2002; Rudolph, Flynn, Abaied, Groot, &Thompson, 2009; Schaefer, Kornienko, & Fox, 2011). More specifi-cally, some of these mechanisms may underlie selection processes(i.e., children with problems actively seek out others to befriendwith similar problems) and ‘others’ socialization’ (i.e., close friendsreinforce each other’s problem behaviors). Due to the develop-mental level of 4–8-year-olds, not all of these mechanisms mayapply. Mechanisms focusing on emotional contagion of internal-izing problems may be most apt. Negative affect induction statesthat depressed individuals display maladaptive interactional stylesthat may induce depressed mood in others (Coyne, 1976), andrelatedly, interpersonal stressor generation posits that depressedchildren generate stress in their relationships which may not onlypromote or exacerbate their symptoms but also affect those of theirrelational partners (Rudolph et al., 2009). Indeed, more conflictand less collaboration are displayed in interactions of depressedchildren with peers in comparison to those of non-depressedchildren (Rudolph et al., 2000). In addition, ‘age-specific manifes-tation’ of symptoms of internalized distress, such as irritability,sadness, guilt, anhedonia and somatic complaints (Luby, 2010),may produce negative reactions in peers (Coyne, 1976). In linewith this reasoning is a recent study that showed depressioncontagion in 8-year-olds and adolescents. However, whereas co-rumination mediated depression contagion among adolescents, itdid not in children (Schwartz-Mette & Rose, 2012). Co-rumination,which is thought to affect friends’ internalizing problems byextensively discussing and re-visiting problems, speculating aboutproblems, and focusing on negative feelings (Rose, 2002), requiresadvanced cognitive capabilities such as perspective-taking andabstract thought, and may thus be not applicable to young chil-dren. Children below the ages of 7–8 have generally not beenfound to possess complex cognitive capabilities (Best & Miller,2010), although a meta-analysis showed that 6-year-olds’ attribu-tional style and depression are correlated (Joiner & Wagner, 1995).However, young children’s friendships are less intimate and char-acterized by less self-disclosure (McDougall & Hymel, 2007). Thus,chances are limited that children will engage in complex interac-tions that reveal their depressogenic attributional styles, and thatfriends will adopt these attributional styles (Stevens & Prinstein,2005). All in all, we argue that the latter cognitive mechanisms maynot be as plausible as the former emotional mechanism underlyinghomophily of internalizing problems during early childhood.

Similarity in internalizing problems between children involvedin a friendship may result from socialization processes, such asemotional contagion, but also from selection processes. That is,similarity in internalizing problems may affect children’s forma-tion of new relationships. In this regard, children with internalizingproblems have been found to have problematic interactions withpeers, which is also associated with social withdrawal, and socialwithdrawal with internalizing problems, although directionality ofeffects is somewhat unclear (Rubin et al., 2009). Withdrawal is alsoassociated with a reduced pool of friends to befriend, less chancesof transivity (i.e., friendships with a friend of a friend), and fewer

friendship ties through popularity. Thus, withdrawn children, whomay also display internalizing problems, are likely to befriend chil-dren outside the normative network processes (Schaefer et al.,2011). Following this reasoning, this may lead to higher chances

2 Resear

owscdmcule

1

sfrictslninwapvtaih2pscbatrhe(

2

2

4wPtrcifoctgyCw

12 L.L. Stone et al. / Early Childhood

f friendship with other marginalized children, such as childrenho exhibit internalizing problems and less chances of friend-

hip with children in the normative peer group. Thus, althoughhildren with internalizing problems may wish to befriend chil-ren who do not necessarily have internalizing problems, theyay not succeed. Therefore, in a sort of ‘default selection process’,

hildren with internalizing problems may be more likely to endp being friends with other children showing internalizing prob-

ems (Hektner, August, & Realmuto, 2000). As such, homophily isxpected in reciprocated, but not unilateral friendships.

.2. Study objectives

The present study aimed to identify homogeneity in friend-hips concerning internalizing problems in 4–8 year old childrenrom a large nation-wide Dutch sample. As research is very scarceegarding similarity in friendships of young children, this studys of an explorative nature. First, based on theory and empiri-al findings during adolescence, we expected to find support forhe homophily hypothesis in friendships of young children, withtronger homophily expected in reciprocal friendships versus uni-ateral friendships (Giletta et al., 2011, 2012). Thus, children whoominate a peer as a friend who does not return the nomination,

.e., unilateral friendships, are thought to be less similar to theominee’s internalizing problems. On the contrary, two childrenho nominate each other as friends, i.e., reciprocal friendships,

re thought to be more similar to each other in their internalizingroblems. Second, as gender is one of the most powerful and uni-ersal dimensions for homophily, it is deemed important to assesshe moderating role of gender (Mehta & Strough, 2009). Duringdolescence, gender differences are evident due to the increasen internalizing symptoms in females but not males, and femalesaving more intimate friendships than males (Rose & Rudolph,006; Twenge & Nolen-Hoeksema, 2002). These differences inrevalence rates of internalizing problems and intimate friend-hips between boys and girls are less accentuated during earlyhildhood, thus, we hypothesize that gender differences woulde absent in homophily of internalizing problems. In addition tossessing the role of same-gender friendships, we also investigatehe moderating role of mixed-gender friendships as this is veryarely studied (Bleske-Rechek & Buss, 2001), especially in child-ood and regarding internalizing problems. Finally, we control forxternalizing problems, as comorbidity rates are high in childhoodCostello et al., 2003).

. Method

.1. Participants

Participants were 1584 Dutch children (M age = 5.9, SD = 1.26,9.6% boys) enrolled in the Kind in Zicht study. These childrenere selected from 2558 children who were eligible for the study.

arental consent was obtained for 2360 children. Subsequently,eachers administered questionnaires for all children in their class,esulting in a number of returned questionnaires of 2182. Of thehildren with parental consent, 1832 participated in sociometricnterviews, 1682 could be included in a dyadic friendship. Finally,or 49 dyads no teacher data were available leading to a final samplef 792 dyads or 1584 children. The majority of teachers (99.5%) andhildren (94.7%) were of Dutch origin. Children were recruited forhe study via 29 primary schools throughout the Netherlands. Four

rades were included in the study; the preschool and kindergartenear, first and second grade. In total, 110 classrooms participated.lass sizes varied from 7 to 36. Parents were informed via lettershich were handed out from the schools. Subsequently, passive

ch Quarterly 28 (2013) 210– 217

informed consent was collected with 8.4% of parents refusingto participate. Friendship nomination interviews were conductedfrom January until March 2010 outside of the classroom. A totalof 1832 children (83.9%), ranging in age from 3.75 to 9.08 years(M = 6.04, SD = 1.28) at the time of nomination, had complete friend-ship nomination interview data. Of these children 50.3% were boys.

2.2. Procedures

The Kind in Zicht study is a cohort study of Dutch children aged4–8. During the school year prior to data collection, 2008–2009,schools were randomly selected from the population of elementaryschools in the Netherlands. Schools in the larger provinces, Noord-Holland, Zuid-Holland, Noord-Brabant and Gelderland and the fourlargest cities, Amsterdam, Rotterdam, The Hague, and Utrecht, wereoversampled. In total, 440 schools were selected. Principals ofthese schools received a letter inviting them for the study. Sub-sequently, principals were asked for participation by phone, whichled to participation of 29 schools (6.6%), containing 2558 childrenin preschool and kindergarten classes, grade 1 and 2 classes. Onlythese grades were eligible for participation. Schools received 1000Euros for their participation. During the school year of data col-lection, 2009–2010, teachers handed out information and consentletters to parents. Consent of 2360 (92%) parents was obtained.Teachers completed the questionnaires digitally or by paper andpencil for all the children in their class, whose parents gave consent.There were no differences depending on mode of administrationfor the emotional problems t(1308) = −0.94, p = 0.35, somatic com-plaints t(1291) = 1.51, p = 0.13, anxious-depressed t(1290) = 1.18,p = 0.24, withdrawn t(1280) = 1.56, p = 0.12, and externalizing scalest(1216) = −0.466, p = 0.64. Sociometric interviews were held atschools from January until March 2010 in which 1832 (77.6%) chil-dren participated.

Children were considered to be a member of a reciprocalfriendship dyad when the classmate they nominated as a friendreciprocated the nomination as friend. Children were consideredunilateral friends when the classmate they nominated did notreciprocate the nomination. As children could name up to fivefriends, some participants were involved in more than one friend-ship at a time. There were 1047 children (62.4%) who named fewerthan 5 friends. The inclusion of all friendship dyads in the analy-ses would have introduced the potential for bias due to unequalcontributions by individuals. Thus, to avoid these problems, anal-yses focused on the highest ranked reciprocal friends and eachparticipant was restricted to one and only one friendship dyad(n = 1682, M age = 5.85, SD = 1.26, 49.6% boys). The first criterionused to create dyads was the highest rank; the second was thatwe aimed to create as many dyads as possible, that is to match asmany participants as possible in friends’ dyads to avoid excludingthem from the analyses. Finally, when it was not possible to decide,we randomly chose in which dyads participants were included. Ofthese dyads, 49 did not have complete teacher reports leading toa final sample of 792 dyads. Chi-square tests were performed inorder to examine whether children with a reciprocal or unilateralfriendship differed from excluded children on socio-demographiccharacteristics (i.e., gender, age, ethnicity, educational level, fam-ily structure and marital status) and t-tests were used to testwhether level of internalizing and externalizing problems dif-fered between included and excluded children. Significant effectsemerged for age �2(8, N = 2182) = 43.72, p < 0.00, ethnicity �2(4,N = 1453) = 58.48, p < 0.00, educational level �2(7, N = 1503) = 38.29,p < 0.00, internalizing t(2236) = 2.66, p = 0.008, and externalizing

problems t(2236) = 3.57, p < 0.001. Young children with the Dutchnationality, mothers with a high educational level and low levelsof internalizing and externalizing symptoms were overrepresentedamong participants who had a reciprocal or unilateral friend versus

Resear

ce

2

2

aviai(taatrRrlrsw

2

i(1iTc2aeswi(Sttciostp

2

stdffiffwfcd

L.L. Stone et al. / Early Childhood

hildren who did not have these friendships. No significant effectsmerged for gender, family structure, and marital status.

.3. Measures

.3.1. Strengths and Difficulties QuestionnaireThe Dutch teacher informant version of the SDQ was used to

ssess internalizing and externalizing behavior (SDQ; Widenfelt,an Goedhart, Treffers, & Goodman, 2003). The subscales measur-ng emotional symptoms (e.g. many worries, often seems worried)nd conduct problems (e.g. often lies or cheats) each contain fivetems. Teachers rated children on a 3-point scale ranging from 0not true) to 2 (certainly true). The scoring procedures used inhis study are available online (Widenfelt et al., 2003). Cronbach’slpha was ̨ = 0.71 for emotional symptoms (M = 0.97, SD = 1.54)nd ̨ = 0.53 for conduct problems (M = 0.76, SD = 1.23). Reliabili-ies based on Structural Equation Modeling – known as Jöreskogho or McDonalds OmegaH (Jöreskog, 1971; McDonald, 1978, 1999;evelle & Zinbarg, 2009; Stone et al., in press) were 0.83 and 0.82,espectively. To control for conduct problems, the emotional prob-ems scale was regressed on the conduct problems scale, creating aesidual emotional problems score. The distribution of the residualcore was skewed, and therefore scores above the 99th percentileere rescaled to the 99th percentile value.

.3.2. (Caregiver-)Teacher report formThe Dutch versions of the C-TRF and TRF were used to assess

nternalizing and externalizing behavior as reported by teachersAchenbach & Rescorla, 2000, 2001; Verhulst, Van der Ende, & Koot,997). The C-TRF, used for children aged 1.5–5 years, comprises 100

tems; the TRF targets 5–18-year-olds and consists of 118 items.hese items are rated using a 3-point Likert scale, where 0 indi-ates responses of “not true”, 1“somewhat or sometimes true”, and

“very true or often true”. The C-TRF and TRF somatic complaints,nxious/depressed and withdrawn subscales were aggregated tonable comparability with the SDQ. The emotionally reactive sub-cale from the C-TRF was removed from the analyses as this scaleas not present in the TRF version. The three subscales compris-

ng the internalizing scale (i.e., somatic complaints ̨ = 0.51, 0.65M = 0.27, SD = 0.87), anxious/depressed ̨ = 0.75, 0.78, (M = 1.48,D = 2.31), and withdrawn ̨ = 0.74, 0.70, (M = 1.11, SD = 1.77), andhe externalizing scale ̨ = 0.93, 0.93, (M = 3.83, SD = 6.49) respec-ively for C-TRF and TRF were used in the current study. Again, toontrol for externalizing problems, the somatic complaints, anx-ous/depressed and withdrawn scales were, separately, regressedn the externalizing problems scale, thus creating three residualcores. The distributions of the residual scores were skewed, andherefore scores above the 99th percentile were rescaled to the 99thercentile value.

.3.3. Friendship nominationA sociometric one-on-one interview was used to identify friend-

hip dyads. A trained research assistant picked the children up fromheir classroom and took them to a quiet place in the school to sitown. The child was asked to nominate five of his/her best friendsrom class: “Who are your best friends in class? You can name

ve.” After naming classmates, children were asked to order these

riends: “Who of these friends you just named is your very bestriend?” This question was repeated until the five named friendsere ordered. The child was allowed to name fewer than five

riends. The child then returned to the classroom and took the nexthild to the research assistant. Responses of the child were writtenown verbatim by the interviewer.

ch Quarterly 28 (2013) 210– 217 213

2.3.4. EthnicityParents were asked to indicate what the child’s nationality was,

with the question “What is your child’s nationality?”. Answeringoptions were ‘Dutch’, ‘Moroccan’, ‘Turkish’, ‘Surinam’ or ‘different’.

2.4. Strategy for analysis

First, correlations were computed between the TRF SomaticComplaints, Anxious-Depressed, Withdrawn and ExternalizingProblems and the SDQ Emotional Problems scales. The unmo-dified scales were used when we computed these correlations.Subsequently, intraclass correlations between friends’ scores (bothreciprocal and unilateral) were computed with the residual scoresof each scale. To deal with the random assignment of some childrenas targets and others as friends (i.e., the indistinguishable nature ofthe dyad partners), a pairwise approach recommended by Griffinand Gonzalez (1995) was used. This requires all relationships to beentered twice, once with the target child’s scores entered first andthe friend’s score entered second, and once with the friend’s scoresentered first and the target child’s scores entered second. There-fore, intraclass correlations were computed to assess the degree ofsimilarity between dyad members. In order to account for inter-dependency between dyad members, the statistical significance ofthe pairwise correlations was adjusted according to the procedureproposed by Griffin and Gonzalez (1995) (see p. 432).

Subsequently, a Confirmatory Factor Analysis (CFA) was per-formed in MPlus with the internalizing subscales as indicators.Residual scores of the TRF Somatic Complaints, Anxious-Depressed,Withdrawn, and the SDQ Emotional Problems scales were used asinput for the CFA. CFA enables (a) obtaining one latent indicatorof internalizing problems and (b) removal of measurement errorleading to a more pure internalizing factor (Bentler & Kano, 1990;Crocker & Algina, 1986). For these analyses, we used maximumlikelihood estimation with the Huber–White covariance adjust-ment (MLR in Mplus 5.0; Muthén & Muthén, 1998–2007). Model fitwas assessed with various fit indices, including robust chi-squarewith estimated degrees of freedom (df), comparative fit index (CFI;Bentler, 1990), root mean squared error of approximation (RMSEA;Byrne, 1998), and Tucker–Lewis index (TLI; Tucker & Lewis, 1973).

Second, the latent internalizing factor scores were saved andused in subsequent dyadic analyses. Correlations between friend-ship types (reciprocal, unilateral) were computed for the latentinternalizing factor and the four indicators. In order to test gen-der moderation, correlations were computed separately for thesevariables, while keeping the distinction between reciprocal andunilateral friends. A series of analyses involving Fisher’s r to ztransformations were conducted to test the differences betweencorrelations. Differences were tested for (a) friendship types (b)gender between friendship types and (c) gender within friendshiptypes.

3. Results

3.1. Descriptive statistics

Correlations and intraclass correlations are presented for 792dyads (557 reciprocal, 235 unilateral) in Table 1. In order to con-trol for Externalizing Problems, partial intraclass correlations ofEmotional Problems, Somatic Complaints, Anxious-Depressed, andWithdrawn were computed and reported in Table 1. All studyvariables were positively related, with the internalizing subscales

emotional problems, somatic complaints, anxious-depressed andwithdrawn variables generally showing higher correlations witheach other, than with the externalizing scale. The intraclass cor-relations at the subscale level show that, across all dyads, there is

214 L.L. Stone et al. / Early Childhood Research Quarterly 28 (2013) 210– 217

Table 1Correlations between Indicators of the latent internalizing factor (N = 1584) and intraclass correlations (N = 792).

1 2 3 4 5 Intraclass r Intraclasspartial r

1 Emotional problems – 0.07 0.072 Somatic complaints 0.44** – 0.20** 0.15**

3 Anxious-depressed 0.72** 0.37** – 0.17** 0.15**

4 Withdrawn 0.38** 0.26** 0.39** – 0.17** 0.10*

5 Externalizing 0.21** 0.16** 0.21** 0.38** – 0.33** –

scpfi

3

twa(wifdca

Ffl(f(nsl

3

ccr

TIs

* p < 0.05.** p < 0.01.

ome evidence for similarity, specifically for the subscales somaticomplaints, anxious-depressed, withdrawn, and externalizing. Theartial intraclass correlations show that similarity for the threeormer subscales remains significant after controlling for external-zing problems.

.2. Latent dyadic factor model results

Confirmatory factor analysis with four indicators, emo-ional problems, somatic complaints, anxious-depressed, andithdrawn loading on a latent internalizing factor yielded

cceptable fit (�2(2) = 18.56, p = 0.000; CFI = 0.981; RMSEA = 0.071CI = 0.044–0.103); TLI = 0.944). To assess the degree of homophilyithin dyads involving reciprocated and unilateral friendships,

ntraclass correlations were computed for the latent internalizingactor scores, emotional problems, somatic complaints, anxious-epressed, and withdrawn scores. The intraclass correlations wereomputed separately within group, and also separately by gender,nd are shown in Table 2.

For reciprocal friends, significant correlations were found, andisher’s r to z analyses revealed that these differed significantlyrom correlations found for unilateral friends, respectively, for theatent internalizing factor (z = 2.23, p < 0.05), somatic complaintsz = 2.52, p < 0.05), and anxious-depressed (z = 2.44, p < 0.05), but notor the emotional problems (z = 1.94, p > 0.05) and withdrawn scalez = −0.21, p > 0.05). Correlations for unilateral friendships were allon-significant. Thus, children within a reciprocated friendshiphowed greater behavioral similarity than children within a uni-ateral friendship.

.3. Moderation between friendship types

In order to examine whether differences between recipro-al and unilateral friendships also differed for boys and girls,orrelations were computed separately for gender. For boys’eciprocal friendships, significant correlations were found, and

able 2ntraclass correlations between internalizing problems of children and their friends,eparately by dyad type (boy, girl, mixed dyads).

Friendship types Friend dyads

Total Boys Girls Mixed

Reciprocal (n = 557, 242 boys, 64 mixed)Latent internalizing 0.15** 0.19** 0.01 0.23Emotional problems 0.11** 0.16** 0.05 0.15Somatic complaints 0.20** 0.24** 0.23** 0.00Anxious-depressed 0.21** 0.23** 0.18** 0.22Withdrawn 0.09* 0.01 0.12 0.27*

Unilateral (n = 235, 96 boys, 45 mixed)Latent internalizing −0.02 −0.12 0.03 −0.03Emotional problems −0.04 −0.11 −0.02 −0.05Somatic complaints 0.01 0.08 0.03 −0.16Anxious-depressed 0.02 −0.07 0.06 0.01Withdrawn 0.11 −0.08 0.16 0.41**

* p < 0.05.** p < 0.01.

these differed significantly from correlations found for unilateralfriends, respectively, for the latent internalizing factor (z = 2.54,p < 0.05) and anxious-depressed (z = 2.45, p < 0.05), but not forthe emotional problems (z = 0.44, p > 0.05), somatic complaints(z = 1.33, p > 0.05) and withdrawn scales (z = 0.76, p > 0.05). Forgirls’ reciprocated friendships, significant correlations were foundfor two subscales, but these did not differ significantly from corre-lations for girls’ unilateral friendships, for the somatic complaints(z = 1.60, p > 0.05), and anxious-depressed (z = 1.01, p > 0.05) scales.Further, correlations for both boys’ and girls’ unilateral friendshipswere all non-significant. For mixed-gender reciprocated friend-ships a significant correlation was found only for the withdrawnscale and, remarkably, for unilateral mixed-gender friendshipsthe correlation for the withdrawn scale was also significant. Thecorrelations for reciprocated and unilateral friends did not differsignificantly (z = −0.95, p > 0.05), indicating that children in recip-rocal mixed-gender friendships show equally great behavioralsimilarity as children in unilateral mixed-gender friendships.

3.4. Moderation within friendship types

Analyses conducted separately by gender show that both boys’and girls’ reciprocated friendships are characterized by similarityin the somatic complaints and anxious-depressed scales. Further,reciprocal male friendships are similar in the latent internalizingand, emotional problems scales. None of the correlations differedbetween boys’ and girls’ friendships, nor for reciprocal or unilat-eral friendships. These findings indicate that boys and girls do notdiffer within friendship type in their similarity. The mixed-genderreciprocated friendships are characterized by similarity in the with-drawn scale. The correlations did not differ between mixed-genderand boys’, and girls’ reciprocal or unilateral friendships, exceptfor the correlations between unilateral mixed-gender and unilat-eral male friendships (z = 2.86, p < 0.01). These findings indicatethat mixed-sex gender friendships do not seem to differ withinfriendship type in their similarity, although they do differ regardingwithdrawn behavior. Unilateral mixed-gender dyads are more sim-ilar to each other regarding withdrawn behavior than unilateralmale dyads.

4. Discussion

Homophily of normative and maladaptive developmental char-acteristics has been very rarely studied in early or middle childhood(Dunn & Cutting, 1999; Eivers, Brendgen, Vitaro, & Borge, 2012;Gleason, Gower, Hohmann, & Gleason, 2005; Haselager et al., 1998),let alone homophily of internalizing problems (Mercer & DeRosier,2010). Furthermore, no studies yet have tested homophily in inter-nalizing problems and controlled for externalizing problems while(a) comorbidity rates are high in childhood (Costello et al., 2003)and (b) similarities in externalizing problems in friendships of chil-

dren have been reported (e.g., Dishion, Patterson, & Griesler, 1994).The current study extends prior research by exploring similarity ininternalizing problems between young children and their friends,with the goal of increasing understanding of social risk factors

Resear

iscasaSZDdyst4soPtaiTlec

niaDpsiaei

prsindbcwlsMaiacneFgf

hrhrifet

L.L. Stone et al. / Early Childhood

mplicated in the etiology of internalizing problems. Our resultsupport our hypotheses, in that similarity was present in recipro-al, but not unilateral, friendships of young children. These resultsre in accordance with the adolescent literature, where evidencehowing that homophily in internalizing problems is present, isccumulating (Brendgen et al., 2010; Giletta et al., 2011; Hogue &teinberg, 1995; Prinstein, 2007; Stevens & Prinstein, 2005; Vanalk et al., 2011), and with studies focusing on children (Mercer &eRosier, 2010; Rubin et al., 2006). However, in these studies, chil-ren were substantially older than in the current study (i.e., 8–10ears of age) and the sample size in the Rubin et al. study was quitemall. To our best knowledge, this study is the first to show thathere is a link between internalizing problems and friendships of–8-year-olds. Although the correlations reported in the currenttudy are not as high as those found in studies with older childrenr adolescents (Brendgen et al., 2010; Hogue & Steinberg, 1995;rinstein, 2007; Rubin et al., 2006; Stevens & Prinstein, 2005), wehink the magnitude of these correlations may be explained by thege of the children. As internalizing problems tend to peak dur-ng adolescence (e.g., Cole et al., 2002; Keenan-Miller et al., 2007;wenge & Nolen-Hoeksema, 2002), more variance in these prob-ems can be explained. In our opinion the limited variance found inarly childhood internalizing problems may account for the smallorrelations reported in this study.

Second, as internalizing problems often co-occur with exter-alizing problems, it was deemed important to establish whether

nternalizing problems cluster in friendships of young children,s externalizing problems do (Angold, Costello, & Erkanli, 1999;ishion et al., 1994). Because we controlled for externalizingroblems, our results support homophily of purely internalizingymptoms among friends. Prior studies on homophily of internal-zing problems did not take concurrent externalizing problems intoccount (Mercer & DeRosier, 2010; Rubin et al., 2006), and reportedffects might have been, at least partially, attributable to external-zing problems.

Further, our results show that homophily in friendships isresent regardless of gender, indicating the robustness of theeported findings. Regarding mixed-gender friendships, there wasome evidence for homophily. To our best knowledge, homophily innternalizing symptoms within mixed-gender dyads of children hasot been specifically examined yet. Also, we found that these chil-ren resembled each other most strongly in withdrawn behavior,oth in reciprocated and unilateral friendships. This suggests thathildren in mixed-gender friendships (i.e., reciprocal) and childrenho want to be friends with a child of the opposite sex (i.e., uni-

ateral) are both reported to show withdrawn behaviors. Althoughpeculative in nature, extant literature may explain these findings.ost children consider same-sex friendships and play styles more

cceptable than being friends with children of the other sex or hav-ng a play style of the other sex (Mehta & Strough, 2009). Having

mixed-gender friendship could be associated with gender atypi-al behavior, and evidence supports the notion that children reactegatively to atypical gender behavior of other children (e.g., Rublet al., 2007). This may be associated with withdrawn behavior.inally, as a substantial part (13.7%) of all friendships were mixed-ender in the current study, future research on mixed-genderriendships is warranted.

Although we were not able to test the mechanisms explainingomophily (i.e., selection and socialization processes), the cur-ent findings provide some indirect insights into the nature ofomophily in friendships of young children. Even young children’seciprocal friendships are characterized by similarity in internal-

zing problems, whereas their unilateral friendships are not. Thus,riendship may not always serve a protective function (cf. Rubint al., 2006), and possibly, for some children, it may be concep-ualized as a risk factor for early internalizing problems. Indeed,

ch Quarterly 28 (2013) 210– 217 215

although socially withdrawn children have been found to be aslikely as non-withdrawn children to have mutual friends, the qual-ity of their friendships has been reported to be lower (Rubinet al., 2006). Future studies should address whether all children inhomophilous internalizing friendships continue to develop symp-toms of depression and anxiety, or whether some do not, and thusto what extent homophilous internalizing problems pose a risk forsubsequent development (e.g., shy and socially withdrawn chil-dren; Gazelle & Rudolph, 2004). Further, from these findings, wemight infer that processes other than co-rumination might under-lie homophily, as it is unlikely that children this age are cognitivelyable to co-ruminate (Best & Miller, 2010). Although it is not evi-dent from our findings that these children may actually socializeeach other into higher levels of internalizing symptoms, the adoles-cent literature points to consistent evidence for socialization effects(e.g., Giletta et al., 2011) as has been reported in 8–10-year-oldsalso (Mercer & DeRosier, 2010). Children with internalizing prob-lems may induce depressed mood in their friends, and vice versa,and generate stress in their relationship, which may increase boththe children’s symptoms (Coyne, 1976; Rudolph et al., 2009).

As for selection processes, whether children with internalizingproblems actively seek out others to befriend with similar prob-lems, or whether they unintentionally end up with those childrenis not clear yet. The similarity attraction hypothesis explicitly, andthe homophily hypothesis more implicitly, assume that childrenbefriend based on preferences for similarity (Berndt, 1982; Byrne,1971; Kandel, 1978). In contrast, the default selection hypothe-sis states that friendship selection does not occur on the basisof similarity, but is due to a lack of availability instead of actualpreference for similarity (Hektner et al., 2000). Although amplestudies support the homophily hypothesis regarding similarityof behavioral characteristics (e.g. Giletta et al., 2011; Mercer &DeRosier, 2010; Schaefer et al., 2011), support for its preferencethesis is limited. Research on the default selection hypothesis isscarce, but two recent studies support the tenet that childrenprefer others as friends who are not similar to them in termsof psychopathology, but still are friends with those children thatdo show similar psychopathology (Schaefer et al., 2011; Sijtsema,Lindenberg, & Veenstra, 2010). By testing these two competinghypotheses against each other (cf. Sijtsema et al., 2010), the causesfor formation of homophilous friendships regarding internalizingproblems could be elucidated.

Although this study has important features, some limitationsshould be noted. First, this study has a cross-sectional design, whichdid not allow us to test possible mechanisms, such as selectionand socialization. A longitudinal design would allow us to exam-ine who were not friends at Time 1 but become friends at Time 2(i.e., selection) and to examine if friends become more similar toeach other over time (i.e., socialization) (Giletta et al., 2011). Fur-ther, future longitudinal studies are needed to assess when friendsbecome similar to each other in internalizing problems, as is shownin adolescence (Brendgen et al., 2010; Giletta et al., 2011; Hogue &Steinberg, 1995; Prinstein, 2007; Stevens & Prinstein, 2005; VanZalk et al., 2011). Second, our study was conducted in a generalpopulation where rates of internalizing problems are relativelylow. Replication of this study in a clinical sample is important to(a) explore whether clinically depressed children at this age havefriends who are similar to themselves and if so, (b) to investigatewhy children befriend other depressed children.

Third, this study used teacher reports to assess problembehaviors. The low concordance between teacher and self-reportsregarding internalizing problems is well-established (Achenbach,

McConaughy, & Howell, 1987), so self-reports may be used in futurestudies to investigate the homophily principle as well. Reliable andage-appropriate instruments that use self-reports are available andcan be used for this purpose, for children from age 4, such as the

2 Resear

BaVpsLotoaenccsntNtbwot

pthitcatawatopcwfciTlefHgiti

R

A

A

A

A

16 L.L. Stone et al. / Early Childhood

erkeley Puppet Interview, and the Dominic Interactive for school-ged children (BPI; Measelle, Ablow, Cowan, & Cowan, 1998; DI;alla, 2000). It is recommended to use parent reports to com-lement teacher- and self-reports, as a multi-informant approacheems to be most reliable in assessing internalizing problems (Deos Reyes & Kazdin, 2005; Stanger & Lewis, 1993). Fourth, reportsf symptoms of internalizing problems via questionnaires may beoo distal for measuring homophily in young children. Observationsf interactions between friends and what these interactions char-cterize may be more suited, as non-verbal and verbal behaviorncompasses information related to cognitions underlying inter-alizing problems. Relatedly, the reliability of the scale somaticomplaints was low, therefore results for this subscale specifi-ally should be interpreted with caution. Finally, the dyads wereelective in terms of age, nationality, educational level, and inter-alizing and externalizing symptoms. These findings may be due tohe relatively low participation rate of schools in our study. In theetherlands, these participation rates are quite typical, but with

he use of more incentives for the school, participation rates mayecome higher. Future research should attempt to include childrenith different nationalities, low SES backgrounds, and higher levels

f internalizing and externalizing problems in order to generalizehe findings to these populations.

Despite these limitations, results from this study may haveotential implications for practice. Although it is not clear fromhese findings whether children actually socialize each other intoigher levels of internalizing problems, the finding that internaliz-

ng problems cluster in friendships of these young children markshe importance of early intervention. Pairing children with psy-hopathological symptoms with those without symptoms seems

fruitful approach. Results from a summer treatment programargeting aggressive children are positive, such that children’sggression decreased when paired to a non-aggressive counterpart,hile the aggressive child did not evoke aggression in the non-

ggressive child (Hektner, August, & Realmuto, 2003). Of course,hese interventions targeting aggressive children need to be tail-red to the specific needs of children exhibiting internalizingroblems. Specifically, some techniques which may be effective forhildren with externalizing symptoms, may not work for childrenith internalizing symptoms (Granic, 2012). Treatment manuals

ocusing on internalizing problems mention the importance forhildren to get habituated to anxiety adequately before terminat-ng exposure of fear-inducing stimuli (Kendall & Hedtke, 2006).herefore, treatments targeting children with internalizing prob-ems should utilize a fear-hierarchy, whereby the child is onlyxposed to an interaction with the paired child if this feels safeor the child with internalizing symptoms (Chorpita, 2007). In theektner et al. study, paired children in the summer treatment pro-ram played a competitive game together. This competitive gamen itself may be fear-inducing for children with internalizing symp-oms. Depending on the specific symptoms of the target child, annteraction should be tailored such that this is not fear inducing.

eferences

boud, F., & Mendelson, M. (1996). Determinants of friendship selection and quality:Developmental perspectives. In W. M. Bukowski, A. F. Newcomb, & W. W. Hartup(Eds.), The company they keep: Friendships during childhood and adolescence (pp.87–112). New York, NY: Cambridge University Press.

chenbach, T., McConaughy, S., & Howell, C. T. (1987). Child/adolescent behavioraland emotional problems: Implications of cross-informant correlations for situ-ational specificity. Psychological Bulletin, 101, 213–232.

chenbach, T. M., & Rescorla, L. A. (2000). Manual for the ASEBA Preschool Forms &

Profiles. Burlington, VT: University of Vermont, Research Center for Children,Youth, & Families.

chenbach, T. M., & Rescorla, L. A. (2001). Manual for ASEBA School-age Forms &Profiles. Burlington, VT: University of Vermont, Research Center for Children,Youth, & Families.

ch Quarterly 28 (2013) 210– 217

Angold, A., Costello, E. J., & Erkanli, A. (1999). Comorbidity. Journal of Child Psychologyand Psychiatry, 40, 57–87.

Bagwell, C. L., Newcomb, A. F., & Bukowski, W. M. (1998). Preadolescent friend-ship and peer rejection as predictors of adult adjustment. Child Development, 69,140–153.

Bentler, P. M. (1990). Comparative fit indexes in structural models. PsychologicalBulletin, 107, 238–246.

Bentler, P. M., & Kano, Y. (1990). On the equivalence of factors and components.Multivariate Behavioral Research, 25, 67–74.

Berndt, T. J. (1982). The features and effects of friendship in early adolescence. ChildDevelopment, 53, 1447–1460.

Berndt, T. J. (1996). Friendship quality affects adolescents’ self esteem and socialbehavior. In W. M. Bukowski, A. F. Newcomb, & W. W. Hartup (Eds.), The companythey keep: Friendship during childhood and adolescence (pp. 346–365). New York,NY: Cambridge University Press.

Best, J. R., & Miller, P. H. (2010). A developmental perspective on executive function.Child Development, 81, 1641–1660.

Bleske-Rechek, A. L., & Buss, D. M. (2001). Opposite-sex friendship: Sex differencesand similarities in initiation, selection, and dissolution. Personality and SocialPsychology Bulletin, 27(10), 1310–1323.

Brendgen, M., Lamarche, V., Wanner, B., & Vitaro, F. (2010). Links between friendshiprelations and early adolescents’ trajectories of depressed mood. DevelopmentalPsychology, 46, 491–501.

Brendgen, M., Vitaro, F., Turgeon, L., Poulin, F., & Wanner, B. (2004). Is there a darkside of positive illusions? Overestimation of social competence and subsequentadjustment in aggressive and nonaggressive children. Journal of Abnormal ChildPsychology, 32, 305–320.

Bukowski, W. M., Newcomb, A. F., & Hartup, W. W. (1996). The company they keep:Friendship in childhood and adolescence. New York, NY: Cambridge UniversityPress.

Byrne, B. M. (1998). Structural equation modeling with Lisrel, Prelis, and Simplis: Basicconcepts, applications and programming. Mahwah, NJ: Erlbaum.

Byrne, D. (1971). The attraction paradigm. New York, NY: Academic Press.Canino, G., Shrout, P., Rubio-Stipec, M., Bird, H., Bravo, M., Ramirez, R., et al. (2004).

The DSM-IV rates of child and adolescent disorders in Puerto Rico. Archives ofGeneral Psychiatry, 61, 85–93.

Chorpita, B. (2007). Modular cognitive behavioral therapy for childhood anxiety disor-ders. NewYork, NY: The Guilford Press.

Cicchetti, D., & Cohen, D. (2006). Developmental psychopathology: Risk, disorder andadaptation. New York, NY: Wiley.

Cole, D. A., Tram, J. M., Martin, J. M., Hoffman, K. B., Ruiz, M. D., Jacquez, F. M., et al.(2002). Individual differences in the emergence of depressive symptoms in chil-dren and adolescents: A longitudinal investigation of parent and child reports.Journal of Abnormal Psychology, 111, 156–165.

Costello, E. J., Mustillo, S., Erkanli, A., Keeler, G., & Angold, A. (2003). Prevalence anddevelopment of psychiatric disorders in childhood and adolescence. Archives ofGeneral Psychiatry, 60, 837–844.

Coyne, J. C. (1976). Toward an interactional description of depression. Psychiatry:Journal for the Study of Interpersonal Processes, 39(1), 28–40.

Crocker, L., & Algina, J. (1986). Introduction to classical and modern test theory. FortWorth, TX: Holt, Rinehart & Winston.

De Los Reyes, A., & Kazdin, A. E. (2005). Informant discrepancies in the assessmentof childhood psychopathology: A critical review, theoretical framework, andrecommendations for further study. Psychological Bulletin, 131, 483–509.

Dishion, T. J., Patterson, G. R., & Griesler, P. C. (1994). Peer adaptations in thedevelopment of antisocial behavior: A confluence model. In R. Huesmann (Ed.),Aggressive behavior: Current perspectives (pp. 61–95). New York, NY: PlenumPress.

Dunn, J., & Cutting, A. L. (1999). Understanding others, and individual differences infriendship interactions of young children. Social Development, 8, 201–219.

Egger, H. L., & Angold, A. (2006). Common emotional and behavioral disorders inpreschool children: Presentation, nosology epidemiology. Journal of Child Psy-chology and Psychiatry, 47, 313–337.

Eivers, A., Brendgen, M., Vitaro, F., & Borge, A. I. H. (2012). Concurrent and longitudi-nal links between children’s and their friends’ antisocial and prosocial behaviorin preschool. Early Childhood Research Quarterly, 27, 137–146.

Engels, R. C. M. E., Vitaro, F., Den Exter Blokland, E., De Kemp, R., & Scholte, R. H. J.(2004). Influence and selection processes in friendships and adolescent smokingbehavior: The role of parental smoking. Journal of Adolescence, 27, 531–544.

Ford, T., Goodman, R., & Meltzer, H. (2003). The British child and adolescent mentalhealth survey 1999: The prevalence of DSM-IV disorders. Journal of the AmericanAcademy Child and Adolescent Psychiatry, 42, 1203–1211.

Gazelle, H., & Rudolph, K. D. (2004). Moving toward and moving away from theworld: Social approach and avoidance trajectories in anxious youth. Child Devel-opment, 75, 829–849.

Giletta, M., Scholte, R. H. J., Burk, W. J., Engels, R. C. M. E., Larsen, J. K., Prinstein,M. J., et al. (2011). Similarity in depressive symptoms in adolescents’ friendshipdyads: Selection or socialization? Developmental Psychology, 47, 1804–1814.

Giletta, M., Scholte, R. H. J., Prinstein, M. J., Engels, R. C. M. E., Rabaglietti, E., & Burk,W. J. (2012). Friendship context matters: Examining the domain specificity ofalcohol and depression socialization among adolescents. Journal of Abnormal

Child Psychology.,. Advance online publication, doi: 10.1007/s10802-012-9625-8.

Gleason, T., Gower, A., Hohmann, L., & Gleason, T. (2005). Temperament and friend-ship in preschool-aged children. International Journal of Behavioral Development,29, 336–344.

Resear

G

G

H

H

H

H

H

H

H

H

J

J

K

K

K

L

L

L

L

M

MM

M

M

M

M

M

O

P

P

P

L.L. Stone et al. / Early Childhood

ranic, I. (2012). Don’t worry, be angry: Anxiety, aggression and repair inparent–child relationships. Unpublished inaugural speech, Radboud UniversityNijmegen, The Netherlands.

riffin, D., & Gonzalez, R. (1995). Correlational analysis of dyad-level data in theexchangeable case. Psychological Bulletin, 118, 430–439.

ankin, B. L., Badanes, L. S., Abela, J. R., & Watamura, S. E. (2010).Hypothalamicpituitary–adrenal axis dysregulation in dysphoric children andadolescents: Cortisol reactivity to psychosocial stress from preschool throughmiddle adolescence. Biological Psychiatry, 68, 484–490.

artup, W. W. (1996). The company they keep: Friendships and their developmentalsignificance. Child Development, 67, 1–13.

artup, W. H., & Stevens, N. (1999). Friendships and adaptation across the life span.Current Directions in Psychological Science, 8, 76–79.

aselager, G. J. T., Hartup, W. W., Van Lieshout, C. F. M., & Riksen-Walraven, J. M. A.(1998). Similarities between friends and nonfriends in middle childhood. ChildDevelopment, 69, 1198–1208.

ektner, J. M., August, G. J., & Realmuto, G. M. (2000). Patterns and temporalchanges in peer affiliation among aggressive and nonaggressive children par-ticipating in a summer school program. Journal of Clinical Child Psychology, 29,603–614.

ektner, J. M., August, G. J., & Realmuto, G. M. (2003). Effects of pairing aggressiveand nonaggressive children in strategic peer affiliation. Journal of Abnormal ChildPsychology, 31, 399–412.

odges, E. V., Boivin, M., Vitaro, F., & Bukowski, W. M. (1999). The power of friend-ship: Protection against an escalating cycle of peer victimization. DevelopmentalPsychology, 35, 94–101.

ogue, A., & Steinberg, L. (1995). Homophily of internalized distress in adolescentpeer groups. Developmental Psychology, 31, 897–906.

oiner, T., & Wagner, K. (1995). Attributional style and depression in children andadolescents: A meta-analytic review. Clinical Psychology Review, 15, 777–798.

öreskog, K. G. (1971). Statistical analysis of sets of congeneric tests. Psychometrika,36, 109–133.

andel, D. B. (1978). Similarity in real-life adolescent friendship pairs. Journal ofPersonality and Social Psychology, 36, 306–312.

eenan-Miller, D., Hammen, C. L., & Brennan, P. A. (2007). Health outcomes relatedto early adolescent depression. Journal of Adolescent Health, 41, 256–262.

endall, P. C., & Hedtke, K. A. (2006). The coping cat workbook. Ardmore, PA: Work-book Publishing.

add, G. W., & Burgess, K. B. (1999). Charting the relationship trajectories of aggres-sive, withdrawn, and aggressive/withdrawn children during early grade school.Child Development, 70, 910–929.

add, G. W., & Kochenderfer, B. J. (1996). Linkages between friendship and adjust-ment during early school transitions. In W. M. Bukowski, A. F. Newcomb, & W.W. Hartup (Eds.), The company they keep: Friendship in childhood and adolescence(pp. 322–345). New York, NY: Cambridge University Press.

uby, J. L. (2010). Preschool depression: The importance of identification of depres-sion early in development. Current Directions in Psychological Science, 19, 91–95.

uby, J. L., Si, X., Belden, A., Tandon, M., & Spitznagel, E. (2009). Preschool depression:Homotypic continuity and course over 24 months. Archives of General Psychiatry,66, 897–905.

cDonald, R. P. (1978). Generalizability in factorable domains: Domain validity andgeneralizability. Educational and Psychological Measurement, 38, 75–79.

cDonald, R. P. (1999). Test theory: A unified treatment. Mahwah, NJ: Erlbaum.cDougall, P., & Hymel, S. (2007). Same-gender versus cross-gender friendship con-

ceptions: Similar or different? Merrill-Palmer Quarterly, 53, 347–380.easelle, J. R., Ablow, J. C., Cowan, P. A., & Cowan, C. P. (1998). Assessing young

children’s views of their academic, social, and emotional lives: An evaluation ofthe Self-Perception Scales of the Berkeley Puppet Interview. Child Development,69, 1556–1576.

ehta, C. M., & Strough, J. (2009). Sex segregation in friendships and normativecontexts across the lifespan. Developmental Review, 29, 201–220.

ercer, S. H., & DeRosier, M. E. (2010). Selection and socialization of internaliz-ing problems in middle childhood. Journal of Social and Clinical Psychology, 29,1031–1056.

rug, S., Hoza, B., & Bukowski, W. M. (2004). Choosing or being chosen by aggressive-disruptive peers: Do they contribute to children’s externalizing and internalizingproblems? Journal of Abnormal Child Psychology, 32, 53–65.

uthén, L. K., & Muthén, B. O. (1998–2007). Mplus user’s guide. Los Angeles, CA:Muthén & Muthén.

llendick, T. H., & King, N. J. (1994). Diagnosis, assessment, and treatment of inter-nalizing problems in children: The role of longitudinal data. Journal of Consultingand Clinical Psychology, 62, 918–927.

oelen, E. A. P., Engels, R. C. M. E., Van der Vorst, H., Scholte, R. H. J., & Vermulst, A. A.(2007). Best friends and alcohol consumption in adolescence: A within-familyanalysis. Drug and Alcohol Dependence, 88, 163–173.

oulin, F., Cillessen, A. H. N., Hubbard, J. A., Coie, J. D., Dodge, K. A., & Schwartz, D.

(1997). Children’s friends and behavioral similarity in two social contexts. SocialDevelopment, 6, 224–236.

rinstein, M. J. (2007). Moderators of peer contagion: A longitudinal examination ofdepression socialization between adolescents and their best friends. Journal ofClinical Child and Adolescent Psychology, 36, 159–170.

ch Quarterly 28 (2013) 210– 217 217

Prinstein, M. J., Boergers, J., & Spirito, A. (2001). Adolescents’ and their friends’health-risk behavior: Factors that alter or add to peer influence. Journal of PediatricPsychology, 26, 287–298.

Prinstein, M. J., Meade, C. S., & Cohen, G. L. (2003). Adolescent oral sex, peer pop-ularity, and perceptions of best friends’ sexual behavior. Journal of PediatricPsychology, 28, 243–249.

Reitz, E., Dekovic, M., Meijer, A. M., & Engels, R. C. M. E. (2006). Longitudinal relationsamong parenting, best friends, and early adolescent problem behavior: Testingbidirectional effects. The Journal of Early Adolescence, 26, 272–295.

Revelle, W., & Zinbarg, R. E. (2009). Coefficients alpha, beta, omega, and the GLB:Comments on Sijtsma. Psychometrika, 74, 145–154.

Rose, A. J. (2002). Co-rumination in the friendships of girls and boys. Child Develop-ment, 73, 1830–1843.

Rose, A. J., & Rudolph, K. D. (2006). A review of sex differences in peer relationshipprocesses: Potential trade-offs for the emotional and behavioral developmentof girls and boys. Psychological Bulletin, 132, 98–131.

Rubin, K. H., Coplan, R. J., & Bowker, J. C. (2009). Social withdrawal in childhood.Annual Review of Psychology, 60, 141–171.

Rubin, K. H., Lynch, D., Coplan, R., Rose-Krasnor, L., & Booth, C. L. (1994). “Birds ofa feather. . .”: Behavioral concordances and preferential personal attraction inchildren. Child Development, 65, 1778–1785.

Rubin, K. H., Wojslawowicz, J. C., Rose-Krasnor, L., Booth-LaForce, C., & Burgess,K. B. (2006). The best friendships of shy/withdrawn children: Prevalence,stability, and relationship quality. Journal of Abnormal Child Psychology, 34,143–157.

Ruble, D. N., Taylor, L. J., Cyphers, L., Greulich, F. K., Lurye, L. E., & Shrout, P. E. (2007).The role of gender constancy in early gender development. Child Development,78, 1121–1136.

Rudolph, K. D., Flynn, M., Abaied, J. L., Groot, A., & Thompson, R. (2009). Why is pastdepression the best predictor of future depression? Stress generation as a mech-anism of depression continuity in girls. Journal of Clinical Child and AdolescentPsychology, 38, 473–485.

Rudolph, K. D., Hammen, C., Burge, D., Lindberg, N., Herzberg, D., & Daley, S. E.(2000). Toward an interpersonal life-stress model of depression: The devel-opmental context of stress generation. Development and Psychopathology, 12,215–234.

Schaefer, D. R., Kornienko, O., & Fox, A. M. (2011). Misery does not love company: Net-work selection mechanisms and depression homophily. American SociologicalReview, 76, 764–785.

Schwartz-Mette, R. A., & Rose, A. J. (2012). Co-rumination mediates contagion ofinternalizing symptoms within youths’ friendships. Developmental Psychology,48, 1355–1365.

Sijtsema, J. J., Lindenberg, S. M., & Veenstra, R. (2010). Do they get what they wantor are they stuck with what they can get? Testing homophily against defaultselection for friendships of highly aggressive boys. The TRAILS study. Journal ofAbnormal Child Psychology, 38, 803–813.

Stanger, C., & Lewis, M. (1993). Agreement among parents, teachers, and childrenon internalizing and externalizing behavior problems. Journal of Clinical ChildPsychology, 22, 107–115.

Stevens, E. A., & Prinstein, M. J. (2005). Peer contagion of depressogenic attribu-tional styles among adolescents: A longitudinal study. Journal of Abnormal ChildPsychology, 33, 25–37.

Stone, L.L., Otten, R., Ringlever, L., Hiemstra, M., Engels, R.C.M.E., Vermulst, A.A.,et al. The parent version of the Strengths and Difficulties Questionnaire:Omega as an alternative to alpha and a test for measurement invariance.European Journal of Psychological Assessment. Advance online publication,http://dx.doi.org/10.1027/1015-5759/a000119, in press

Tucker, L. R., & Lewis, C. (1973). The reliability coefficient for maximum likelihoodfactor analysis. Psychometrika, 38, 1–10.

Twenge, J. M., & Nolen-Hoeksema, S. (2002). Age, gender, race, socioeconomic status,and birth cohort difference on the Children’s Depression Inventory: A meta-analysis. Journal of Abnormal Psychology, 111, 578–588.

Valla, J. P. (2000). Instruction manual for the Dominic Interactive. In J. P. Valla (Ed.),The Dominic Interactive. Montreal, Canada: DIMAT.

Van Zalk, N., Van Zalk, M., Kerr, M., & Stattin, H. (2011). Social anxiety as a basis forfriendship selection and socialization in adolescents’ social networks. Journal ofPersonality, 79, 499–526.

Verhulst, F. C., Van der Ende, J., & Koot, H. M. (1997). Handleiding voor de Teacher’sReport Form (TRF) [Manual for the Teachers’ Report Form (TRF)]. Rotterdam, Neder-land: Afdeling Kinder- en Jeugdpsychiatrie, Sophia Kinderziekenhuis/ErasmusMC.

Vitaro, F., Tremblay, R. E., Kerr, M., Pagani, L., & Bukwoski, W. M. (1997). Disrup-tiveness, friends’ characteristics, and delinquency in early adolescence: A testof two competing models of development. Child Development, 68, 676–689.

Widenfelt, B. M., van Goedhart, A. W., Treffers, P. D. A., & Goodman, R. (2003). Dutchversion of the Strengths and Difficulties Questionnaire (SDQ). European Child

and Adolescent Psychiatry, 12, 281–289.

Zahn-Waxler, C., Klimes-Dougan, B., & Slattery, M. C. (2000). Internalizing problemsof childhood and adolescence: Prospects, pitfalls, and progress in understandingthe development of anxiety and depression. Development and Psychopathology,12, 443–466.