This article was downloaded by: [Fondren Library, Rice University ]On: 18 November 2014, At: 12:34Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Measurement in Physical Education andExercise SciencePublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/hmpe20

Validation of the Child Sport CohesionQuestionnaireLuc J. Martin a , Albert V. Carron a , Mark A. Eys b & Todd Loughead ca Western University London , Ontario , Canadab Wilfrid Laurier University Waterloo , Ontario , Canadac University of Windsor Windsor , Ontario , CanadaPublished online: 11 Mar 2013.

To cite this article: Luc J. Martin , Albert V. Carron , Mark A. Eys & Todd Loughead (2013) Validationof the Child Sport Cohesion Questionnaire, Measurement in Physical Education and Exercise Science,17:2, 105-119, DOI: 10.1080/1091367X.2013.761023

To link to this article: http://dx.doi.org/10.1080/1091367X.2013.761023

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Measurement in Physical Education and Exercise Science, 17: 105–119, 2013Copyright © Taylor & Francis Group, LLCISSN: 1091-367X print / 1532-7841 onlineDOI: 10.1080/1091367X.2013.761023

Validation of the Child Sport Cohesion Questionnaire

Luc J. Martin and Albert V. CarronWestern University

London, Ontario, Canada

Mark A. EysWilfrid Laurier UniversityWaterloo, Ontario, Canada

Todd LougheadUniversity of Windsor

Windsor, Ontario, Canada

The purpose of the present study was to test the validity evidence of the Child Sport CohesionQuestionnaire (CSCQ). To accomplish this task, convergent, discriminant, and known-group differ-ence validity were examined, along with factorial validity via confirmatory factor analysis (CFA).Child athletes (N = 290, Mage = 10.73 ± 1.13 years) from six elementary schools completed theCSCQ (Martin, Carron, Eys, & Loughead, 2012), a sport satisfaction questionnaire (Duda & Nicholls,1992), the Competitive State Anxiety Inventory—2 Children (Stadulis, MacCracken, Eidson, &Severance, 2002), and a Self-Efficacy Questionnaire—Soccer (Munroe-Chandler & Hall, 2004).Results indicated that cohesion was positively related to satisfaction and negatively related to anxietyin children, providing support for convergent validity. Additionally, as hypothesized, social cohe-sion was less correlated to self-efficacy than task cohesion, supporting discriminant validity. Finally,support for the factorial validity of the CSCQ was demonstrated through acceptable fit indices.

Key words: cohesiveness, group dynamics, children, measurement, validity

INTRODUCTION

Within the sport and exercise psychology domain, a great deal of research has focused on cohe-sion, which is defined as “a dynamic process that is reflected in the tendency for a group tostick together and remain united in the pursuit of its instrumental objectives and/or for the sat-isfaction of member affective needs” (Carron, Brawley, & Widmeyer, 1998, p. 213). Cohesionplays an important role in the dynamics of all groups, so much so that some social scientists havedescribed it as the most important small group variable (Golembiewski, 1962; Lott & Lott, 1965).

Correspondence should be sent to Luc J. Martin, Department of Kinesiology and Physical Education, University ofLethbridge, Lethbridge, AB, T1K 3M4, Canada. E-mail: luc.martin@uleth.ca

Dow

nloa

ded

by [

Fond

ren

Lib

rary

, Ric

e U

nive

rsity

] a

t 12:

34 1

8 N

ovem

ber

2014

106 MARTIN ET AL.

Results from research with adult populations examining the correlates of cohesion highlight thisimportance. For example, researchers have found cohesion to have a positive relationship withcollective efficacy (Kozub & McDonnell, 2000), athlete satisfaction (Widmeyer & Williams,1991), and adherence (Prapavessis & Carron, 1997), and a negative (i.e., beneficial) relation-ship with both state anxiety (Prapavessis & Carron, 1996) and depression (Terry, Carron, Pink,Lane, Jones, & Hall, 2000).

The examination of these cohesion correlates was facilitated by the development of the GroupEnvironment Questionnaire (GEQ; Carron, Widmeyer, & Brawley, 1985). However, the GEQwas originally developed to measure perceptions of cohesion with athletes between the ages of18 and 30. Due to the perceived importance of cohesion in sport and exercise groups, researchersdeveloped specific measures for different populations. In 2000, Estabrooks and Carron developedthe Physical Activity Group Environment Questionnaire (PAGEQ) for use in exercise and phys-ical activity classes containing older adults (60 years or greater). More recently, Eys, Loughead,Bray, and Carron (2009) developed the Youth Sport Environment Questionnaire (YSEQ) for ado-lescent sport populations (13–17 years). Finally, and of most relevance to the present study,Martin, Carron, Eys, and Loughead (2012) developed the Child Sport Cohesion Questionnaire(CSCQ)—an inventory used to assess cohesion in children’s (9–12 years) sport teams.

The CSCQ is a 16-item inventory measuring perceptions of cohesion on a 5-point Likert scale.Seven items measure task cohesion (i.e., the extent to which a team is united during competitionand collectively works toward the attainment of team goals), and seven items measure socialcohesion (i.e., the extent to which individuals on a team get along and stick together away fromthe sport). The remaining two items are negatively worded spurious items used to detect partici-pant response acquiescence. This newly created questionnaire demonstrated strong model fit withgood inter-factor correlations and internal consistency values (Martin et al., 2012). Although theseinitial results are promising, establishing construct validity is an ongoing process. Therefore,the purpose of the present study was to further examine the CSCQ for four manifestations ofvalidity—convergent, discriminant, known-group difference, and factorial validity.

Convergent validity is demonstrated when constructs that are theoretically related are in factshown to be related (e.g., Smith, Cumming, & Smoll, 2008; Trochim, 2006). Athlete satisfactionhas been found to be positively related to cohesion in adult populations (e.g., Aoyagi, Cox, &McGuire, 2008; Martens & Peterson, 1971; Spink, Nickel, Wilson, & Odnokon, 2005; Widmeyer& Williams, 1991). For example, Widmeyer and Williams (1991) found athlete satisfaction tobe highly correlated to perceptions of team cohesion in 85 NCAA Division 1 female golfers.Similarly, Spink et al. (2005) found a comparable relationship between the constructs of satis-faction and cohesion in a sample of 194 competitive male ice hockey players. Therefore, for thepresent study, it was hypothesized that children perceiving higher amounts of task and socialcohesion in their teams would also express greater amounts of satisfaction with their sportingexperience.

Another construct included to test convergent validity for the CSCQ was competitive stateanxiety. Prapavessis and Carron (1996) found that athletes on teams with higher levels of taskcohesion experienced lower levels of pre-competition state anxiety. Building on these findings,Eys, Hardy, Carron, and Beauchamp (2003) examined whether athletes perceived their compe-tition anxiety as facilitative or debilitative. They found athletes who perceived their cognitiveand somatic anxiety as being debilitative had lower levels of task cohesion. Therefore, consis-tent with this general pattern of results, it was hypothesized that individuals perceiving greater

Dow

nloa

ded

by [

Fond

ren

Lib

rary

, Ric

e U

nive

rsity

] a

t 12:

34 1

8 N

ovem

ber

2014

COHESION IN CHILD SPORT 107

levels of cohesion in their teams would experience lower levels of competitive state somatic andcognitive anxiety.

Discriminant validity is considered to be present when theoretically plausible differences aredemonstrated between constructs (e.g., Smith et al., 2008; Trochim, 2006). This type of valid-ity can be assessed both externally, through relationships with theoretically unrelated constructssuch as cohesion and IQ, and internally, through relationships between subscales—for example,between task and social cohesion (e.g., Cumming, Smith, Smoll, Standage, & Grossbard, 2008).Perceptions of cohesion in children have been found to have task and social orientations (Martin,Carron, Eys, & Loughead, 2011; Martin et al., 2012). Task cohesion is concerned with teamgoals and objectives, while social cohesion is concerned with friendships and affiliation (Carron& Eys, 2012). Self-efficacy is defined as an individual’s perceptions of his or her ability to per-form a task successfully (Bandura, 1977). Given the task focus of self-efficacy, its relationship totask and social cohesion could be expected to differ. Lent, Schmidt, and Schmidt (2006) found asmall relationship between self-efficacy and cohesion; however, this small, albeit significant (p <

.05), relationship is perhaps not surprising, considering the cohesion subscales were combined.Thus, it is predicted that task cohesion, which assesses a group’s closeness and unity towardscompleting a task or objective, should be more positively correlated with an individual’s level ofself-efficacy than social cohesion.

Known-group difference validity is demonstrated when populations that are hypothesized todiffer are in fact shown to have statistically significant and meaningful mean differences withrespect to the variable under examination (Rowe & Mahar, 2006). When Carron et al. (1985)validated the GEQ, and Heuzé and Fontayne (2002) validated their French-language cohesioninventory (Questionnaire sur l’Ambiance du Groupe), they suggested that perceptions of cohesionwould differ based on sport type and team tenure. Insofar as sport type was concerned, theseauthors predicted that cohesion would be stronger in participants from team (e.g., volleyball)versus individual (e.g., track and field) sports. With regard to team tenure, they predicted thatcohesion would be stronger in longstanding team members, compared to newly recruited teammembers. Therefore, we hypothesized that an individual’s type of sport participation, and theirlength of tenure on their team, could be differentiated based on their perceptions of both task andsocial cohesion.

Finally, confirmatory factor analysis (CFA) was used to test the factorial validity of the CSCQ.In an initial study with 298 child sport participants, Martin et al. (2012) demonstrated a strongmodel fit for the CSCQ. However, as Tabachnick and Fidell (2007) pointed out, “cross-validationwith another sample is performed whenever possible” (p. 682). Thus, it was hypothesized thatanalyses undertaken with the sample in the present study would again provide support for thefactorial validity of the CSCQ.

METHOD

Participants

A heterogeneous sample of 290 children (n = 131 males, n = 159 females) ranging in age from9 to 12 years (Mage = 10.73 ± 1.13) volunteered for the present study. The child sport participantscame from six elementary schools and represented a variety of different interdependent team

Dow

nloa

ded

by [

Fond

ren

Lib

rary

, Ric

e U

nive

rsity

] a

t 12:

34 1

8 N

ovem

ber

2014

108 MARTIN ET AL.

sports such as soccer (n = 83), hockey (n = 74), basketball (n = 27), American football (n = 16),baseball (n = 16), volleyball (n = 10), rugby (n = 8), ringette (n = 5), and cheerleading (n = 4),and individual sports such as track and field (n = 11), swimming (n = 9), gymnastics (n = 7),martial arts (n = 5), figure skating (n = 5), bowling (n = 3), dance (n = 3), tennis (n = 2), diving(n = 1), and equestrian (n = 1). Interdependent team sports (n = 243) required interaction amongteam members during play (e.g., volleyball, hockey), while individual sports (n = 47) were thoseperformed independently (e.g., track and field, swimming). Athletes also identified the length oftime that they had participated on their current team (i.e., team tenure). Of the 191 participantswho provided responses, 106 had been on a team for one year or less and 85 had been on a teamfor two years.

The reasoning behind the sample size chosen for the current study was based on two of thetypes of analyses undertaken. First, for Pearson-product moment correlations, Tabachnick andFidell (2007) suggested that “for variables in the social sciences where reliability is often around.80, about 10 cases are needed for every variable” (p. 570). The four questionnaires administeredin the study had a combined seven variables; therefore, based on these suggestions, a minimum of70 subjects were required. Second, there are no fixed prescriptions in sample sizes for CFA, justguidelines. Tabachnick and Fidell (2007) suggested it “is comforting to have at least 300 cases”but acknowledged that, “solutions that have several high loading marker variables (>.80) do notrequire such large sample sizes (about 150 cases should be sufficient)” (p. 613). Therefore, thesample size (N = 290) for the present study was deemed acceptable.

Measures

Cohesion

The 16-item CSCQ (Martin et al., 2012) was employed to assess cohesion. As indicated above,of the 16 questions, 14 relate to task (7 items) and social (7 items) cohesion and two are spuriousitems included to assess response acquiescence. Responses are obtained on a 5-point Likert-typescale anchored at the extremes by 1 (Strongly Disagree) and 5 (Strongly Agree). Thus, higherscores reflect stronger perceptions of cohesion.

Satisfaction

Participant satisfaction was measured using items generated by Duda and Nicholls (1992)to assess satisfaction in sport. These items fall within two sub-scales—satisfaction and bore-dom. For the present study, only the subscale containing five items targeting satisfaction (e.g.,“I usually find playing sport interesting”) was incorporated. Responses were obtained on a 5-point Likert-type response scale, anchored at the extremes with 1 (Strongly Disagree) and 5(Strongly Agree). Thus, higher scores reflected greater satisfaction. These items were originallyused with an adolescent population (Mage = 15.10 years), and demonstrated a Cronbach’s alphavalue of .94. Although previously used with an older sample, all but one of the items had Flesch-Kincaid readability grade levels between 2.4 to 7.6. Note that while one item was slightly higherthan a grade seven reading level, the potential for readability-produced measurement error wasconsidered low enough to maintain the item. Furthermore, a CFA was conducted—using the

Dow

nloa

ded

by [

Fond

ren

Lib

rary

, Ric

e U

nive

rsity

] a

t 12:

34 1

8 N

ovem

ber

2014

COHESION IN CHILD SPORT 109

statistical software package AMOS 18 (Arbuckle, 2009) with maximum likelihood estimation—with the present sample and demonstrated a non-significant chi-squared test χ2(5) = 3.027,p = .69, good fit indices (CFI = 1.00, TLI = 1.00, RMSEA = .00, SRMR = .01; for fur-ther information on fit indices and criterion values used in the present study, please refer tothe results section), and had factor loadings above .55 (i.e., “good” values; Comrey & Lee,1992).

Competitive State Anxiety

The Competitive State Anxiety Inventory—2 Children (CSAI-2C ; Stadulis, MacCracken,Eidson, & Severance, 2002) was administered in order to assess competitive state anxiety. Theoriginal inventory allows for the inclusion of words to indicate the desired activity; therefore,words relating to sport were inserted (e.g., “concerned that I may not play as well as I can today”).For the purpose of the present study, small adaptations to the CSAI-2C were implemented. First,the 15-item CSAI-2C contains three subscales: somatic anxiety, cognitive anxiety, and confi-dence. The five items pertaining to confidence were omitted, resulting in 10 items overall. Second,the CSAI-2C is based on a 4-point Likert-type scale. In the present study, a 5-point Likert-typeresponse scale anchored at the extremes with 1 (Strongly Disagree) and 5 (Strongly Agree) wasused in order to insure consistency of format throughout the questionnaire package. Higher scoresreflected greater levels of anxiety. In past research, the CSAI-2C has demonstrated good modelfit indices (e.g., Goodness-of-Fit-Index (GFI) = .959, Adjusted Goodness-of-Fit-Index (AGFI) =.943, SRMR = .042) and Cronbach’s alpha values (α = .78, somatic anxiety; α = 75, cognitiveanxiety) with a child population (N = 623) ranging in age from 8 to 12 years (Stadulis et al.,2002). A CFA with the present sample resulted in a significant chi-squared test χ2(34) = 115.715,p < .001 (this is a likely outcome with large sample sizes), an acceptable model fit (CFI = .93,TLI = .92, RMSEA = .09, SRMR = .05), factor loadings above .55 (Comrey & Lee, 1992), andan inter-factor correlation (between somatic and cognitive anxiety) of .80.

Self-efficacy

Self-efficacy was measured using the modified version of the Self-Efficacy Questionnaire—Soccer (SEQ-S; Munroe-Chandler & Hall, 2004) used by Hall, Munroe-Chandler, Fishburne,and Hall, (2009). The questionnaire is composed of five items (e.g., I am confident I canwork through difficult situations) and responses are obtained on a 0% to 100% rating scaledesigned to represent degree of efficacy. Again, however, in order to maintain consistency inthe format throughout the total questionnaire package, responses were obtained on a 5-pointLikert-type scale anchored at the extremes with 1 (Strongly Disagree) and 5 (Strongly Agree).Higher scores reflected greater perceptions of self-efficacy. This questionnaire has previouslybeen used with a similar child population (Mage = 11.53 years; Hall et al., 2009). A CFA demon-strated a non-significant chi-squared test χ2(5) = 10.067, p = .07, good model fit (CFI = .99,TLI = .98, RMSEA = .06, SRMR = .02), and factor loadings above .70 (i.e., “excellent”values; Comrey & Lee, 1992). Finally, it should be noted that all sub-scales within the mea-sures described above had internal consistency values greater than .80 in the present study (seeTable 1).

Dow

nloa

ded

by [

Fond

ren

Lib

rary

, Ric

e U

nive

rsity

] a

t 12:

34 1

8 N

ovem

ber

2014

110 MARTIN ET AL.

TABLE 1Descriptive Statistics (Mean, Standard Deviation, and Cronbach’s alpha) for

Study Variables

Factor M SD α

1. Task cohesion 4.06 .73 .902. Social cohesion 3.75 .80 .903. Satisfaction 4.41 .76 .894. Cognitive anxiety 2.39 .83 .805. Somatic anxiety 2.41 .91 .856. Self-efficacy 4.04 .82 .89

Note. Mean scores for all factors were measured on a 5-point Likert scale with higherscores reflecting higher perceptions of that particular construct. M = mean; SD = standarddeviation; α = Cronbach’s alpha coefficient.

Procedure and Analysis

Once ethical approval was obtained from the lead author’s non-medical research ethics board, theresearchers contacted the local school board’s research ethics committee for permission to enterelementary schools. Permission was obtained, and six elementary schools agreed to participatein the study. The lead researcher entered classrooms in order to provide a brief description of thestudy to the children and distribute parental and participant consent/assent forms. Once parentalconsent forms were returned to the teacher, the lead researcher returned to the school to distributequestionnaires to the eligible participants. The questionnaires were administered in the schoollibrary at the beginning of the lunch hour to insure that class time was not disrupted. Participantswere asked to relate the questions to their current sport team and to pretend they were gettingready to play a game or perform their sport. The questionnaires took approximately 15 minutesto complete.

Convergent and discriminant validity were assessed using Pearson’s product-moment correla-tions to determine the relationships between cohesion, satisfaction, competitive state anxiety, andself-efficacy. Known-group difference validity was assessed using discriminant function analysisto determine whether perceptions of cohesion (task and social) could discriminate between sporttype (individual and team sport) and/or team tenure (one year and two years). Factorial validitywas assessed by conducting a CFA on the CSCQ, using the statistical software package AMOS18 (Arbuckle, 2009) with maximum likelihood estimation.

RESULTS

Descriptive Statistics

Means, standard deviations, and Cronbach’s alpha values for the six sub-scales analyzed in thestudy are provided in Table 1. In general, participants indicated high levels of cohesion (task andsocial), satisfaction, and self-efficacy, and lower levels of competitive state anxiety. The internalconsistency values for all of the sub-scales were acceptable (i.e., >.70).

Dow

nloa

ded

by [

Fond

ren

Lib

rary

, Ric

e U

nive

rsity

] a

t 12:

34 1

8 N

ovem

ber

2014

COHESION IN CHILD SPORT 111

Normality assumptions and tests for outliers were conducted for all of the variables (i.e., cohe-sion, satisfaction, competitive state anxiety, and self-efficacy). Specifically, values of skewnessand kurtosis ranged between acceptable values of –2 and 2, and Tabachnick and Fidell (2007)suggest that, with larger sample sizes, it is beneficial to examine the shape of distribution usinghistograms, which in the current study appeared to be normally distributed (e.g., Field, 2009).Finally, Box’s M tests were non-significant (p > .05) and Mahalanobis distance was used in orderto probe for potential outliers with each variable. Based on a conservative criterion (p < .001),we identified three potential outliers whose chi-square tests were greater than χ2 = 22.46. Thethree cases were inspected, and it was found that their individual responses varied considerablythroughout the questionnaires. These three cases were subsequently removed from the analyses;however, their removal led to the identification of three new outliers. After repeating this process,it is suggested to run the analysis with and without the outliers to determine whether they are trulyinfluencing the results (Tabachnick & Fidell, 2007). Their removal did not improve the model fitfor the variables and the parameter estimates; therefore they were retained in the analysis.

Validity Analyses

Convergent Validity

The first test of convergent validity involved cohesion (task and social) and satisfaction.Convergent validity could be assumed to be present if task and social cohesion demonstratedmoderate relationships with satisfaction. The prescriptions advanced by Cohen (1988), in whicha correlation of .50 is considered “large,” .30 “medium,” and .10 “small,” were used to interpretthe magnitude of the relationships in the present study. The results indicated that both task andsocial cohesion had large significant positive (p < .001) correlations (r = .68 and .52, respec-tively) with satisfaction (see Table 2). Therefore, the hypothesis that cohesion and satisfactionwould be related was supported.

The second test for convergent validity involved an examination of the relationships amongtask and social cohesion and both somatic and cognitive anxiety. Convergent validity could beassumed if the two cohesion measures demonstrated negative correlations with the two com-petitive state anxiety measures. The findings from Table 2 indicated that task cohesion had large

TABLE 2Pearson Correlations between the Subscales from the CSCQ and the Sport

Satisfaction Questionnaire, CSAI-2C, and SEQ-S

Factor 1. Task Cohesion 2. Social Cohesion

1. Task cohesion — .53∗2. Social cohesion .53∗ —3. Satisfaction .68∗ .52∗4. Cognitive anxiety −.49∗ −.37∗5. Somatic anxiety −.49∗ −.32∗6. Self-efficacy .73∗ .46∗

Note. ∗p < .001

Dow

nloa

ded

by [

Fond

ren

Lib

rary

, Ric

e U

nive

rsity

] a

t 12:

34 1

8 N

ovem

ber

2014

112 MARTIN ET AL.

significant (p < .001) negative correlations with cognitive (r = –.49) and somatic (r = –.49) anxi-ety. Social cohesion had moderate significant (p < .001) negative correlations with both cognitive(r = –.36) and somatic (r = –.32) anxiety. Thus, our a priori hypothesis was supported.

Discriminant Validity

Discriminant validity was determined by assessing the difference between the relationshipsfor both types of cohesion (task and social) and self-efficacy. It was hypothesized that socialcohesion would have a weaker correlation with self-efficacy. The results (see Table 2) showed alarge significant (p < .001) correlation to be present between task cohesion and self-efficacy (r= .73) and a moderate to large significant (p < .001) correlation between social cohesion andself-efficacy (r = .46). A test of these correlations (Chen & Popovich, 2002) confirmed that theywere statistically different (t(287) = 8.00, p < .001). Therefore, the hypothesis was supported.

Known-group difference validity

The first test of known-group difference validity involved a direct discriminant function anal-ysis (DFA) with team tenure (first year players versus second year players) as the groupingvariable and task and social cohesion as the independent variables. The results from the DFAdemonstrated that individuals’ tenure was successfully discriminated based on perceptions ofcohesion (χ2 (2) = 9.05, p < .001). The squared canonical correlation (indicating the degreeof relationship between sport type and perceptions of cohesion) was .05 (p = .01). Based onthe group centroids for the first year players (–.19) and the second year players (.25), thediscriminant function was able to successfully separate these two groups (e.g., Tabachnick &Fidell, 2007). In addition, 82.2% of the original grouped cases were correctly classified. Finally,the structural coefficients and the standardized discriminant function coefficients (see Table 3)suggest that the best predictor for discriminating between team tenure is perceptions of socialcohesion.

The second test of known-group difference validity involved a direct DFA with sport type(team versus individual sport) as the grouping variable and task and social cohesion as the inde-pendent variables. It was hypothesized that sport type could be differentiated by perceptions ofboth task and social cohesion. Although the group centroid values for team (–.24) and individual(.05) sport did suggest a separation of the groups (e.g., Tabachnick & Fidell, 2007), perceptions

TABLE 3Cohesion Means (Standard Deviations), Discriminant Function Coefficients,

Structural Coefficients, and Group Centroids for Team Tenure

Tenure

Factor1 year

(n = 106)2 years

(n = 85)

DiscriminantFunction

CoefficientStructuralCoefficient

Task cohesion 3.98 (.83) 4.16 (.72) −.06 .49Social cohesion 3.55 (.92) 3.91 (.66) 1.03 .99

Note. Mean scores for cohesion were measured on a 5-point Likert scale (1 = low cohesionand 5 = high cohesion).

Dow

nloa

ded

by [

Fond

ren

Lib

rary

, Ric

e U

nive

rsity

] a

t 12:

34 1

8 N

ovem

ber

2014

COHESION IN CHILD SPORT 113

TABLE 4Cohesion Means (Standard Deviations), Discriminant Function Coefficients, and Structural

Coefficients for Sport Type

FactorTeam sport(n = 243)

Individualsport (n = 47)

DiscriminantFunction

CoefficientStructuralCoefficient

Task cohesion 4.09 (.72) 3.89 (.78) .92 .99Social cohesion 3.77 (.79) 3.64 (.85) .14 .62

Note. Mean scores for cohesion were measured on a 5-point Likert scale (1 = low cohesion and 5 =high cohesion).

of cohesion did not successfully discriminate between sport type (χ2(2) = 3.19, p = .25; seeTable 4). The squared canonical correlation was .01 (p = .20), and 60.2% of original groupedcases were correctly classified.

Factorial Validity

A CFA was conducted to test the hypothesis that the CSCQ possessed factorial validity.Figure 1 contains the inter-factor correlation and the standardized factor loadings. In addition,Table 5 provides the correlation matrix of the items for the CSCQ. A statistically significant (p <

.001) chi-squared test (χ2(76) = 174.531) was found. However, again, it is highly likely to obtaina significant chi-square result with large sample sizes. The scale for the range of the error termswas specified automatically by the statistical software package (e.g., Kline, 2011), and resultsindicated that the error terms were between .39 and .76 (see Figure 1). The Comparative FitIndex (CFI), Tucker-Lewis Index (TLI), Root Mean Square Error of Approximation (RMSEA),and Standardized Root Mean Square Residual (SRMR) were chosen to assess model fit. For theCFI and TLI, some argue that cut-off values for good model fit should be greater than .90 (e.g.,Bentler, 1990), while others call for a more conservative cut-off of .95 (e.g., Hu & Bentler, 1999).For the RMSEA, a value of .06 indicates good fit (Hu & Bentler, 1999), while values between.08 and .10 indicate mediocre fit, and values greater than .10 indicate poor fit (MacCallum,Browne, & Sugawara, 1996). Finally, for the SRMR, values should be below .08 (Browne &Cudeck, 1993). Based on these guidelines, our results indicated a good model fit (CFI = .96, TLI= .95, RMSEA = .07, and SRMR = .04). The internal consistency values were .90 for both taskand social dimensions, the composite reliability scores (Bacon, Sauer, & Young, 1995) were .93(task) and .92 (social), and the inter-factor correlation was moderate (φ = .59). Thus, the a priorihypothesis was supported.

DISCUSSION

The purpose of the study was to examine the construct validity of the CSCQ. To carry out thispurpose, four types of validity were tested—convergent, discriminant, known-group difference,and factorial. Two general issues associated with our findings warrant discussion.

The first involves the validity of the CSCQ. Insofar as convergent validity is concerned, wetested two relationships: cohesion-satisfaction and cohesion-pre-competition anxiety. Results

Dow

nloa

ded

by [

Fond

ren

Lib

rary

, Ric

e U

nive

rsity

] a

t 12:

34 1

8 N

ovem

ber

2014

114 MARTIN ET AL.

Task

Cohesion

T1

T2

T3

T4

T5

T6

T7

S1

S2

Social

Cohesion

S3

S4

S5

S6

S7

.45

.67

.62

.80

.87

.95

.75

.85

.75

.70

.61

.71

.76

.85

.89

.60

.39

.49

.32

.67

.52

.56

.48

.80

.65

.70

.53

.46

.59

FIGURE 1 Inter-factor correlation, standardized parameter estimates,and errors of the Child Sport Cohesion Model.

from both sets of analyses provided support for convergent validity. As indicated above, acohesion-satisfaction link has been established previously in adult (e.g., Aoyagi et al., 2008;Spink et al., 2005; Widmeyer & Williams, 1991) and adolescent (e.g., Paradis & Loughead,2011) populations. Therefore, the presence of this relationship in children is probably not sur-prising. However, it does indicate the potential importance of cohesion for children involvedin sport. More specifically, not only is team cohesion related to satisfaction in children, butcohesive environments are also likely to facilitate many of the reasons children have cited forjoining and maintaining membership in sport: to have fun, to improve their skills, and to developfriendships (Weiss & Petlichkoff, 1989). Our results also have practical implications for thisage group. By targeting and increasing the levels of task and social cohesion in children’s sportteams, coaches and practitioners could increase the likelihood that young athletes would be moresatisfied with their sport experience and therefore be more likely to continue participation.

Dow

nloa

ded

by [

Fond

ren

Lib

rary

, Ric

e U

nive

rsity

] a

t 12:

34 1

8 N

ovem

ber

2014

COHESION IN CHILD SPORT 115

TABLE 5Correlation Matrix of the Items in the CSCQ

Items Task 1 Task 2 Task 3 Task 4 Task 5 Task 6 Task 7 Social 1 Social 2 Social 3 Social 4 Social 5 Social 6

Task 2 .57Task 3 .49 .45Task 4 .56 .55 .58Task 5 .47 .45 .60 .62Task 6 .55 .49 .72 .66 .61Task 7 .55 .49 .64 .58 .58 .75Social 1 .35 .28 .36 .36 .40 .43 .31Social 2 .24 .20 .27 .30 .29 .36 .26 .51Social 3 .27 .26 .30 .32 .26 .33 .27 .60 .47Social 4 .33 .32 .30 .34 .27 .32 .27 .58 .57 .55Social 5 .35 .29 .41 .43 .46 .52 .38 .59 .58 .53 .53Social 6 .34 .27 .40 .41 .38 .41 .32 .58 .55 .57 .59 .63Social 7 .36 .33 .34 .41 .37 .40 .24 .57 .54 .55 .62 .63 .72

Note. All correlations are significant at the p < .01 level.

The inverse relationships found between cohesion (i.e., task and social) and pre-competitionanxiety (i.e., cognitive and somatic) are also consistent with results obtained with an adult sample(e.g., Prapavessis & Carron, 1996). The presence of these relationships in children has importantimplications. Research shows that anxiety can decrease enjoyment in children (e.g., Gould &Krane, 1992; Scanlan & Lewthwaite, 1986) and cause them to avoid organized sport (Passer,1988; Pierce, 1980). Children with high levels of competitive anxiety are concerned with thepossibility that others will evaluate their performance negatively (e.g., Brustad, 1988; Passer,1983). It is possible that anxiety may be reduced when cohesion is increased because membersfeel closer to their teammates and believe them to be more supportive (as opposed to threatening).In fact, it could be argued that a cohesive group shares many similarities with a “caring climate,”which has been defined as an environment that is “interpersonally inviting, safe, supportive, andable to provide the experience of being valued and respected” (Newton et al, 2007, p. 70), and isproposed to evoke less anxiety in children (Fry, 2010).

Support was obtained for the discriminant validity of the CSCQ. As was pointed out above,it was hypothesized that task cohesion would have a significantly greater relationship withself-efficacy than would social cohesion. This is an important finding for this population, as self-efficacy is considered to be one of the most consistent predictors of physical activity (Weinberg& Gould, 2011). Therefore, theoretically, by targeting task cohesion, and improving individualperceptions of self-efficacy, a coach could increase athlete participation and adherence rates.In fact, peer-support—which would invariably be present in a task cohesive group—is one ofthe sources for improved self-efficacy (Bandura, 1986, 1997). Similarly, this established relation-ship could have a reciprocal effect. As feelings of self-efficacy increase, so would the willingnessto become involved and to participate in the task aspects of the group, thereby contributing toincreased levels of group task cohesion.

A discriminant function analysis was conducted to determine known-group difference validity.In this instance, the analysis was unable to categorize team and individual sport athletes basedon their perceptions of cohesion (both task and social). Therefore, for this test, known-group

Dow

nloa

ded

by [

Fond

ren

Lib

rary

, Ric

e U

nive

rsity

] a

t 12:

34 1

8 N

ovem

ber

2014

116 MARTIN ET AL.

difference validity was not supported. This is an interesting finding, as research with olderpopulations (e.g., Brawley, Carron, & Widmeyer, 1987; Heuzé & Fontayne, 2002) founddifferences in perceptions of cohesion based on sport type. This serves to highlight the impor-tance of a cohesion inventory for children. Although this age group can identify cohesionand understands the benefits of a cohesive group, some of the implications and presenceof the phenomenon may differ compared to adolescent and adult populations. Additionalresearch with different samples may be needed to further assess cohesion with regard to sporttype.

With regard to team tenure, only partial support was present for our hypothesis. Specifically,social cohesion was significantly greater in athletes who had participated on their current team fortwo years versus first year participants. Task cohesion did not differ significantly between the twocategories of tenure. A potential reason for this finding relates to the age of the children involvedin this study (i.e., 9–12 years). Perhaps at this age, children are too young to have establishedroles as veterans and rookies on their teams. In fact, it is common in many sports for childrento change teams with each passing year. A possible avenue to better test known-group differencevalidity with this age group in the future may be to assess continued sport involvement (e.g.,Spink, Wilson, & Odnokon, 2010). That is, it can be hypothesized that athletes who perceivetheir teams as highly cohesive will be more likely to adhere and return to their sport team thefollowing season—perceptions of cohesion may discriminate between those who return to theirteam from those who do not.

Finally, our study has provided support for the factorial validity of the CSCQ. As previouslydiscussed, factorial validity also was supported in an earlier study with a different sample (Martinet al., 2012). The fit indices for the present study were as good (some identical) as those in theprevious study. A proposed model is suggested to be valid when: (a) items targeting a specificfactor have high factor loadings for that factor, and (b) the correlations between the factors arenot excessively high (Kline, 2011). All of the factor loadings (see Figure 1) with the exception oftwo task items (r = .67, .62) were equal to or above .70 (excellent; Comrey & Lee, 1992), and theinter-factor correlation (φ = .59) was well below the recommended .90 (Kline, 2011). Therefore,the CSCQ has demonstrated initial factorial validity with two independent samples.

The second general point that warrants discussion relates to future directions. Overall, thepresent study has provided support for the construct validity of the CSCQ. Thus, it can now beused with confidence to better understand the impact cohesion has on many different aspectsof child sport. For example, the present study shows that cohesion is correlated with a numberof important constructs—satisfaction, competitive state anxiety, and self-efficacy. Future researchcould examine the causal nature of these relationships. Also, as another example, one could arguethat cohesion and adherence are to some extent tautological (i.e., both reflect, to varying degrees,how well the group sticks together). Thus, causal relationships between task and social cohesionand adherence measures such as drop out behavior, absenteeism, and intention to return shouldbe examined (e.g., Estabrooks, 2000; Loughead, Colman, & Carron, 2001; Spink, 1995).

Finally, research in the area of child sport has consistently emphasized the importance ofsocial factors for children’s enjoyment, adherence, feelings of self-worth, and competence (e.g.,Page, Frey, Talbert, & Falk, 1992; Smith, 2007; Ullrich-French & Smith, 2009; Weiss & Smith,2002). The CSCQ enables researchers to quantify the degree to which children perceive the social(and task) bonds within their sport teams. Our study demonstrates that children feel both a taskand social unity in their teams; therefore, with this information, and a valid measurement tool,

Dow

nloa

ded

by [

Fond

ren

Lib

rary

, Ric

e U

nive

rsity

] a

t 12:

34 1

8 N

ovem

ber

2014

COHESION IN CHILD SPORT 117

researchers are presented with fertile grounds to continue to determine the positive influences thata cohesive environment can provide for participating children.

AUTHOR NOTE

The authors would like to acknowledge the Social Sciences and Humanities Research Council ofCanada for its support of this project. The authors would also like to thank Dr. Yan Liu for herassistance with the study.

REFERENCES

Aoyagi, M. W., Cox, R. H., & McGuire, R. T. (2008). Organizational citizenship behaviour in sport: Relationships withleadership, team cohesion, and athlete satisfaction. Journal of Applied Sport Psychology, 20, 2–41.

Arbuckle, J. L. (2009). Amos 18 user’s guide. Chicago, IL: SPSS Inc.Bacon, D. R., Sauer, P. L., & Young, M. (1995). Composite reliability in structural equation modeling. Educational and

Psychological Measurement, 55, 394–406.Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioural change. Psychological Review, 84, 191–215.Bandura, A. (1986). Social foundations of thought and actions: A social cognitive theory. Englewood Cliffs, NJ: Prentice

Hall.Bandura, A. (1997). Self-efficacy: The exercise of control. New York, NY: Freeman.Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238–246.Brawley, L. R., Carron, A. V., & Widmeyer, W. N. (1987). Assessing the cohesion of teams: Validity of the group

environment questionnaire. Journal of Sport Psychology, 9, 275–294.Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A. Bollen & J. S. Long (Eds.),

Testing structural equation models (pp. 136–162). Newbury Park, CA: Sage.Brustad, R. J. (1988). Affective outcomes in competitive youth sport: The influence of intrapersonal and socialization

factors. Journal of Sport & Exercise Psychology, 10, 307–321.Carron, A. V., Brawley, L. R., & Widmeyer, W. N. (1998). The measurement of cohesiveness in sport groups. In J.

L. Duda (Ed.), Advances in sport and exercise psychology measurement (pp. 213–226). Morgantown, WV: FitnessInformation Technology.

Carron, A. V., & Eys, M. A. (2012). Group dynamics in sport (4th ed.). Morgantown, WV: Fitness InformationTechnology, Inc.

Carron, A. V., Widmeyer, W.N., & Brawley, L. R. (1985). The development of an instrument to measure cohesion in sportteams: The Group Environment Questionnaire. Journal of Sport Psychology, 7, 244–266.

Chen, P. Y., & Popovich, P. M. (2002). Correlation: Parametric and non-parametric measures. Thousand Oaks, CA:Sage.

Cohen, J. (1988). Statistical power analysis for the behavioural sciences (2nd ed.). Hillsdale, NJ: Erlbaum.Comrey, A. L., & Lee, H. B. (1992). A first course in factor analysis (2nd ed.). Hillsdale, NJ: Erlbaum.Cumming, S. P., Smith, R. E., Smoll, F. L., Standage, M., & Grossbard, J. R. (2008). Development and validation of the

achievement goal scale for youth sports. Psychology of Sport and Exercise, 9, 686–703.Duda, J. L., & Nicholls, J. G. (1992). Dimensions of achievement motivation in schoolwork and sport. Journal of

Educational Psychology, 84, 290–299.Estabrooks, P. A. (2000). Sustaining exercise participation through group cohesion. Exercise and Sport Sciences Reviews,

28, 63–67.Estabrooks, P. A. & Carron, A. V. (2000). The Physical Activity Group Environment Questionnaire: An instrument for

the assessment of cohesion in exercise classes. Group Dynamics: Theory, Research and Practice, 4, 230–243.Eys, M. A., Hardy, J., Carron, A. V., & Beauchamp, M. R. (2003). The relationship between task cohesion and competitive

state anxiety. Journal of Sport & Exercise Psychology, 25, 66–76.Eys, M. A., Loughead, T. M., Bray, S. R., & Carron, A. V. (2009). Development of a cohesion questionnaire for youth:

The Youth Sport Environment Questionnaire. Journal of Sport & Exercise Psychology, 31, 390–408.

Dow

nloa

ded

by [

Fond

ren

Lib

rary

, Ric

e U

nive

rsity

] a

t 12:

34 1

8 N

ovem

ber

2014

118 MARTIN ET AL.

Field, A. (2009). Discovering statistics using SPSS. (3rd ed.). Thousand Oaks, CA: Sage.Fry, M. D. (2010). Creating a positive climate for young athletes from day 1. Journal of Sport Psychology in Action, 1,

33–41.Golembiewski, R. (1962). The small group. Chicago, IL: University of Chicago.Gould, D. & Krane, V. (1992). The arousal-athletic performance relationship: Current status and future directions. In T.

S. Horn (Ed.), Advances in sport psychology (pp. 119–142). Champaign, IL: Human Kinetics.Hall, C. R., Munroe-Chandler, K. J., Fishburne, G. J., & Hall, N. D. (2009). The sport imagery questionnaire for children

(SIQ-C). Measurement in Physical Education and Exercise Science, 13, 93–107.Heuzé, J. P., & Fontayne, P. (2002). Questionnaire sur l’Ambiance du Groupe: A French-language instrument for

measuring group cohesion. Journal of Sport and Exercise Psychology, 24, 42–67.Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus

new alternatives. Structural Equation Modeling, 6, 1–55.Kline, R. B. (2011). Principles and practice of structural equation modeling (3rd ed.). New York, NY: Guilford.Kozub, S. A., & McDonnell, J. F. (2000). Exploring the relationship between cohesion and collective efficacy in rugby

teams. Journal of Sport Behavior, 23, 120–129.Lent, R. W., Schmidt, J., & Schmidt, J. (2006). Collective efficacy in student work teams: Relation to self-efficacy,

cohesion, and performance. Journal of Vocational Behavior, 68, 73–84.Lott, A. J., & Lott, B. E. (1965). Group cohesiveness as interpersonal attraction: A review of relationships with antecedent

and consequent variables. Psychological Bulletin, 64, 259–309.Loughead, T. M., Colman, M. M., & Carron, A. V. (2001). Investigating the meditational relationship of leadership, class

cohesion, and adherence in an exercise setting. Small Group Research, 32, 558–575.MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power analysis and determination of sample size for

covariance structure modeling. Psychological Methods, 1, 130–149.Martens, R., & Peterson, J. A. (1971). Group cohesiveness as a determinant of success and member satisfaction in team

performance. International Review for the Sociology of Sport, 6, 49–61.Martin, L. J., Carron, A. V., Eys, M. A., & Loughead, T. M. (2011). Children’s perceptions of cohesion. The Sport and

Exercise Psychology Review.7, 11–25.Martin, L. J., Carron, A. V., Eys, M. A., & Loughead, T. M. (2012). Development of a cohesion inventory for children’s

sport teams. Group Dynamics: Theory, Research, & Practice, 16, 68–79.Munroe-Chandler, K. J., & Hall, C. R. (2004). Enhancing the collective efficacy of a soccer team through motivational

general-mastery imagery. Imagination, Cognition and Personality, 24, 51–67.Newton, M., Fry, M., Watson, D., Gano-Overway, L., Kim, M., Magyar, M., & Guivernau, M. (2007). Psychometric

properties of the caring climate scale in a physical activity setting. Revista de Psiocologia del Deporte, 16, 67–84.Page, R. M., Frey, J., Talbert, R., & Falk, C. (1992). Children’s feelings of loneliness and social dissatisfaction:

Relationship to measures of physical fitness and activity. Journal of Teaching in Physical Education, 11, 211–219.Paradis, K. F., & Loughead, T. M. (2011). Examining the mediating role of cohesion between athlete leadership and

athlete satisfaction in youth sport. Manuscript submitted for publication.Passer, M. W. (1983). Fear of failure, fear of evaluation, perceived competence and self-esteem in competitive-trait

anxious children. Journal of Sport Psychology, 5, 172–188.Passer, M. W. (1988). Determinants and consequences of children’s competitive stress. In F. L. Smoll, R. A. Magill, &

M. J. Ash (Eds.), Children in sport (3rd ed., (pp. 203–227). Champaign, IL:Human Kinetics.Pierce, W. J. (1980). Psychological perspectives of youth sport participants and non-participants. Unpublished doctoral

dissertation. Virginia Polytechnic Institute and State University, Blacksburg.Prapavessis, H., & Carron, A. V. (1996). The effect of group cohesion on competitive state anxiety. Journal of Sport &

Exercise Psychology, 18, 64–74.Prapavessis, H., & Carron, A. V. (1997). Sacrifice, cohesion, and conformity to norms in sport teams. Group Dynamics:

Theory, Research, and Practice, 1, 231–240.Rowe, D. A., & Mahar, M. T. (2006). Validity. In T. M. Wood & W. Zhu (Eds.), Measurement theory and practice in

kinesiology (pp. 9–25). Champaign, IL: Human Kinetics.Scanlan, T. K., & Lewthwaite, R. (1986). Social psychological aspects of competition for male youth sport participants:

IV. Predictors of enjoyment. Journal of Sport Psychology, 8, 325–335.Smith, A. L. (2007). Youth peer relationships in sport. In S. Jowett & D. Lavallee (Eds.), Social psychology in sport (pp.

41–54). Champaign, IL: Human Kinetics.

Dow

nloa

ded

by [

Fond

ren

Lib

rary

, Ric

e U

nive

rsity

] a

t 12:

34 1

8 N

ovem

ber

2014

COHESION IN CHILD SPORT 119

Smith, R. E., Cumming, S. P., & Smoll, F. L. (2008). Development and validation of the motivational climate scale foryouth sports. Journal of Applied Sport Pscyhology, 20, 116–135.

Spink, K. S. (1995). Cohesion and intention to participate of female sport team athletes. Journal of Sport & ExercisePsychology, 17, 416–427.

Spink, K. S., Nickel, D., Wilson, K., & Odnokon, P. (2005). Using a multilevel approach to examine the relationshipbetween task cohesion and team task satisfaction in elite ice hockey players. Small Group Research, 36, 539–554.

Spink, K. S., Wilson, K. S., & Odnokon, P. (2010). Examining the relationship between cohesion and return to team inelite athletes. Psychology of Sport and Exercise, 11, 6–11.

Stadulis, R. E., MacCracken, M. J., Eidson, T. A., & Severance, C. (2002). A children’s form of the competitive stateanxiety inventory: The CSAI-2C. Measurement in Physical Education and Exercise Science, 6, 147–165.

Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston, MA: Allyn & Bacon.Terry, P. C., Carron, A. V., Pink, M. J., Lane, A. M., Jones, G. J. W., & Hall, M. P. (2000). Perceptions of group cohesion

and mood in sport teams. Group Dynamics, 4, 244–253.Trochim, W. (2006). Convergent and discriminant validity. Retrieved November 20, 2011, from http://www.

socialresearchmethods.net/kb/convdisc.phpUllrich-French, S., & Smith, A. L. (2009). Social and motivational predictors of continued youth sport participation.

Psychology of Sport and Exercise, 10, 87–95.Weinberg, R. S., & Gould, D. (2011). Children and sport psychology. In R. S. Weinberg & L. Petlichkoff (eds),

Foundations of Sport and Exercise Psychology (5th ed.), (pp. 491–510). Champaign, IL: Human Kinetics.Weiss, M. R., & Petlichkoff, L. (1989). Children’s motivation for participation in and withdrawal from sport: Identifying

missing links. Pediatric Exercise Sciences, 1, 195–211.Weiss, M. R., & Smith, A. L. (2002). Friendship quality in youth sport: relationship to age, gender, and motivation

variables. Journal of Sport & Exercise Psychology, 24, 420–437.Widmeyer, W. N., & Williams, J. M. (1991). Predicting cohesion in a coacting sport. Small Group Research, 22, 548–570.

Dow

nloa

ded

by [

Fond

ren

Lib

rary

, Ric

e U

nive

rsity

] a

t 12:

34 1

8 N

ovem

ber

2014