Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Perceptual similarity and member functioning in exercise groups Dunlop, William 2009

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata


ubc_2009_fall_dunlop_william.pdf [ 2.13MB ]
JSON: 1.0070963.json
JSON-LD: 1.0070963+ld.json
RDF/XML (Pretty): 1.0070963.xml
RDF/JSON: 1.0070963+rdf.json
Turtle: 1.0070963+rdf-turtle.txt
N-Triples: 1.0070963+rdf-ntriples.txt
Original Record: 1.0070963 +original-record.json
Full Text

Full Text

Perceptual Similarity and Member Functioning in Exercise GroupsbyWilliam DunlopBA., The University of Western Ontario, 2007A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF ARTSinThe Faculty of Graduate Studies(Human Kinetics)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)August 2009© William Dunlop, 2009ABSTRACTThis study explored the relationships between perceived intra-group similarity, cohesion,and adherence among exercise group members. Participants (N = 402) from 46 registeredhealth and wellness courses in a large city in Western Canada completed a questionnaireassessing their perceived level of similarity with the other group members and cohesionthree times during the first eight weeks of their course. Data were collected following thesecond, fifth, and eighth classes (this coincided with the second, fifth, and eighth week ofeach course). Participants’ initial perception of the proportion of group members thatwere similar to themselves was found to significantly (and positively) predict programadherence. In contrast, early measures of class cohesion did not predict programadherence. A secondary aim of this study was to apply a theoretical framework developedwithin the domain of organizational psychology to understand some of the contextualdeterminants of cohesion in group-based exercise programs. This framework had beenused to explain the emergence of cohesion within work groups through the considerationof (a) similarity among members’ surface- and deep-level attributes, and (b) the relativestage of group development (i.e., combined amount of time the group has spent together).Dimensions of task and social cohesion were predicted by both surface- and deep-levelsimilarity perceptions. Findings are discussed in relation to theory development,measurement, and the application of group dynamics principles to behavioral medicineresearch.11TABLE OF CONTENTSAbstract iiTable of Contents iiiList of Tables viList of Figures viiAcknowledgements viiiDedication ix1 Introduction 11.1 Perspectives on Intra-group Similarity 31.2 Group-based Exercise Programs 71.3 Intra-Group Similarity and Adherence 91.4 Cohesion and Exercise Groups 91.5 Surface-level and Deep-level Similarity in Groups 131.6 Similarity, Attraction, and Cohesion 151.7 Assumptions Present in Surface-Level Homogeneous and Heterogeneous Groups. 181.8 Exercise and Intra-Group Similarity Studies 211.9 Can Exercise Groups really be Considered ‘Groups?’ 241.10 How is Intra-Group Similarity Measured within ExerciseGroups’?262Method 312.1 Participant Recruitment 312.2 Participants 332.3 Measures 352.4 Procedure 401113 Results.433.1 Response Rate 433.2 Analysis 443.3 Individual or Group-level Analysis9 543.4 Consideration of Group Size 553.5 Global Similarity and Adherence 563.6 Surface- and Deep-Level Perceptual Similarity and Cohesion 593.7 Cohesion and Adherence 643.8 Test of Mediation 663.9 Assumption of Deep-level Similarity and Cohesion 674Discussion 714.1 Hypothesis 1 714.2 Hypotheses 2 and 3 734.3 Hypothesis 4 764.4 Hypothesis 5 764.5 Hypotheses 6 and 7 774.6 General Discussion 784.6.1 Is Intra-Group Similarity Beneficial to Member Functioning in Exercise Groups? 784.6.2 The Importance of Considering Subjective Intra-Group Similarity 804.6.3 Perceptual Intra-Group Similarity, Cohesion, and Adherence 824.6.4 Limitations 844.6.5 Summary & Conclusion 87References 89ivAppendices.108Appendix A 108Appendix B 109AppendixC 110AppendixD 111Appendix E 112Appendix F 113AppendixG 114Appendix H 116VLIST OF TABLESTable 1 Descriptive Statistics 34Table 2 Intercorrelations Among SLS and DLS Perceptions (Time 1) 49Table 3 Intercorrelations Among SLS and DLS Perceptions (Time 2) 50Table 4 Intercorrelations Among Global Measures of Perceptual Similarityand Adherence 58Table 5 Intercorrelations Among SLS and DLS Perceptions and Cohesion (Time 1) 61Table 6 Intercorrelations Among SLS and DLS Perceptions and Cohesion (Time 2) 63Table 7 Intercorrelations Among Cohesion and Adherence 65Table 8 Intercorrelations Among SLS and DLS Perceptions (Time 1) 68Table 9 Intercorrelations Among DES Factor Change Score and Cohesion 70viLIST OF FIGURESFigure 1 A Conceptual Model of Group Cohesion (Adapted from Carron et al., 1985) 11Figure 2 Representation of the Hypothesized Mediated Relationship Between Similarity,Cohesion, and Adherence 17viiACKNOWLEDGEMENTSI would like to extend my sincere gratitude to my advisor, Dr. Mark Beauchampfor his exceptional mentoring and guidance. Over the last two years Mark has taught me agreat deal about the business of academia, the notion of personal responsibility andcommitment, and the importance of striking a healthy balance between these twodomains. For this I am eternally grateful.I would also like to thank Dr. Brian Wilson for his dispositional open door policy.Although I was fortunate enough to take a class taught by Brian, I contend that I learnedmuch more from him outside of this class. The many conversations Brian and I have hadover the last two years have been extremely rewarding. Thanks for always listeningBrian...Special thanks go out to my mother, father, and brother. You three haveinfluenced my personal development tremendously. Who knows where I would bewithout you. Technically speaking, I guess I would not even ‘be’ without you. Thank youfor all you have done for me.Finally, I would like to thank my fellow graduate students at UBC. Though themeans that we adopt vary greatly, we are united in our passion for the pursuit ofunderstanding life in all its many facets. I hope this passion never withers.viiiDEDICATIONTo my father, whose passion for knowledge and critical reflection is eclipsed onlyby his ability to be present in, and enjoy, the moment no matter how simple or routine.ix1 INTRODUCTION“Never doubt that a small group of thoughtful, committed citizens can change the world.Indeed, it is the only thing that ever has.”Margaret Mead (1901-1978)Groups, and group functioning, have a pervasive influence on how we live ourlives. As a result, much work within social (Alcock, Carmens, & Sadava, 1998; Kenrick,Neuberg, & Cialdini, 2005, organizational (Robbins & Langton, 1999), sport, andexercise (Canon, Hausenblas, & Eys, 2005; Harwood & Beauchamp, 2007) psychologyhas explored the many variables that influence group functioning. To a large degree, thisexploration has been fostered by the continued application of concepts and theoriesacross these sub-disciplines. For example, through the study of the performance ofprofessional cyclists (a topic that could be considered to fall under the discipline of sportand exercise psychology), Norman Triplet (1897) identified the phenomenon now knownas social facilitation. In turn, this phenomenon has been researched heavily by social(e.g., Guerin, 1993; Spence, 1956; Zajonc, 1965) and organizational (e.g., Bond & Titus,1983; Robbins & Langton, 1999) psychologists.Within social, organizational, and sport psychology, a considerable amount ofresearch attention has been focused on the study of group composition. Groupcomposition is defined as “the relationships among the characteristics of individuals whocompose the group” (Shaw, 1981,p.454). The attention this concept has received withinthese disciplines is likely a result of the fact that group composition has been linked toseveral important outcome measures, most notably performance (Knippenberg &Schippers, 2007; Milliken & Martins, 1996).1Researchers who study group composition are often concerned with identifyingthe combination of group members that will likely result in optimal group and memberfunctioning. Consequently, the primary focus of researchers in this area has centered onthe study of diversity within groups and teams. Intra-group diversity is defined as “acharacteristic of the social grouping (i.e., group organization, society) that reflects thedegree to which there are objective and subjective differences between people within thegroup” (Knippenberg & Schippers, 2007,p.519). Consistent with this definition, intragroup diversity and similarity are understood to be diametrically opposed, falling onopposite ends of the “similarity-difference continuum” (Harrison, Price, Gavin, & Florey,2002,p.906). That is, similarity is conceptualized as being diametrically opposite todiversity (higher levels of one correspond to lower levels of the other).Intra-group similarity has been found to relate to many variables relevant to groupfunctioning including (but not limited to) cohesion (e.g., Back, 1951; Shaw & Shaw,1962), creativity (e.g., Dose, 1999) member satisfaction (e.g., Tsui, Egan, & O’Reilly,1992), social integration (e.g., Tsui, Egan, & Xin, 1995), communication (e.g., Lincoln &Miller, 1979), commitment (e.g., Knippenberg, Haslam, & Platlow, 2007), and thequality of interpersonal relations, (e.g., Milliken & Martins, 1996; Knippenberg &Schippers, 2007; Triandis, Kurowski, & Gelfand, 1994). Given this seemingly pervasiveinfluence, as well as the tendency for researchers within social, organizational, and sportand exercise psychology to incorporate concepts and theories across sub-disciplines, it issurprising to note that intra-group similarity has received scant attention within thedomain of exercise psychology. Indeed, only one known study (Shapcott, Carron, Burke,Bradshaw, & Estabrooks, 2006) has considered the importance of this topic within2exercise groups. Thus, the overall purpose undertaken in this thesis was to address thisgap in the research literature by exploring the relationship between intra-group similarityand member functioning in exercise groups.1.1 Perspectives on Intra-group SimilarityEssentially, two perspectives on intra-group similarity exist: the first purports thatintra-group similarity is detrimental to group functioning, the second that it is beneficial.This first perspective on intra-group similarity is based largely on the beliefs that (a)similar or homogeneous groups have less varied resources on which to draw upon and (b)greater and more varied resources enhance group processes (e.g., group creativity; Dose,1999). The majority of empirical work providing support for this perspective hasexplored the effect that varied backgrounds, training, and perspectives can have on workgroup performance and communication (e.g., Cox, 1993; Cox & Blake, 1991). Greaterdiversity in these factors is thought to result in an expanded pool of information (andknowledge) available to group members (Milliken & Martins, 1996). This perspective hasbeen fostered under the moniker of the ‘information processing model’ (Knippenberg, DeDreu, & Homan, 2004; Mannix & Neale, 2005). While recognizing that this model wasconceived with access to diverse information in mind it has also been theorized thatgreater diversity with respect to many other, non-informational attributes/variables mayalso benefit a group’s task and social functioning (see Carron et al., 2005). Although thename of this model implies a restricted focus on information or knowledge diversity, forthe purpose of the present discussion, the ‘information processing model’ will be used torefer to the perspective that intra-group similarity (be it informational or noninformational in nature) is detrimental to group functioning.3The second perspective on intra-group similarity views within-group homogeneityas beneficial to group functioning. This perspective is largely informed by Byrne’s (1971)Similarity-Attraction Hypothesis and Turner’s (1984; 1985; 1987) Self CategorizationTheory. Simply put, the similarity-attraction hypothesis predicts that one will be attractedto similar others and repelled (see Chen & Kenrick, 2002) by those that are dissimilar.This hypothesis has been supported by a great deal of anecdotal and empirical evidence(Montoya, Horton, & Kirchner, 2008). For example, attraction has been correlated withmany different types of similarities including attitudinal (e.g., Byrne, Baskett, & Hodges,1971; Tan & Singh, 1995), personality (e.g., Banikotes & Neimeyer, 1981; Bleda, 1974),and physical (e.g., Peterson & Miller, 1980; Stevens, Owens, & Schaefer, 1990)similarity.Byrne (1971) proposed that the relationship between similarity and attractionoccurs as a result of each individual’s desire to have their own attitudes, beliefs, andworld views validated. According to Byrne, this desire stems from a fundamental needfor a logical and consistent view of the world. By being attracted to, and associated with,similar others, one increases the likelihood that this need will be satisfied. This is becausesimilar others are more likely to share, and thus validate, one’s world views. In thismanner, interactions with similar others are believed to be positively reinforcing (Byrne,1971). Indeed, as Struass, Barrick, and Connerly (2001) state, “similar attitudes, forexample, are perceived to be rewarding and are therefore viewed in the model as positivereinforcements, whereas dissimilar attitudes function as negative reinforcements” (p.638).4The proposed relationship between similarity and attraction can be positionedwithin the larger conceptual framework of self-categorization theory (Turner, 1984; 1985;1987). This theory, created to explain how and when individuals place themselves andothers into social categories, purports that people are more likely to adopt membership ofa group if they perceive congruencies in salient qualities between themselves and theother members (e.g., members may be perceived to be similar in age). Thus, a person ismore likely to be drawn to, and adopt membership in, a group in which they believe theyshare relevant attributes with the other members.Self-categorization theory contends that “our self-concept is based on the socialcategories we place ourselves in (e.g., age, gender, race)” (Strauss et al., 2001,p.638). Itfollows that membership of a group in which one is dissimilar to others will not alignwith that individual’s sense of self-identity. As Riordan and Shore (1997) state, “to theextent that self-identity is important to a person, the lack of continuity in self-identity dueto employment in a. . . group may prevent the individual from positively evaluatingthe. . . group and feeling a great deal of support and commitment toward the group” (p.344). This is because individuals have a disposition to evaluate the categories that they donot occupy (i.e., categories inconsistent with one’s self-identity) negatively and evaluatethe categories that they do occupy positively (Kramer, 1991; Tajfel, 1981; Strauss et al.,2001). Due to these evaluative processes, people are drawn to others, and reinforced bythose, who occupy the same categorizations as themselves and deterred from others whodo not fit in these categories. For the purposes of the present discussion, the‘categorization model’ will be used to refer to the perspective that individuals are5attracted to similar others and, as a result, intra-group similarity is beneficial to groupfunctioning.It would be a gross over-simplification to claim that either of these two competingperspectives can explain all processes and functions that occur within groups. Indeed, asEly (2004) states, “the link between diversity.., and the group’s performance... is neithersimple nor direct” (p. 755). However, in large part, intra-group similarity has been foundto have a positive effect on “the psychological relationship between the individual and thegroup (i.e., identification, commitment, cohesion) and affective/evaluative responses tothe group” (Knippenberg et al., 2007,p.207). Among other outcome variables, moresimilar groups have been found to report higher levels of group attachment (Tsui et al.,1992), communication (Riordan & Shore, 1997), and cooperation (Turner 1982; 1984),better interpersonal relations (Triandis et al., 1994), less turnover and absences (Jacksonet al., 1991; Tsui et al., 1992) and less conflict (Pelled, 1996) than comparably diversegroups. As a result, Harrison, Price, and Bell (1998) identified the perspective that intragroup similarity has a positive impact on functioning as “the primary thesis” amongdiversity researchers(p.96). Although homogeneous groups may reap theaforementioned benefits, groups with a low degree of similarity may actually outperformmore homogeneous groups when tasks require creative solutions or the consideration ofmultiple perspectives (i.e., perspective taking; Dose, 1999).From the research reviewed thus far, more heterogeneous groups seem to hold thepotential to outperform more homogeneous groups on certain tasks, such as those thatrequire creativity or perspective taking. However, on tasks that do not require a high levelof creativity or perspective taking, homogeneous groups will likely function better than6more diverse groups. This is due to the fact that homogeneous groups are likely to reporta higher level of cohesion, communication, member satisfaction, and commitment thancomparably heterogeneous groups (Tsui et a!., 1992). This conclusion provides supportfor the information processing model as well as the categorization model. However, thequestion remains, which processes are more applicable to exercise groups?Although heterogeneous groups may have the potential to outperform morehomogeneous groups on some tasks, such as those requiring creativity (Dose, 1999),these benefits (e.g., creativity) are likely limited in their applicability to exercise groups.Instead, processes such as cohesion, member satisfaction, and adherence derived fromgreater within-group homogeneity seem much more applicable to groups of this kind(Castellani, lanni, Ricca, Mannucci, & Rotella, 2003; Estabrooks, 2007). Since themajority of evidence from social and organizational psychology suggests that groupscomposed of similar members tend to report improved interpersonal relationships whencompared to members in more diverse groups, the central thesis proposed in this projectwas that intra-group similarity would be beneficial to the functioning of exercise groupmembers. In this thesis, member functioning was operationalized based on (a) each groupmember’s level of adherence and (b) the level of cohesion reported by each member.1.2 Group-based Exercise ProgramsDishman (1988) observed that within six months of enrolling in an exerciseprogram one in two individuals will typically withdraw from the program and drop out.As a result, maintaining participant adherence throughout the duration of a programrepresents one of the greatest challenges faced by applied health researchers andcoordinators (Castellani et al., 2003). In an attempt to understand and enhance adherence7to exercise programs many different approaches have been adopted (e.g., Carels et. al,2008; Marcus et. a!, 2007; Napolitano et. al, 2008; Booth, Nowson, & Matters, 2008;Hong, Hughes, & Prohaska, 2008). However, from the perspective of the researchdesigns inherent in these programs, the vast majority of programs can be classified aseither individual-, or group-based in nature.Individual-level exercise programs are most often designed to enable participantsto exercise within a personally convenient location, such as the home (e.g., King et al.,2008). Although there is some variability in the types of individual-based programsoffered, by definition these programs are undertaken without membership in a formalizedexercise group. In contrast, group-based programs typically consist of a collection ofparticipants within communal exercise environments that require them to either (a)perform activities together (e.g., Annesi, 1999; Carron & Spink, 1993; Estabrooks &Carron, 1 999a,b), and/or (b) work independently towards the pursuit of a collective goal(e.g., Shapcott et al., 2006).On average, group-based programs are more cost effective than individual-levelinterventions (Estabrooks, 2007). In addition, group-based approaches offer the addedpotential of providing a beneficial social experience (Carron et al., 2005), seem to bepreferred by the majority of individuals (Beauchamp, Carron, McCutcheon, & Harper,2007; Burke, Carron, & Eys, 2006; Heinzelmann & Bagley, 1970; Stephens & Craig,1990) and, in several prominent meta-analyses have been found to be more effective insupporting exercise adherence than individual-based programs (i.e., Carron, Hausenbias,& Mack, 1996; Dishman & Buckworth, 1996).81.3 Intra-Group Similarity and AdherenceA large body of research within social and organizational psychology reports thatwhen individuals are acutely similar to the other members of the group, they are morelikely to remain a part of that group than when they differ markedly from other groupmembers (e.g., Jackson et al., 1991; Milliken & Martins, 1996; Tsui et al., 1992). As anexample, Jackson et a!., (1991) reported that members of executive management teamswho were similar to their teammates in terms of age, education level attained, collegecurriculum, and experience outside the industry were more likely to maintain theirmembership in their work group than those who were dissimilar on the aforementionedvariables.The finding that people are more likely to remain a part of a group when they aresimilar to the other group members accords with the categorization perspective of intragroup similarity (i.e., similarity is beneficial), and is also consistent with the “primarythesis” (Harrison et al., 1998,p.96) purported by diversity researchers. In accordancewith this evidence, the principal hypothesis proposed in this thesis was that members whoperceived that they are similar to the other members of their group would be more likelyto adhere to the group-based exercise program than those who perceived a comparablylow level of similarity (Hypothesis 1).1.4 Cohesion and Exercise GroupsWithin the exercise psychology literature a considerable amount of researchattention has focused on the role of exercise class cohesion. Cohesion is defined as “adynamic process reflected by the tendency of a group to stick together and remain unitedin the pursuit of its instrumental objectives and/or for the satisfaction of member affective9needs” (Carron, Widmeyer, & Brawley, 1988,P.213). This construct is thought topossess four main characteristics (Carron et al., 2005). First, cohesion ismultidimensional in nature as there are multiple factors that may lead a group to remaintogether and these factors may not necessarily be the same in every group. Second,cohesion is thought to be dynamic and can change as the group develops. Third, cohesionis instrumental in nature insofar as it is related to the reasons for the group’s initialformation. Finally, cohesion also has an affective component. This is because thesatiation of member’s affective needs is thought to influence the likelihood that the groupwill remain together (Carron, Shapcott, & Burke, 2007).Consistent with these characteristics, the conceptual model of cohesion advancedby Carron et al. (1985) includes a task and social foci, as well as individual and grouporientations, thus resulting in a four dimension model. The task focus represents themotivation or desire to achieve the group’s instrumental objectives. The social focusrepresents the motivation to build and maintain social relationships and activities withinthe group and among group members. An individual orientation is represented by anindividual’s attractions to the group. These attractions represent the personal motivationsand feelings about the group that act to attract and retain the member. Finally, the grouporientation is represented by members’ perceptions of group integration (Carron et al.,2007).Four conceptually distinct dimensions of cohesion result when the task and socialfoci are combined with the attraction and group orientations (see Figure 1). Theindividual attractions to the group — task dimension (ATG-T) represents each member’sperception of his or her desire to be involved with the group’s task. The individual10attractions to the group — social dimension (ATG-S) represents a member’s perception ofthe level of social interaction, as well as the degree of social acceptance, they experiencein the group. Both group integration dimensions reflect perceptions regarding the degreeof unanimity or harmony within the group as a whole. However, the group integration —task dimension (GI-T) conceptualizes this perception around the group’s collective tasks,whereas the group-integration — social (GI-S) dimension does so around social concerns.Figure 1 A conceptual model of group cohesion (Adapted from Carron et a!., 1985)Individual attractions Individual Attractionto group - Task to group — SocialDimensions ofcohesion in sport andexercise groupsGroup integration - -_________________- Group integration —Task SocialIn his recent review of group integration interventions in exercise, Estabrooks(2007) identified cohesion as a “fundamental consideration in physical activityinterventions”(p.143). The significance attached to group cohesion likely stems from thepositive relationship that has consistently been found between cohesion and adherencebehaviours in exercise groups (e.g., Carron & Spink, 1993; Estabrooks & Carron,1999a,b). For example, Spink and Carron (1994) found that the ATG-T, GI-T, and GI-Sdimensions of cohesion could be used to discriminate program adherers from non11adherers. In this study a higher score on these dimensions was positively related to thelikelihood that the participant would remain in the exercise program. In a similar manner,Spink and Carron (1992) reported that one’s level of program adherence relatedpositively to the ATG-T and ATG-S dimensions of cohesion among group-based exerciseclasses. In a similar vein, Estabrooks and Carron (1 999a,b) found a positive relationshipbetween both task and social cohesion and adherence behaviours and, when examiningthe relationship between cohesion and adherence across multiple studies, Carron et a!.(1996) identified a general positive trend between task cohesion and exercise adherence.The positive relationship between cohesion and program adherence in group-basedphysical activity settings has also been observed within young (e.g., Spink & Carron,1994), middle-aged (e.g., Annesi, 1999) and older (e.g., Estabrooks & Canon, 1999a,b)adult populations (i.e., across the lifespan).Although the relationship between group cohesion and program adherence amongexercise group members is relatively well established, the specific antecedents, or causes,of cohesion in this setting has received limited attention. In spite of the paucity ofresearch designed to identify the antecedents of group cohesion within exercise settings,considerable attention within social and organizational psychology has sought to identifythe sources of cohesion within experimental and work groups. In social andorganizational psychology, a strong link has been established between intra-groupsimilarity and cohesion that is largely consistent with the categorization perspective ofintra-group similarity (e.g., Harrison et a!., 1998; 1998; Jackson et. al, 1991; Knippenberget al., 2007; Molleman, 2005; Tsui & O’Reilly, 1989; Terborg, Castore, & DeNinno,121976; Wiersema & Bird, 1993). That is, these studies generally report a positiverelationship between the degree of intra-group similarity and group cohesion.1.5 Surface-level and Deep-level Similarity in GroupsThere are multiple ways in which one can characterize similarity within groups(for a review see Harrison & Sin, 2005; Riordan & Wayne, 2008). Much of the pastresearch quantifying intra-group similarity has placed an over-riding emphasis on thephysical qualities of group members, particularly with respect to group members’ age,ethnicity, and gender makeup (e.g., Colquitt, Noe, & Jackson, 2002; Turban, Dougherty,& Lee, 2002; Tsui et al., 1995). However, psychological variables have also beenconsidered (e.g., Chatman & Flynn, 2001; Harrison et al., 2002; Klein, Conn, Smith, &Sorra, 2001). Furthermore, it has recently been suggested that a more completeunderstanding of intra-group similarity is likely to be gained by considering the physicalas well as the psychological composition of groups (Harrison & Sin, 2005). In line withthis suggestion, an emerging stream of group-based research has sought to examine bothphysical and psychological similarity though the consideration of a group’s degree ofsurface-level similarity (SLS) and deep-level similarity (DLS; Harrison et al., 1998;Phillips & Loyd, 2006).Surface-level variables, as defined by Harrison et al. (1998), refer to “overt,biological characteristics that are typically reflected in physical features” (p. 97).Examples of surface-level variables include age, gender, and ethnicity. In comparison,deep-level variables consist of characteristics that are not overtly observable and areusually discovered through extended personal communication. Examples of deep-levelvariables include attitudes, beliefs, and personal values. Surface-level variables are13thought to be analogous to physical attributes and deep-level variables are thought to beanalogous to psychological attributes (Harrison & Sin, 2005).Harrison and colleagues (1998) conducted a study on the effects of SLS and DESon group cohesion and social integration in work groups. They found that the group’slevel of SLS had the strongest effect on group cohesion initially (i.e., early in the stagesof group development), with those groups high in SLS reporting higher levels of cohesionwhen compared to the more surface-level heterogeneous groups. However, as timeprogressed, the effect of SLS on cohesion greatly decreased. The opposite was true forthe groups’ level of DLS. Initially, intra-group DES had little effect on cohesion but, overtime, higher levels of DLS were found to positively relate to group cohesion at anincreasingly strong degree.When interpreting the above results, Harrison and colleagues proposed that thepattern between SLS, DLS, cohesion, and time observed occurred as a result of eachmember’s acquisition of knowledge related to their group’s deep-level composition.Harrison and colleagues suggested that individuals in recently formed groups may haveto rely on surface-level attributes when initially assessing the degree of similarity presentwithin the group (i.e., the deep-level composition of the group is largely unknown at thistime). As a result, it is the degree of SLS that influences group cohesion initiallyfollowing the group’s formation. Over time, and as group members get to know eachother, they begin to discern the other members’ deep-level qualities (attitudes, beliefs,and values). As this information becomes known, the degree of DLS present within thegroup begins to influence the group’s level of cohesion at an increasingly strong degree.Coincidentally, as a function of obtaining information regarding the deep-level14composition of the group, the group’s degree of SLS begins to decrease in importance.Once this has occurred, the relationship between SLS and cohesion is thought to diminishto the level of non-existence (i.e., non-significance; Harrison et al., 1998).The secondary purpose of this thesis was to integrate the above SES and DESparadigm within group-based exercise settings in an attempt to better understand some ofthe social processes that occur among participants in these types of environments. Theimplementation of this conceptual model, based on surface- and deep-level variablesholds great potential as this system may more accurately capture the group- andindividual-level characteristics that relate to factors such as cohesion and programadherence.In this thesis five secondary hypotheses were tested. First, it was hypothesizedthat the extent to which exercise group members believed that they were similar to othermembers of the group in terms of surface-level qualities would be a strong positivepredictor of the level of cohesion they report initially following the group’s formation(Hypothesis 2A) and that perceived DLS would not significantly predict cohesion at thistime (Hypothesis 2B). It was also hypothesized that the predictive ability of perceivedSLS would decrease in strength over time, as the groups developed (Hypothesis 3A) andthat perceived DLS would positively predict cohesion in later stages of groupdevelopment (Hypothesis 3B). Finally, and in line with the research reviewed thus far, itwas hypothesized that group cohesion would positively predict program adherence(Hypothesis 4).1.6 Similarity, Attraction, and CohesionAs previously discussed, there is strong evidence to suggest that people are15generally attracted to others that they perceive as being similar to themselves (i.e.,Bryrne, 1971; Chen & Kenrick, 2002). Indeed, the relationship between similarity andattraction is considered to be of such strength that Byrne and Rhamey (1965) refer to thisphenomenon as the law of attraction.Within the group dynamics literature it has been found that similarity amongmembers enhances the level of attraction to the group (Pilkington & Lydon, 1997; Davis,1984; Royal & Golden, 1981). With respect to deep-level qualities (Harrison et al., 1998),intra-group interpersonal attraction has been found to correlate positively with attitude(Harrison et al., 2002; Lott & Lott, 1965; Singh, Ng, Ong, & Lin, 2008; Singh, Ho, Tan,& Bell, 2007), belief (Sachs, 1975), and value (Hobman, Bordia, & Gallous, 2004;Husian & Kureshi, 1983; Knippenberg & Schippers, 2007; Lee & Duck, 1982; Rokeach,1970) similarity.The strong relationship between similarity and attraction is closely mirrored bythe relationship between attraction and cohesion within groups. In fact, the latter twoconstructs share such a close relationship that several models of group cohesion haveconceptualized this variable based (sometimes primarily) on the level of attraction to theother group members (e.g., Bovard, 1951; Carron, et al., 2005; Deep, Bass, & Vaughn,1967; Dimock, 1941; Fiedler, 1954; Klein & Christiansen, 1969; Stokes, 1983;Wilkenson, 2007). As previously stated, cohesion within exercise classes is thought to becomposed of a task and social foci as well as individual and group orientations(Estabrooks & Carron, 2000). Accounting for the close relationship between attractionand cohesion, two of the dimensions resulting from this conceptualization of cohesion,ATG-T and ATG-S, make explicit reference to an individual’s attraction to the group16(Carron et al., 2007).To summarize the research reviewed in this section, similarity and attraction sharea close, positive relationship. Indeed perceived similarity with respect to attitudes, beliefs,and values (the three types DLS qualities proposed by Harrison et al., 1998) have eachbeen found to relate directly to attraction. The relationship between attraction andcohesion is also close, with current conceptual models (and measures) of cohesionincorporating attraction as a fundamental dimension of cohesion.As a result of the theoretical relationship between intra-group similarity, cohesion,and adherence it was hypothesized that cohesion would mediate the relationship betweenthe degree of perceived similarity and level of program adherence among participants(Hypothesis 5). This hypothesis stems from the established relationship betweensimilarity and cohesion (as described by Harrison et al., 1998) and cohesion andadherence (e.g., Estabrooks & Carron, 1999a) as well as the relationship betweensimilarity and adherence theoretically proposed in this thesis (see Figure 2).Figure 2 Representation of the hypothesized mediated relationship betweensimilarity, cohesion, and adherenceCohesionSimilarityCAdherence171.7 Assumptions Present in Surface-Level Homogeneous and HeterogeneousGroupsOne of the properties of group membership is that people’s sense of ‘self’ canbecome influenced by their mere involvement in that group. As a result, members in suchsocial contexts have a propensity to assume that other members share similar attitudesand beliefs as themselves (Allen & Wilder, 1979; Holtz & Miller, 1985; Wilder, 1984).As an example, Allen and Wilder (1979) asked university students to rate their preferencefor eight pairs of slides of paintings. The participants were then told that they were beingassigned to one of two groups, allegedly based on their preference for the paintings. Afterthis assignment the participants were asked to rate a hypothetical member of their group’sopinion on topics relevant (e.g., “what color combination do you prefer?”) and irrelevantto alleged group placement (e.g., “the national government is too conservative”). In thisstudy, participants expected other group members to hold similar opinions as themselveson all topics presented, regardless of whether the topic was used to determine groupplacement.Although Allen and Wilder found an effect stemming solely from groupmembership it should be noted that participants in their study did not get to see, orinteract with, the other group members (i.e., these members were hypothetical). Whengroup members do get the chance to interact with each other (and are exposed to theirsurface-level qualities) the degree of perceived SLS will likely influence perceptions ofDLS within the group (Jackson et a!., 1991; Levinger & Breedlove, 1966). After all, inthe vast majority of normal (i.e., non-experimental) situations only one’s surface-levelattributes are immediately accessible, even in the absence of social interaction (Zelimer18Bruhn, Maloney, Bhappu, & Salvador, 2008). As a result, initially, superficial/physicalcharacteristics provide the only basis of information from which one can attempt todetermine another person’s deep-level qualities (Harrison, et a!., 1998). Furthermore, asPhillips and Loyd (2006) suggest, “once a particular surface-level characteristic is madesalient, people generally assume that they hold more similar attitudes and beliefs withindividuals who share their surface-level characteristics then with people who do not, ontopics both relevant and irrelevant to the salient distinction”(p.146). In line with thissuggestion, it has been found that surface-level homogeneous groups generally assumethat their members have common deep-level qualities and are often surprised when theyfind evidence to the contrary (Chen & Kenrick, 2002; Graves & Powell, 1995). Chen andKenrick (2002) suggest that this surprise greatly increases the likelihood that dissimilarmembers will eventually leave the group.This inference of deep-level composition based on surface-level composition hasbeen referred to as the “congruence assumption” (Mannix & Neale, 2005,p.44). It ischaracterized by the belief that those group members who have a high level ofdemographic similarity (as a result of their surface-level attributes) with the respondentalso have similar deep-level attributes including attitudes (Jackson et al., 1991), beliefs(Fiske, 2000), values (Elsass & Graves, 1997), educational history (Milliken & Martins,1996) and past experiences (Pfeffer, 1983). These stereotypes likely permeate given that,in many cases a relationship between surface- and deep-level attributes exists (Harrison eta!., 1998; Jackson et a!., 1991). For example, age shares a positive relationship with risktaking propensity (Vroom & Pahi, 1971). Indeed, to a certain extent, all analyses of intragroup demographic similarity treat these observable qualities as indicators of the degree19of DLS present within the group (Knippenberg & Schippers, 2007; Priem, Lyon, & Dess,1999).The importance of the congruence assumption becomes reinforced when oneconsiders its relation to the rationale for the similarity-attraction hypothesis. As Byrne(1971) suggests, we are drawn to similar others in an attempt to reaffirm our logic andviews of the world (vis a vis attitudes, beliefs, and values). These variables may beconsidered ‘deep’ in nature as they are, for the most part, unobservable. However, giventhe relationship between many surface-, and deep-level variables, surface-level attributesare often used as proxy indicators for inferences regarding DLS. It is this DES that isthought to relate to attraction. In other words, following Byrne’s rationale, it is not thesurface-level makeup of an individual per se that lies at the heart of attraction. Instead itis the level of DLS present that influences attraction which is inferred in the first instanceby perceptions of one’s own and others’ surface-level attributes.In this study two final (tertiary) hypotheses were tested. First, it was hypothesizedthat group members who initially perceived a high degree of similarity betweenthemselves and the other group members with respect to surface-level qualities wouldalso perceive a high degree of similarity among deep-level qualities early in the group’sdevelopment (i.e., before deep-level information is likely known). In other words, apositive correlation was expected between the initial SLS and DES perceptions(Hypothesis 6).The second hypothesis concerned the relationship between changes in perceptionsof similarity across data collection periods. As previously mentioned, in group settings,members have a tendency to assume that other members of the group will hold similar20attitudes to themselves (e.g., Allen & Wilder, 1979) especially if the group has ahomogeneous surface-level composition (Phillips & Loyd, 2006). In line with thisfinding, Chen and Kenrick (2002) have reported that when this assumption is violated(i.e., attitudes between an individual and other group members are perceived to benotably dissimilar) the level of attraction between dissimilar members decreases. Thus,(regardless of the level of SLS within a group), if a member’s expectation of the degreeof intra-group DLS is violated (i.e., one’s initial perception of deep-level similarity withinthe group is interpreted as being over-estimated/inflated) a low level of attraction to thegroup is likely to be experienced by that member. Consequently, it was hypothesized thata lower degree of cohesion would be reported among those members who, over time,perceived reduced levels of DLS within their group (i.e., they perceived less DLS withinthe group than they initially assumed; Hypothesis 7).1.8 Exercise and Intra-Group Similarity StudiesAs mentioned earlier, a surprisingly limited amount of research has explored therelationship between intra-group similarity, adherence, and cohesion in exercise groups.Recently, Beauchamp et al., (2007) reported that individuals generally display apreference for exercising within groups that are comprised of members of a similar ageand a general dislike of exercising with those much younger or older than oneself. In thisstudy participants from across the adult age span were asked to rate their preference forone individual-based, and three group-based exercise contexts. The group-based contextspresented in this study varied solely in the mean age of the members in each hypotheticalgroup. Specifically, participants were asked to rate their preference for exercising in agroup comprised mostly of people in their: 20s and 30s, 40s and 50s, or 60s and 70s. In21accordance with Beauchamp et al.’s hypothesis, a preference for group-based contextscomposed of members of a similar age as the respondent was identified in all age groups.Also, as the disparity in the age between the hypothetical group members and theparticipant increased, personal preference for the given group-based context decreased. Ina more recent study employing a similar research design, Dunlop and Beauchamp (2008)found that, across the age-spectrum, people reported a greater preference for gender-segregated classes when compared to gender-integrated classes (i.e., males preferredmale-only classes, females preferred female-only classes).Although these two studies did not link these preferences to outcome measuressuch as cohesion or program adherence, the results reported are none the less consistentwith the literature reviewed. Age and gender each represent surface-level qualities. Basedon the position forwarded in this thesis, intra-group similarity on each of these variablesis predicted to correlate with each member’s attraction to the other group members.The sole study, to date, that has explored the relationships between intra-groupsimilarity, cohesion, and adherence in exercise groups was conducted by Shapcott andcolleagues (2006). In this study, personal and group attributes were conceptualized asbeing either task-related (self-efficacy, level of previous activity, and personal goals) ortask-unrelated (ethnicity and gender). Groups of six people completed an eight-weekcourse in which a collectively agreed upon number of miles were to be walked by thegroup. The authors reported that only similarity in the level of previous physical activityamong the group members correlated (positively) with both cohesion and adherence.Thus, the more similar members were in their previous physical activity behaviours, themore cohesive the group was and the higher the level of adherence in the group.22Although only one of the intra-group similarity variables investigated by Shapcottand colleagues (2006) was found to relate to cohesion and adherence, severaldissimilarities between their investigation and the proposed study should be noted. First,the ‘groups’ in the Shapcott et al. study differed from the groups proposed for inclusionin the current study with respect to the manner in which the prescribed activities wereperformed. While each of the walking groups in Shapcott et al.’s study were invited tospecify a group goal (i.e., miles walked collectively), participants did not have to pursueand accomplish this goal together. Instead, each member was responsible for a certainproportion of the total amount of miles to be walked and was free to complete thisactivity with others or by themselves based on their own preference. Under theseconditions, the degree of intra-group similarity may have had a less pronounced effect onadherence, given that if a particular member did not feel as though he or she was similarto the other group members, this member could still complete the required activitywithout interacting with the rest of the group.Second, the indices of intra-group similarity used by Shapcott et al. wereobjective (i.e., non-perceptual) in nature whereas the measures of intra-group similarityused in the current study were subjective (perceptual) in nature (this point is discussedmore thoroughly below). Recent research (Riordan & Wayne, 2008) has found thatperceptual measures of intra-group similarity have a stronger relationship with certainoutcome measures when compared to more objective measures. Indeed, perceptions ofreality have been argued to hold a dominant influence on outcome measures such asattitudes and involvement (Lawrence, 1997). It follows that perceptions of certain intragroup similarities may have influenced cohesion and adherence within Shapcott and23colleagues’ groups; however, the researchers would have been unable to document theseeffects given the type of measures they employed.Finally, no attention was paid by Shapcott et al., to the effect that time may haveplayed in the relationship between similarity, cohesion, and adherence. Given the strongfoundation within organizational psychology suggesting that the relationship betweenintra-group similarity and cohesion is dynamic in nature (e.g., Harrison et al., 1998;Turban et al., 2002) it remains possible that Shapcott and colleagues failed to identify therelationship between these variables in its entirety.In summary, the study proposed in this thesis attempts to apply a framework fromsocial and organizational psychology to understand some of the determinants of cohesionand adherence within a group-based exercise setting. Given that many differences existbetween work and exercise groups several integrative issues require consideration. Theseissues are outlined below.1.9 Can Exercise Groups Really be Considered ‘Groups?’Needless to say, the question of whether an exercise group can be considered atrue group will ultimately come down to how one defines a group. One universallyaccepted characteristic of a group is that it must be composed of two or more individuals(Carron et al., 2005). Building on this requisite, Alcock et al., (1998) suggested that inorder for a collection of individuals to be considered a group the members must also beaware of each other, influence one another, share common goals, engage in ongoingrelationships, and perceive themselves as belonging to the group. Typical exercise groupswill likely meet these conditions (though to varying degrees). However, it has also beenargued that a collection of individuals must share a common fate in order to be24considered as true group (Fielder, 1967). This last criterion potentially poses a seriouschallenge to recognizing exercise groups as true groups.Although each member of a given exercise group may share a common goal (e.g.,to lose weight or increase muscle mass) one’s progress to that goal is largely independentof the other members of the group (in any given exercise group some members willachieve their goal, others will not). It is unlikely that members who attain their personalgoals would consider the program a failure if all other individuals in the group did notreach their own goals. For that reason the characteristic of common fate does not seem tobe met by typical exercise classes.Given that exercise groups do not meet the qualification specified above, Spinkand Carron (1992) likened exercise groups to what Taj fel and Turner (1979) consideredminimal groups. Specifically, Spink and Carron (1992) noted, “these are socialcategorizations that develop because humans possess a need to enhance and protect theirself-esteem. Because of this need to enhance and protect self-esteem, people have astrong motivation to develop social bonds and social identities from their memberships incollectives”(p.9).Several parallels can be drawn between this perspective and the rationale for thesimilarity-attraction hypothesis proposed by Byrne (1971; i.e., the reinforcement of worldviews). Regardless, even though exercise groups would fail to meet the qualificationsuggested by Fielder (1967) they nonetheless may still be considered a type of group. Inaddition, Burke et al. (2005) have argued that exercise groups may be considered to begroups on statistical grounds. This argument is supported by the finding that exercisegroups show agreement regarding the degree of cohesion present within the group and a25divergence on this construct between groups.1.10 How is Intra-Group Similarity Measured within Exercise Groups?In the majority ofprevious research that has attempted to quantify interpersonalsimilarity within groups one of two approaches has typically been adopted. The firstapproach requires that group members complete measures that assess their individual-level attributes (Mohammed & Angell, 2004). For example, participants’ age, ethnicity,satisfaction with supervision (e.g., Harrison et al, 1998), level of previous physicalactivity, self-efficacy (e.g., Shapcott et. a!, 2006) or other variables relevant to theresearch questions posed may be assessed. With these data, indices of variability amongthe group members on each of these variables are then calculated. Typically these indicesof variability provide an indication of the degree of ‘groupness’ (e.g., intra-clustercorrelation; see Klein & Kozlowski, 2000), or a measure of disparity between eachmember and the rest of the group (e.g., Euclidean distance scores; see Harrison & Sin,2005; Riordan & Wayne, 2008). Examples of previous applications of these indicesinclude the quantification of a group’s personality composition (e.g., Barry & Stewart,1997) and age diversity (e.g., Tsui et al., 1992). Among diversity researchers thesemeasures are referred to as ‘objective’ or ‘actual’ assessments of intra-group similarity(Harrison & Sin, 2005). This operationalization is used in recognition of the fact that thecharacteristics of others in the “group represent part of the objective contextualenvironment in which an individual operates” (Riordan & Wayne, 2008, p. 566).A second approach used by researchers interested in assessing intra-groupsimilarity is to measure each group member’s perception of the level of similarity withinthe group. Perceived intra-group similarity is conceptualized as the degree to which an26individual believes that members of the group are similar to himself or herself(Knippenberg & Schippers, 2007; Strauss et al., 2001). Several researchers haveoperationalized perceptual similarity with a high level of specificity by stating thatperceived similarity must be assessed in reference to a specific (Crutchfield, Spake,D’Souza, & Morgan, 2003) or relevant (Riordan & Wayne, 2008) attribute (e.g., age,gender, attitudes). However, general (hereafter referred to as ‘global’) perceptions ofsimilarity have also been investigated (e.g., Hobman, Bordia, & Gallous, 2003; Hobmanet al., 2004).Perhaps unsurprisingly, actual (or objective) and perceptual (or subjective)measures of similarity have been argued to represent different facets of the intra-groupsimilarity construct (Riordan, 2000; Riordan & Wayne, 2008; Zellmer-Bruhn et al., 2008)and several reasons have been given for considering the study of perceptual similaritybeyond actual similarity. By definition, measures of actual similarity fail to account forthe salience and importance that each member ascribes to certain dissimilarities that mayor may not exist objectively in the group (Hobman et al., 2003). Specifically, therespondent is prevented from considering all the attributes and qualities that they deemrelevant to their conceptualization of their group’s degree of similarity (Hobman et al.,2004). Indeed as Randel and Earley (2009) state, “a purely ‘objective’ assessment mayfail to capture how team members perceive similarity within the team, how each teammember views multiple similarity characteristics on that team, and how thesecharacteristics are relied on to varying degrees in describing others on the team” (p. 807).Perceptions of similarity have been found to correlate with subsequent behavior(Hensley, 1981) and these perceptions have often been found to relate more strongly with27outcome measures than measures of actual similarity (Hobman et al., 2003; Montoya etal., 2008; Orpen, 1984; Riordan & Wayne, 2008; Strauss et al., 2001; Turban & Jones,1988). For example, when comparing the predictive strength of perceptual and non-perceptual (i.e., objective) measures, Riordan and Wayne (2008) reported that theperceptual measures were “more often related to and accounted for more variance in theoutcomes” (p. 582) than comparable objective measures. This likely stems from the factthat, as Ferris and Judge (1991) suggest, “people react on the bases of perceptions ofreality, not reality per Se”(p.464). It follows that the level of similarity within a groupthat one perceives and the actual level of intra-group similarity may not necessarily relate(Dose, 1999; Randel & Earley, 2009) and an emphasis on perceptual, as opposed toactual, similarity may be justified.Although perceptual similarity is distinct from actual similarity and, in manycases may be a more significant predictor of salient outcome variables, there is a relativedearth of research examining the effects of perceived similarity in diversity research(Harrison et al., 2002; Riordan & Wayne, 2008; Zellmer-Bruhn et al., 2008). As a result,considering perceptions of intra-group similarity has been identified as an important areaof future research (Harrison & Sin, 2005; Riordan 2000; Riordan & Wayne, 2008).Indeed, exploring perceptions of intra-group similarity has been recognized as a“necessary step towards a more complete understanding of how diversity influences teamoutcomes” (Zellmer-Bruhn et al., 2008,p.52). For these reasons, perceptual, as opposedto actual measures of similarity were employed in this study.Parenthetically, if one were to attempt to describe the variance in deep-levelqualities within the group, using objective measures of intra-group similarity, all28specific/salient qualities would have to be measured. Within the current study, this wouldbe less of a problem for surface-level variables, given that there are a relatively fewreadily apparent surface-level characteristics that are theoretically tenable (e.g., age,ethnicity, gender, physical appearance). However, when attempting to conceptualize thepotential range of deep-level qualities of interest there are countless attitudes, beliefs, andvalues that could be considered equally valid for study inclusion (e.g., political affiliation,environment, views on human rights, welfare legislation, death penalty, legalization ofdrugs, gay and lesbian rights; see Chen & Kenrick, 2002, for an extensive list). Thebenefit of measuring similarity ‘perceptually’ is that perceived similarity with respect togeneral attitudes, beliefs, and values can be assessed. Indeed, by utilizing ageneral/omnibus measure, participants can consider those attitudes, beliefs, and valuesthat are personally salient for them. The limitation of measuring similarity in this way isthat specific information is not provided with respect to which attitudes, beliefs, andvalues the respondent is basing their rating of similarity on (Riordan & Wayne, 2008).Provided that a relationship between these (general) deep-level qualities and exerciseadherence is identified in this study, future researchers may subsequently look to explorewhich specflc attitudes, beliefs, and values contribute to this relationship.Consistent with the recommendations of Riodan and Wayne (2008), theperceptual content of the questionnaire used in this research included multiple items toassess perceived surface-level (e.g., age, gender, ethnicity, physical appearance) anddeep-level (values, attitudes, beliefs, life experiences) similarity. In addition, and also inaccordance with the recommendations of Riodan and Wayne, these perceptions werecollected repeatedly (discussed in the procedure subsection), and related to objective29outcome measures (i.e., the objective measure of adherence described in the nextsection).302 METHOD2.1 Participant RecruitmentThere were four criteria that potential exercise courses had to meet to beconsidered for inclusion in this study. The first criterion was that participant preregistration for the entire course was required (i.e., the course was described as‘registered’ in nature). Thus, ‘drop-in’ classes were excluded from this study as thesetypes of programs require no such commitment. This criterion was adopted in an attemptto ensure that the same group (i.e., the same group members) met on a weekly basis.Several of the hypotheses proposed in this study dealt with the developmental pattern ofcohesion and adherence as a function of group composition. If this composition changedevery class (i.e., the same members were not expected to be present every class) thenthese hypotheses could not be tested. Also, drop-in classes frequently do not commenceand conclude within a single term as these types of programs often run on an on-goingbasis throughout the year. As a result, it would not be possible to study these groups fromtheir inception.The second criterion for study inclusion was that the activities performed in thestudied courses were physical in nature and required some corporal exertion. Manyleisure and community centres have moved away from the term ‘exercise’ courses andhave chosen to adopt the more inclusive term of ‘health and weilness’ courses. The latterterm is more inclusive in that it allows for some attention to be given to one’s mental, asopposed to strictly physical, health. Although many of the courses involved in thisinvestigation were classified as health and weilness, instead of exercise, courses, allcourses included in this study met the second requirement for inclusion. In other words,31although some of the courses included in this study did allot time for the advancement ofpsychological wellness they also dedicated a substantive amount of time to physicalexercise and exertion.The third requirement was that, in each course, classes were scheduled to meetonce per week. Estabrooks and Carron (1999a) have suggested that perceptions ofcohesion may be influenced by the frequency of contact. This third criterion wasintroduced in an attempt to control for any differences in the social interaction that mayoccur between weekly and more frequently meeting classes (i.e., two or more times perweek).The final criterion for study inclusion was that potential courses had to be at leasteight weeks in length. This was necessary to ensure that data could be collected at allthree time points (i.e., during the second, fifth, and eighth week of the course). Inaddition, this length is consistent with past research on the effects of diversity oncohesion and adherence in exercise groups (i.e., Shapcott et al., 2006).The population of interest in this study comprised participants registered in group-based exercise courses meeting the criteria listed above. In an attempt to accuratelyrepresent this diverse population, a list of all courses offered at community centres inVancouver, West Vancouver, North Vancouver, and Richmond meeting the abovecriteria was first compiled. This was done by consulting each centre’s recreation guidefor the ‘winter term’ (January through March) of 2009. Several fitness centres in theGreater Vancouver Area were also contacted regarding the group-based exercise coursesthey offered. Unfortunately, the vast majority of these fitness centres offered group-basedprograms solely on a drop-in basis. As a result, community, as opposed to fitness, centres32were targeted for study inclusion.Upon receiving ethical approval from the University of British Columbia’sBehavioral Research Ethics Board (; seeAppendix A,B) program coordinators at potential sites of data collection were sent aletter of initial contact (See Appendix C) via email. This letter briefly explained thepurpose of the project and stated that the study’s principal investigator would be in touchvia telephone to discuss the possibility of their centre’s participation. A list of the centre’scourses that would be appropriate for study inclusion was also included in the body ofthis email.Within one week of sending this email the study’s principal investigator contactedeach program coordinator via telephone. The intention of this phone call was to furtherexplain the purpose of the study, address any questions or concerns the coordinator mayhave had, and to determine if the program coordinator would be amenable to participatingin this study. If the coordinator agreed to participate, a request for the contact information(i.e., email addresses) of the course instructors working at the coordinator’s centre wasmade. A letter of initial contact specifically designed for the course instructors (seeAppendix D) was then sent via email to each of the instructors (N = 22) at the respectivecommunity centres. This letter briefly explained the study’s purpose and procedures,made the request for each instructor to keep accurate attendance records, and informedthe instructor that they would receive a $30.00 honorarium (per course) in exchange fortheir participation in this study.2.2 ParticipantsParticipants were recruited from 46 gender-integrated exercise courses (from nine33community centres) within the Greater Vancouver Area. Based on the course descriptionprovided in each community centre’s program guide, 48% (n = 22) of the includedcourses were categorized as ‘yoga’ courses, 40% of the included courses (n = 17) werecategorized as ‘pilates’ courses, and 15% ofthe included courses (n = 7) were categorizedas ‘strength or conditioning courses.’ The average class size of these courses was 13.43people (SD = 5.59). This number was determined through the consultation of eachcourse’s registry. 85% of the individuals registered in these courses were female and 15%were male. In total, 402 individuals participated in this study. 84.8% of participants werefemale and 15.2 % were male (note that the sample characteristics closely matched thepopulation characteristics for gender). The average age of respondents was 47.49 years(SD 14.68; see Table 1 for demographic description of the sample).Table 1 Descriptive StatisticsVariable Males FemalesN 60 336Age (years) 49.15 (13.30) 47.14 (14.89)Canadian (%) 66.70 65.20Height (inches) 70.67 (2.43) 64.91 (2.82)Weight (pounds) 179.24 (23.36) 139.15 (23.28)Body Mass Index 25.10 (3.04) 23.25 (3.46)Regularly Active (%) 21.67 18.45Note: Standard deviations provided in brackets342.3 MeasuresActivity Status. Participants were asked to report the typesof physical activitythey engaged in (e.g., cycling, running) during the two weeksprior to completing thequestionnaire. Once these activities had been identified they were prompted toprovidethe number of times they engaged in each activity as well as theduration of these bouts.Finally, participants were asked to specify the intensity level of eachactivity. Thisintensity was indexed based on the changes in breathing and heart rate asexperienced bythe participant. Four levels of intensity werespecified: (a) no, (b) small, (c) moderate,and (d) large increases in heart rate and breathing. Consistent with pastresearch(Beauchamp et al., 2007; Caspersen, Christiansen, & Pollard, 1986; Young, King,& Oka,1996; Wilcox, King, Brassington, & Ahn, 1999)participants were classified as ‘active’ ifthey reported engaging in moderate or vigorous (i.e., moderate tolarge increases in heartrate and breathing) activities for a duration of at least 20 minutes threeor more times perweek. This status was determined using the first questionnaire that each participantcompleted (for a small number of participants activity status wasdetermined using datacollected from the second or third data collection period).Adherence. Program adherence was assessed through two different,complementary methods. The first method was instructor-mediated, the secondmethodwas self-reported in nature. Instructors participating in this study were asked torecordclass attendance over the first eight weeks of the course on alog sheet (see Appendix E).In addition, during the last data collection period participants wereasked to report thenumber of classes that they had missed since the program began. This numberwas thensubtracted from the total number of classes held (eight) to arrive at a measureof the35number of classes the participants believed they had attended. Following the protocol ofprevious research (e.g., Annesi, 1999; Eastabrooks & Carron, 1999 a,b) both measures ofadherence were converted from raw scores based the number of classes attended to thepercentage of classes attended. This percentage was calculated by dividing the number ofclasses that the participant attended by the total number of classes offered. The instructor-mediated and self-report measures were found to correlate significantly (r .63,p<.001).Cohesion. The Physical Activity Group Environment Questionnaire (PAGEQ;Estabrooks & Carron, 2000) was used to measure cohesion within the exercise groupsincluded in this study. This measure is multidimensional in nature and has been found tohave an acceptable level of content, concurrent, factorial and predictive validity(Estabrooks & Carron, 2000). Consistent with past research employing this measure (e.g.,Estabrooks & Carron, 2000), each item of the PAGEQ was assessed using a 9 pointLikert-type scale with responses ranging from ‘strongly disagree’ (1) to ‘strongly agree’(9). Subscale scores were created by averaging the appropriate item responses. Thisresulted in scores ranging from one to nine for each of this measure’s four subscales.Demographic Variables. Participants were asked to report their gender, age,current occupation, level of education achieved, height and weight. Each participant’sBody Mass Index (BMI) was calculated by dividing the square of his or her self-reportedheight (m2)by self-reported weight (kgs). A measure of ethnicity was also included. Thismeasure included a list of the 21 most frequently identified ethnicities/nationalities forVancouver(www. 1 1 /products/highlight/ETO/Table 1 .cfln?LangE&T5360l&Gv2&GID=933) and asked respondents to place a checkmark beside each ethnicitythey identified with.Similarity. Consistent with past research (e.g., Phillips & Loyd, 2006; Phillips,Northcraft, & Neale, 2006), on a nine point Likert-type scale (ranging from stronglydisagree to strongly agree), participants were asked to rate the degree to which they feltthat the other members of their group were similar to them on five surface-level and fivedeep-level attributes (e.g., In my exercise class, I believe that group members are similarto me in terms of age) It should be noted that this approach is markedly different thanassessing the degree to which participants perceived all members of the group to besimilar.To highlight the above this distinction, consider a group with five homogeneousmembers and one radically divergent member. If this diverse member was asked to ratethe degree to which members of the group are similar, he would most likely report thatmembers of the group are indeed very similar. After all, five of the six members of thegroup would be very similar. However, if this same person was asked to rate the degreeto which he felt that members of the group were similar to him a different response wouldlikely emerge given his relation to the rest of the group. Thus, the two measures ofperceptual similarity described above assess two distinct aspects of the perceived degreeof similarity within a group. It follows that the measurement of perceptual similarity usedin this study displays an egocentrically relational characteristic. This is because theapproach used taps into respondents’ perceptions of themselves (which may beconsidered to be egocentric) in relation to the other members of the group (which, bydefinition is relational in nature).37Perceptions of each surface-level and deep-level attribute were assessed using asingle-item methodology. Although use of single-item measures may potentiallycompromise reliability, Riordan and Wayne (2008) suggest that “asking individuals abouttheir similarity in demographic characteristics is similar to asking then about their owndemographic characteristics” (p. 572). Single item measures regarding demographiccharacteristics are extensively used in diversity research and have been found to bereliable (e.g., Crampton & Wagner, 1994). It follows that concerns regarding thereliability of the single-item perceptual measures can be largely dispelled.Harrison et al., (1998) identif’ age, gender, and ethnicity (all three being surface-level qualities) as the three most common attributes studied among diversity researchers.As a result, measures of perceived similarity with respect to the above three attributeswere assessed in this study. The additional surface-level perceptions included in thisstudy were in reference to physical appearance and physical conditioning similarity.Fiske and Taylor (1991) suggest that the salience of member characteristics vary acrossdifferent contexts and situations. The final two surface-level perceptions were includeddue to the physical nature of the activities performed in group-based exercise settings,and due to the likelihood of physical appearance and condition being highly relevantwhen these activities are performed.In contrast to the strong theoretical and empirical foundation upon which toidentify salient surface-level qualities, limited guidance was available when attempting tocreate a comparable list of deep-level attributes (Harrison et al., 1998). This is likely aresult of the fact that, conceivably, hundreds of deep-level attributes could be considered(e.g., Chen & Kenrick, 2002; Meglino, Ravlin, & Adkins, 1989). Given the definition of38‘deep-level’ attributes provided by Harrison et a!. perceptions of similarity with regard toattitudes, personal values, and personal beliefs were included in this study. Also,consistent with Milliken and Martin (1996), educational similarity was classified as adeep-level attribute and included as the fourth deep-level perception assessed in thisstudy. The final deep-level perception included in this study was in reference to previouslife experiences. It was believed that these five perceptions would provide a general senseof the degree to which individuals considered that they were similar to other groupmembers on a ‘deep’ (i.e., non-observable/psychological) level.Two ‘global’ measures ofperceived similarity were also included in this study.These measures were global in the sense that they each assessed a general perception ofinter-personal similarity without reference to a specific attribute or trait. The first globalmeasure (hereafter referred to as the ‘overall’ measure of perceived similarity; OPS)simply asked participants to provide an overall rating of their perceived inter-personalsimilarity with other group members on the same nine-point Likert-type scale describedabove. The second measure of global similarity asked participants to report the number ofgroup members they felt “very similar” to as well as the perceived size of their group. Aperceived proportion of similar others variable (hereafter referred to as the measure ofperceived ‘proportional’ similarity; PPS) was calculated by dividing the total number ofindividuals the participant felt very similar to by the perceived group size.The OPS variable required consideration of all members present within the group(including those that were perceived as dissimilar or incongruent) and was informed bythe notion of prototypicality and depersonalized attraction (Hogg, 1992; 1993; Hogg,Abrams, Otten, & Hinkle, 2004; Turner, 1987). According to Hogg et al. (2004),39members of a group tend to conceive a prototypical group member as a result of agroup’s composition and task. This prototype is thought to embody all positivecategorizations present within the group. As an example, in this case of exercise classes,the prototypical member may be a physically fit and active person. In any manner,individuals are thought to index their attraction to the group based on the degree to whichgroup members accord with this prototype. This attraction is depersonalized in so far as itis indexed based on reference to an aggregated conception of all members within thegroup, and not any one specific person within the group. The notion of prototypicalityand depersonalized attraction relate to the OPS measure since this prototype is created asa result of the composition of the group in its entirety.In comparison, the construction of the PPS variable was informed by the beliefthat individuals would be attracted to similar others — especially within groups (seeByrne, 1971). In other words, individuals may still be attracted to, and retain membershipin, a group composed of dissimilar others provided they can identify with at least some ofthe other members of the group. From this perspective, an overall perception of similaritybetween the respondent and all (or most) group members would not be required forincreased functioning, especially if the task is not excessively interdependent in nature.2.4 ProcedureA research assistant was present at the end of the first class of each of the 46courses included in this study. At this time, the research assistant solicited the groupmembers’ attention and requested a couple of minutes of their time. This researchassistant then informed the class about the general purpose (to explore some of the factorsthat influence adherence to these types of programs) and procedure of the study. With40respect to the procedures associated with this study, each class was told that a researchassistant would be present at the end of the second, fifth, and eighth class with a verybrief, two page questionnaire. Group members were instructed to approach the researchassistant and request a blank questionnaire if they wished to participate in this study. Theresearch assistant also stressed that participation in this study was entirely voluntary andthat the potential participants would not endure any negative repercussions as a result ofdeclining the offer to participate or withdrawing from this study. Following this verbaldescription of the study, the group members were provided with a written description ofthe study, in the form of an information letter (see Appendix F).Consistent with the information provided after the first class, a research assistantwas present at the end of the second class of each of the 46 courses included in this study.This research assistant once again solicited the attention of the group members at thistime, reminded them about the study, and requested that class members fill out a blankquestionnaire if they wished to participate. This questionnaire (see Appendix G) includedthe perceptual measures of similarity outlined above as well as the PAGEQ. In addition,several demographic measures were included in this document. In accordance with theprocedures approved by the University of British Columbia’s Behavioral Research EthicsBoard, written consent on the part of the participants was not required. Instead, consentwas demonstrated through each member’s choice to participate. This process wasrepeated after the fifth and eighth class of the course, with the only difference being thatthe questionnaire administered after the eighth class was slightly modified (see AppendixH). Specifically, this updated questionnaire no longer requested information regardingeach participant’s ethnicity and instead included a self-reported measure of program41adherence.423 RESULTS3.1 Response RateAll course instructors participating in this study were asked to keep accurate andup-to-date attendance records for each of their class members. Unfortunately, while manyinstructors were very diligent in this regard, many others neglected to keep any record ofcourse attendance at all. Of the 22 instructors participating in this study eight (3 6.36%)failed to maintain an accurate attendance log. These eight instructors taught a total of 13of the 46 classes studied. Thus, the instructor-mediated attendance records were availablefor 7 1.74% (n =33) of the classes included in this study. As a result, instructor-mediatedattendance records were available for 68.40% (n = 275) of the sample and unavailable forthe remaining 3 1.60% (n 127).Participation rates were calculated for each course and within each data collectionperiod. This was done by dividing the number of registered group members whocompleted the questionnaire during each data collection period by the total number ofregistered group members in attendance on that collection period (i.e., on the second,fifth, and eighth class). As a result of this method, response rates were only tabulated forthose courses in which instructor-mediated attendance records were kept. During the firstdata collection period (after the second class) the average group response rate was77.59% (SD = 18.20%). The average group response rate for the second (after the fifthclass) and third (after the eighth class) data collection periods were 60.67% (SD22.58%) and 71.49% (SD = 20.18%) respectively. The low response rate observed duringthe second data collection period is addressed below.433.2 AnalysisPrior to the main analyses, the data were screened for any errors that may haveoccurred during imputation using SPSS Frequencies and Descriptives. If an anomaly wasidentified then the original source (i.e., the completed questionnaire) was consulted. Oncethis was done all values were found to reside within the appropriate range for eachvariable. In addition, each variable’s mean and standard deviation was deemed plausible.Next, the data were once again screened to identify those participants whocompleted a questionnaire in two or more different courses. Six such participants wereidentified. Each of these six participants completed the questionnaire in two differentcourses. As a result, one of the two entries for these participants was removed at random(via coin flip) resulting in the sample of 402 participants.Following these case deletions, data corresponding to all of the study variableswere examined using SPSS Missing Values Analysis (MVA). It is generally assumed thatif the percentage of cases with missing values for a given variable is greater than or equalto five percent then the pattern of missing responses for the given variable should beexamined (Tabachnick & Fidell, 2007). The pattern of missing data for a given variablecan be classified in one of three ways: missing completely at random (MCAR), missing atrandom (MAR), and not missing at random (NMAR). As the name suggests, when thedata are MCAR it can not be predicted by any of the other variables in the data set. Inother words, there is no relationship between the pattern of response/non-response for thegiven variable and the other variables (including independent and dependent variables).As Tabachnick and Fidell (2007) state, this is “the best of all possible worlds, if data mustbe missing” (p. 63). If data are found to be MCAR then the researcher is justified in44analyzing the data in a routine way as concerns regarding the pattern of response/non-response can largely be dispelled. When data are MAR the pattern of missing values canbe predicted from at least one of the other independent variables in the data set (as aresult this term is really a “misnomer”; Scheffer, 2002,p.153). Finally, when the data isNMAR the pattern of response/non-response is related to the dependent variable(s)(Scheffer, 1997). Several options exist to deal with data that are classified as MAR orNMAR (for a review see Scheffer, 1997; Tabachnick & Fidell, 2007) however, ifresearchers wish to be conservative in their analyses they can simply delete or remove thedata that are MAR or NMAR. This was the approach adopted in this study.Responses for each data collection period were examined separately to identifythe pattern of missing responses. SPSS MVA provides a test to determine whether one’sdata are MCAR or MAR/NMAR with a significant result indicating that the data areeither MAR or NMAR. The pattern of missing data from the first,x2(301)= 3l6.98,p =.252, and third,2(218) = 195.56,p= .860, collection periods (collected during thesecond and eighth week of the course) was found to be MCAR. In contrast, the pattern ofmissing data from the second collection period (i.e., the fifth week of the course) wasfound to be MAR/NMAR,2l98) = 270.128,p< .001. Subsequently, data from thiscollection period was removed from all subsequent analyses.The data were then screened for potential univariate outliers using SPSSDescriptives (to record the standard scores of each variable in the database) and Explore(to identify the five highest and lowest standard scores for each variable). Six standardscores in excess of 3.29 were identified. Since these outliers were not a result of a dataimputation error and one can expect a few standard scores in excess of 3.29 in a large45sample (Tabachnick & Fidel!, 2007), these cases were left in the data file.Scores for each of the four subscale measures of the PAGEQ were then calculated(following the protocol outlined by Estabrooks and Carron, 2000) for the first and lastdata collection periods and added to the data file. All four of these measures at each ofthe time points (after class 2 and class 8; hereafter referred to as the first, and second datacollection periods respectively) were found to have an acceptable level of reliability(Cronbach’s Alpha .87). The percentage of the group each participant felt “verysimilar” to at both time points was then calculated by dividing the number of people therespondent felt very similar to by the respondent’s perceived group size.Consistent with past research on intra-group similarity (e.g., Harrison et a!., 2002;Turban et al., 2002), the possibility of reducing the surface-level and deep-levelperceptual measures of similarity into broader factors was then explored. Harrison andSin (2005) stressed caution when aggregating individual measures of specific similaritiesinto composite measures or ‘factors.’ They contend that “diversity is meaningful [only]when it is more narrowly defined or dimensionalized”(p. 199). This caution likely stemsfrom the fact that intra-group similarity factors composed of a combination of attributesthat have no conceptual or theoretical relationship are quite difficult to interpret (Riordan& Wayne, 2008) and the haphazard construction of these types of factors runs the risk ofmasking the effects of each of the individual perceptions included (Harrison & Sin,2005). In an attempt to address these reservations, four criteria were adopted to evaluatethe validity of the potential intra-group similarity factors being considered. First, theperceptions included in each factor were required to be significantly inter-correlated atboth time points (i.e., the relationship among components was stable across time points).46Second, and perhaps most importantly, a conceptual/theoretical argument for thecomposition of each factor was required (thus ensuring interpretability). Third, eachmodel was required to fit the data adequately. Finally, the factors that met the proceedingcriteria were required to have an acceptable level of scale reliability at both time points.In this study, five single-item measures were used to assess perceptions ofsimilarity in relation to surface-level qualities (i.e., age, gender, physical condition,physical appearance, and ethnicity). Five items were also used to assess perceptions ofsimilarity in relation to deep-level qualities (i.e., attitudes, education, personal values,personal beliefs, and previous life experience). As a result, the possibility of creatingcomposite ‘surface-level’ and ‘deep-level’ measures based on the combination ofapplicable single-item measures was deemed appropriate. An acceptable case-to-variableratio (Floyd & Widaman, 1995) was achieved as a result of the sample size at the twotime points, and the number of individual-item measures included in each a priori factor.The first criteria that these two potential factors had to satisfy was that each of thefive items composing each factor were related (significantly) at both time points. Both ofthe potential factors (surface and deep) satisfied this requirement when analyzing the datafrom the first data collection period (see Table 2). All five of the single-item deep-levelperceptual measures were also found to correlate significantly with each other among thedata collected during the final time period (see Table 3). However, when analyzing thesingle-item surface-level perceptions collected during the last time period a differentpattern emerged. Among these items, the ethnicity and gender measures did not correlatesignificantly with each of the other three measures. As a result, the possibility of creatinga ‘surface-level’ factor composed of all five of the single-item surface-level perceptions47was compromised. Attention was then focused on testing the validity of the five item‘deep-level’ factor (hereafter referred to as the DLS factor) and three item ‘surface-level’subset factor composed of perceptions regarding member’s age, physical condition, andphysical appearance similarity.48Table2IntercorrelationsAmongSLSandDLSPerceptions(Time1)Sirn.DimensionMeanSDMm.Max.12345678910SLS1.Age4.912.1119.51**.41**.25**.25**.27**.29**•34**.40**.31**2.Appearance5.131.8719.62*.20*.30.50**•54**.51**•53**.64**3.Condition5.141.7818.23**.30*37**.50**39**.56**44**4.Gender6.652.5619.21**33**.13.21**.09.095.Ethnic5.452.2419.12.31**.26**.30**.26**DLS6.Attitude6.371.762945**54**7.Belief5.061.4928———————44**.64*8.Education5.511.5219———————.64**9.LifeExp.4.541.5218—————————10.Value5.611.5419Note:*p<.05,**p<.01Table3IntercorrelationsAmongSLSandDLSPerceptions(Time2)Sirn.DimensionMeanSDMm.Max.12345678910SLS1.Age4.982.0312.Appearance5.231.7713.Condition5.201.7214.Gender6.652.4815.Ethnic5.732.1419 8 9 9 9.48**.62**.38**.65**55**.67**73*•53**.70** .56**—.36**44**.13.24**.30**.18*33**37**.18**73**16*37**52**53**47**55**59**.15*.38**.46.50**44**.63**47**————.17*.30**.14.25**.04.18•33**44**.36**53**.38**DLS6.Attitude6.131.5717.Belief5.091.4918.Education5.521.5519.LifeExp.4.791.59110.Value5.521.561Note:*p<05,**p<019 8 9 8 9The second criterion that these two potential factors were required to satisfSr wasthat a theoretical or conceptual rationale could be given for their construction Attentionis first turned to the proposed surface-level factor, which included perceptions of age,physical appearance, and physical conditioning similarity. A distinguishing feature of thethree perceptions proposed for inclusion in this factor is that these three perceptions relateto the functional ability of exercise group members. ‘Functional ability’ is, of course,context dependent, and in the current discussion, functional ability refers to the capacityof group members to perform the exercises prescribed within that setting.The consideration of a factor comprised of perceived age, physical appearance,and physical condition similarity also allows for a more select focus on the effects ofperceived ethnicity and gender similarity. This select focus is consistent with pastresearch (e.g., Graves & Powell, 1995; Hogg et al., 2004; Turban et al., 2002; Tsui et al.,1995; Tsui, Porter, & Egan, 2002; Zellmer-Bruhn et al., 2008) and may be justified due tothe particular salience of these two attributes when forming impressions of others(Riordan & Shore, 1997; Brickson, 2000; Stangor, Lynch, Duan, & Clas, 1992). Indeedas Randel and Earley (2009) state, “gender and race are the characteristics most relied onin forming perceptions of others”(p.808). As a result of the seeming importance ofgender and ethnicity when categorizing others, and the rationale provided for a surfacelevel factor composed of age, appearance, and condition similarity (hereafter referred toas the ‘physical functionality factor’), dividing the surface-level perceptions into a threeitem factor and two single-item measures was deemed justifiable.With respect to the DLS factor, all five items were conceptualized as collectivelyreflecting each participant’s belief that he or she is similar to other members of in terms51of deep-level characteristics. These items were not conceived with a multidimensionalconceptualization in mind (i.e., reflecting different facets of DLS), and indeed thisunidimensional conceptualization of DLS is consistent with approaches adopted inprevious diversity research (e.g., Harrison et al., 2002; Turban et al., 2002). In short, thecreation of a DLS factor was deemed to be theoretically justified.Potential factors were analyzed using SPSS Factor, and were extracted via thegeneralized least squares method. This method of extraction was chosen due to the factthat it weights variables based on their importance to the solution. This importance isdetermined based on the amount of shared variance accounted for by each variable(Tabachnick & Fidell, 2007). Since the amount of shared, as opposed to total, varianceaccounted for was of primary interest, this method of extraction was deemed justified.The eigenvalue-greater-than-one criteria (see Tabachnick & Fidell, 2007), Scree plot,total amount of common variance accounted for, and factor loadings were consulted asindices of each model’s fit. When analyzing the deep-level perceptual data from the firstdata collection period, one factor was clearly identified through both the eigenvaluegreater-than-one criteria and Scree test. This factor accounted for 51.65% of the commonvariance among the five deep-level items and each item was found to load onto the factor(Tabachnick & Fidell, 2007; factor loadings ranged from .61 to .87). The structure of thisDLS factor remained comparable during the final data collection period. Once againemploying the eigenvalue-greater-than-one criteria, a single factor was clearly identifiedduring this time period. This was confirmed after consulting this DLS factor’s Scree plot.This factor accounted for 57.59% of the common variance among the five deep-levelitems and each of these items was found to load onto the factor (factor loadings ranged52from .66 to .89). This potential factor was determined to have an acceptable level of fit atboth time points as a result of the criteria explored.Next, the physical functionality factor was explored, using SPSS Factor oncemore. Within the data collected during the initial time period one factor was clearlyidentified following the eigenvalue-greater-than-one criteria and the Scree test. Thisfactor accounted for 53.40% of the common variance among the three surface-level itemsand each of these items was found to load onto the factor (factor loadings ranged from .59to .85). This pattern was replicated in the data collected during the last time period. Onceagain, one factor was clearly identified following the eigenvalue-greater-than-one criteriaand consultation of the model’s Scree plot. This factor accounted for 57.20% of thecommon variance among the three surface-level items, with each item loading on the apriori factor (factor loadings ranged from .51 to .90). Given these indices of fit at bothtime points, this unidimensional operationalization of ‘physical functionality’ wasdeemed acceptable.To satisfy the final criteria, the scale reliabilities of each factor at each time pointwere explored using SPSS Reliability Analysis. Among the data collected during the firsttime point, the DLS factor and the surface-level functionality factor were both found tohave an acceptable level of reliability (Cronbach’s Alpha = .83 and .76 respectively).This trend continued during the last data collection period with the DLS factor(Cronbach’s Alpha = .87) and the surface-level functionality factor (Cronbach’s Alpha =.76) each demonstrating an acceptable level of reliability.533.3 Individual-Level, or Group-Level Analyses?An issue that requires addressing is whether the data should be analyzed at theindividual-level or group-level. By analyzing the data at the individual-level, eachparticipant’s responses are free to vary within each group and direct (although not causal)links between individual-level independent and dependent measures can be made.However, given that these data were collected within groups, the choice to analyze thesedata at the individual-level will ultimately, and perhaps unavoidably, be accompanied byquestions regarding the appropriateness of treating ‘nested’ data as independent. Incomparison, a group-level analysis would address the nested nature of the data. However,it would also restrict the variability of responses within each group and may obscure therelationship between the individual-level independent and outcome measures (Bickle,2007; Cohen, Cohen, West, & Aiken, 2003).Given that each type of analysis has its associated benefits and drawbacks, it isperhaps unsurprising that debate regarding whether the individual or the group should bethe unit of analysis has occurred for some time (e.g., Allport, 1924). Carron and Spink(1995) suggest that this debate has been sustained in part because no definite answer as towhich level of analysis should be preferred exists. Instead, these researchers suggest thatthe determination of the appropriate level of analysis “depends on the nature of thequestion” (p. 91) posed.In the current study the research questions primarily corresponded to individualperceptions (e.g., perceived similarity) and individual-level outcomes (e.g., adherencebehaviours). Since individual perceptions and individual behaviours were the criticalconsideration in this study, analysis at the individual-level was deemed justified. This54rationale is consistent with that of Carron and Spink (1995) as well past researchexploring cohesion and adherence in exercise groups (e.g., Anessi, 1999; Carron &Spink, 1993; 1995; Shapcott et al., 2006; Estabrooks & Carron, l999a,b) and perceptionsof similarity in work groups (e.g., Crutchfeild, Spake, D’Souza, & Morgan, 2003;Hobman et a!., 2003; Piasentin & Chapman, 2007).3.4 Consideration of Group SizeIn their recent article, Randel and Earley (2009) suggest that the size of the groupor team may influence the salience of team members’ diversity characteristics.Specifically, these researchers state that in large groups or teams it may be harder toarrive at a perception regarding intra-group similarity. This is due to the fact that it maybe more difficult for members to identify group-composition in larger groups. It followsthat the possibility accounting for group size in the subsequent analyses is worthy ofserious consideration.Group size may indeed have an effect on perceptions of similarity, cohesion, andadherence behaviours in exercise groups. However it would be imprudent to blindlyincorporate group size as a control variable in all subsequent analyses before firstunpacking some of the terms described above. In order to assess the appropriateness ofaccounting for group size when analyzing the present data ‘small’, ‘medium’, and ‘large’groups were first defined. The proportion of groups within the current data set that fellwithin each of these three classifications was then determined and a decision regardingthe inclusion of a group size variable was made.Carron and Spink (1995) define ‘small’ groups as those groups with less than 20participants. In addition, Carron et a!., (1990) define ‘large’ groups as those groups55ranging between 32 and 46 members in size. It follows that ‘medium’ classes may beconsidered to be those classes ranging between 20 and 31 participants in size. Turning tothe data on hand, 44 of the 46 groups (95.65% of all groups; 89.64% of all individual-level data) were ‘small’ in size. Given the relatively small proportion of groups that werenot ‘small’ in size it was determined unnecessary to control for group size in thesubsequent analyses. This approach was consistent with past research exploring cohesionand adherence behaviours among exercise group participants (e.g., Anessi, 1999; Carron& Spink, 1993; 1995; Shapcott et al., 2006; Estabrooks & Carron, 1999a,b).3.5 Global Similarity and AdherenceIn spite ofthe fact that members’ perceptions regarding their degree of globalsimilarity with the other group members were collected during both time points, only datafrom time 1 were analysed in relation to the adherence measures employed in this study.Consistent with person perception theory (Allport, 1954) and attribution theory (Shaver,1975), initial perceptions are considered to influence future behaviours (in this case,adherence behaviours). That is, global similarity was hypothesized to be a theoreticalantecedent of adherence behaviours. In addition, initial perceptions are known to be quite“tenacious” (Zellmer-Bruhn et al., 2008,p.46) as these perceptions have a much strongerinfluence on subsequent behaviours than later perceptions, even if later perceptions runcounter to those made earlier (Zellmer-Bruhn et al., 2008). It follows that a focus on therelationship between initial perceptions of global similarity and adherence within thecurrent study is justified on both theoretical and empirical grounds.A correlation matrix including OPS, PPS (both calculated from data derived fromtime 1), and the two measures of adherence was then created (see Table 4). As is evident56from this table, PPS correlated with the instructor-mediated, r = .21p= .007, and selfreport adherence data, r = .31,p<.001. In comparison, OPS did not correlatesignificantly with either measure of adherence.57Table4IntercorrelationsAmongGlobalMeasuresofPerceptualSimilarityandAdherenceMeanSDMm.Max.12341.OverallPerceptionofSimilarity(OPS)5.421.6619—.52**.09.142.PerceivedProportionalSimilarity(PPS)40.65%27.18%01.0—.21*3.Instructor-MediatedAdherenceData74.75%20.90%01.0———4.Self-ReportedAdherenceData86.15%13.06%.431.0————Note:*p<.05,**p<.0100When the self-reported adherence data were regressed onto these two globalmeasures of perceived similarity a significant equation resulted, Adj. R2 = .09, F(2,124) =6.78,p=.002. However, only PPS contributed significantly to this equation, ,8 = .318, t == .001. A similar result was obtained when the instructor-mediated adherencemeasures were regressed onto these two global measures of perceived similarity. Thisequation was found to account for a significant portion of the variability in the instructor-mediated data, Adj. R2 = .04, F(2,169) 4..07,p = .019. Once again, only PPS contributedsignificantly to this equation, /3= .231, t = 2.70,p= .008. The significant regressionequations provided support for Hypothesis 1. However, the non-significant contributionof OPS to these equations ran counter to the predictions made. Thus, partial support forHypothesis 1 was attained.3.6 Surface- and Deep-Level Perceptual Similarity and CohesionTime 1. Table 5 lists the descriptive statistics for, and the bivariate correlationsamong, perceptions of similarity and cohesion at time 1. Data from scores derived fromthe four cohesion subscales were each regressed onto the gender, ethnicity, physicalfunctionality, and DLS variables. When ATG-T was regressed onto these similarityvariables the equation was significant, Adj. R2 = .12, F(4,258)=lO.28,p < .001, with theperception of gender similarity,/3= .18, t = 2.96,p= .003, and the DLS factor, /3= .260, t= 3.45, p = .00 1, contributing significantly to this equation. ATG-S was then regressedonto the deep-level and three SLS variables. This equation accounted for a significantportion of the variability in ATG-S, Adj. R2 = .09, F(4,258)=‘7.O9,p < .001, with thephysical functionality factor contributing significantly to this equation, /1 = .278, t = 3.54,p<.001. Next, scores derived from the GI-T subscale were regressed onto the same four59similarity variables. A significant result was once again found, Ad]. R2 = ,12, F(4,257) =<.001, with the DLS factor contributing significantly to this equation, /1 = .309, t= 4.03,p< .001. Finally, the GI-S factor was regressed onto the four similarity variables,and again this equation accounted for a significant portion of the variability in GI-S, R2 =.05, Adj. R2 = .03, F(4,255)=3.l’74.,p = .014, with the physical functionality factorsignificantly contributing to this equation,fi= .193, t = 2.38,p= .018. Hypothesis 2Aproposed that cohesion would be predicted by SLS at time 1. Three of the four elementsof cohesion were predicted by SLS variables at this time. Thus, partial support forHypothesis 2A was attained. In comparison, Hypothesis 2B proposed that cohesionwould not be predicted by DLS at time 1. Although two of the four cohesion elementswere not predicted by DLS at this time, the remaining two elements were. As a result,partial support for Hypothesis 2B was attained.60Table5IntercorrelationsAmongSLSandDLSPerceptionsandCohesion(Time1)MeanSDMm.Max.123456781.GenderSimilarity6.652.5619—.21**.29**.25**.25**.06.19**.062.EthnicSimilarity5.452.2419——34**•33**.23**.11.14*.15*3.PhysicalFunctionalityFactor5.011.6519———•57**.19**.29**.25**.24**4.DLSFactor5.621.4119————34**.30**39*115.ATG-T7.871.2319—————.24**.41**.056.ATG-S4.562.0119——————.63**.63**7.GI-T5.771.8119———————8.GI-S3.211.9119————————Note:*p<.05,**p<.01Time 2. Table 6 lists the descriptive statistics for, and the bivariate correlationsamong, perceptions of similarity and cohesion at time 2. Each of the four cohesion factorswere once again regressed onto the deep-level and three surface-level variables. WhenATG-T was regressed onto these four independent variables the equation was significant,Adj. R2 = .04, F(4,178)=2.98,p .020. However, none of the four independent variablescontributed significantly to this equation. The ATG-S factor was then regressed onto thesame four independent variables. This equation was also significant, Ad]. R2 = .22,F(4, 178) = 14.15,p< .001 with the physical functionality factor,fi= .361, t = 4.40,p <.001, and the DLS factor,fi= .205, t=2.39,p = .018, contributing significantly to itspredictive value. Next, GI-T was regressed onto the deep-level and three surface-levelvariables. This equation was significant, Adj. R2 = .25, F(4, 176) 16.15,p< .001, withthe physical functionality factor, Ji .296, t 3.65,p < .001, and the DLS factor,fi =.3 00, t = 3.53,p= .001, contributing significantly to its predictive value. Finally, the GI-Sfactor was regressed onto the four similarity variables. This equation was also significant,Adj. R2 = .13, F(4, 176) = 7.57,p< .001, with the physical functionality factorcontributing significantly to the predictive value of this equation,fi= .370, t 4..23,p <.001.62Table6IntercorrelationsAmongSLSandDLSPerceptionsandCohesion(Time2)MeanSDMm.Max.123456781.GenderSimilarity6.652.4819—.13.18**.23**.14-.02.08.032.EthnicSimilarity6.105.5719——.16*.28**.———.56**.18.41**•43**•37**4.DLSFactor5.511.3718.17*37**44**.22**5.ATG-T7.901.1119—————.19*.38**.15*6.ATG-S4.251.9319——————.62**74**7.GI-T5.951.7919———————57**8.GI-S3.371.9819————————Note:*p<05,**p<01Hypothesis 3A proposed that cohesion would no longer be predicted by SLS attime 2. This hypothesis was not supported as all four of the cohesion elements werepredicted by SLS during this time period. In contrast, Hypothesis 3B proposed thatcohesion would be predicted by DLS at time 2. Two of the four elements of cohesionwere predicted by DLS at this time. Thus, partial support for Hypothesis 3B was attained.3.7 Cohesion and AdherenceTable 7 lists the descriptive statistics for, and the bivariate correlations among, thecohesion and adherence variables for both time points. Although measures of cohesionwere administered repeatedly (i.e., during each data collection period; see methodssection) no predictions regarding these later-administered measures of cohesion andadherence were made. This is because (in a similar manner to initial perceptions of globalsimilarity) cohesion is understood to be a theoretical antecedent of many outcomevariables, including adherence. In accordance with this conceptual understanding,researchers have traditionally focused on measures of cohesion that (temporally) precedeoutcome measures (e.g., Carron & Spink, 1993; Carron et al., 1988; Estabrooks &Carron, 2000; Spink & Carron, 1992). In an attempt to ensure that the results of thecurrent study would be comparable to relevant past research, no attention was given tothe relationship between perceptions of cohesion and past adherence behaviours (i.e.,relating perceptions of cohesion at the end of a course/program to the adherence in theprogram) in this study.64Table7IntercorrelationsAmongCohesionandAdherenceMeanSDMm.Max.1234561.ATG-T7.871.2319.24**.41**.05.15*.062.ATG-S4.562.0119.61**.63**.05.103.GI-T5.771.8119———.46**.07-.014.GI-S3.211.9119————.07.075.Instructor-MediatedAdherence74.75%20.90%01.0—————6.Self-ReportedAdherence86.15%13.06%.431.0——————Note:*p<.05,**p<.01The two measures of adherence employed in this study were each regressed ontothe data derived from the four cohesion subscales. When the instructor-mediatedadherence data were regressed onto these four variables a non-significant equationresulted, Adj. R2 = .02, F(4,205)=l.9l,p = .111. A similar (i.e., a non-significant) resultwas found when the self-reported adherence data were regressed onto these four cohesionvariables, Adj. R2 = .003, F(4,138)=l.09,p = .363. Given these non-significantregression equations, Hypothesis 4 (linking cohesion to adherence) was rejected.3.8 Test of MediationIn order to test for mediation a significant relationship between the independentand dependent variable must first be established. Baron and Kenny (1986) state that avariable functions as a mediator when it meets the following three conditions: first,variations in the presumed mediator (cohesion) are accounted for by variations in theindependent variable (similarity; see path ‘a’ in Figure 2). Second, variations in themediator variable are statistically associated with variations in the dependent variable(adherence; see path ‘b’ in Figure 2). Finally, when paths ‘a’ and ‘b’ are held constant therelationship between the independent and dependent variable becomes non-significant(path ‘c’ in Figure 2). In cases where the relationship between the independent anddependent variable is reduced but does not become non-significant this is taken asevidence of partial mediation, as other mediating factors may also exist.Since neither equation in which a measure of adherence was regressed onto thefour cohesion factors was significant, all hypothesized mediation paths were precluded.As a result Hypothesis 5 was rejected.663.9 Assumption of Deep-level Similarity and CohesionIn order to explore the relationship between initial perceptions of SLS and DLS acorrelation matrix composed of the deep-level and three surface-level variables wascreated (see Table 8). As a result of the prediction that initial perceptions of SLS andDLS would be correlated, data from the first data collection period were used in thesubsequent analysis. Specifically, the DLS factor correlated significantly with thephysical functionality factor, r=.5’7,p < .001, as well as the perceptions of gender, r =<.001, and ethnicity, r=.13,p < .001, similarity.67Table8IntercorrelationsAmongSLSandDLSPerceptions(Time1)SimilarityDimensionMeanSDMm.Max.12341.Gender6.652.5619—.21**.29**2.Ethnic5.542.2419——34**33**3.PhysicalFunctionalityFactor5.011.6519——57**4.DLSFactor5.621.4119———Note:*p<.05,p<.01C.’ 00The DLS factor was then regressed onto these three surface-level perceptions.This equation was significant, Ad]. R2 .42, F(3 ,262) = 64.71,p< .001, with the physicalfunctionality factor,fl= .587, t= ll.54,p <.001, and ethnicity similarity,fl= .119, t2.39,p= .018, contributing significantly to its predictive value. This significant equationwas taken as support of Hypothesis 6.Consistent with past research exploring the development of perceptions acrossmultiple time points (e.g., Polzer, Milton, & Swann, 2002), a change score variable wascreated by subtracting each participant’s DLS factor value for the first data collectionperiod from their reported value during the last data collection period. Thus, a positivevalue on this new variable signified an increase in one’s perception of the degree of DESbetween oneself and the rest of the group whereas a negative value signified a decrease inone’s perception of the degree of this similarity.A correlation matrix was then created to explore the relationship between changesin perceptions of DLS and the cohesion variables reported during the last data collectionperiod (see Table 9). Changes in one’s score on the DLS factor correlated significantlywith ATG-S , r = .18,p= .040, and GI-S, r .24,p.005. In contrast, this DLS changescore did not correlate significantly with ATG-T, r = .002,p= .977, or GI-T, r .024,p= .775. When the DES change score was regressed onto these four elements of cohesion asignificant equation resulted, Ad]. R2 .05, F(4,134) 4.56,p.03 8, with GI-Scontributing significantly to this equation, ,8 = .3 17, t = 2.16,p= .033. Given thissignificant equation and the mixture of significant and non-significant intercorrelations,partial support for Hypothesis 7 was attained.69Table9IntercorrelationsAmongDLSFactorChangeScoreandCohesionMeanSDMm.Max.123451.ChangeinDeep-LevelFactor-.121.35-4.404.40—.002.18*.0242.ATG-T(week8)7.901.1119——.19*.38**15*3.ATG-S(week8)4.251.9419———.61**73**4.GI-T(week8)5.951.7919————57**5.GI-S(week8)3.371.9819—————Note:*p<05,**p<014 DISCUSSIONThis section begins with a discussion of each of the hypotheses tested in thisstudy. This is followed by a general discussion of the thesis findings. During this generaldiscussion the broad objectives and aims of this thesis are revisited, limitations areacknowledged, and suggestions for future researchers are made.4.1 Hypothesis 1Exercise psychologists place a great deal of importance on the notion of programadherence (see Carron et al., 1996; Dishman & Buckworth, 1996). Indeed, the success ofa given program is often judged based on its reported level of adherence. Within thefields of social and organizational psychology a literature has emerged suggesting thatsimilar group members are more likely to remain a part of the group than comparablydissimilar members (e.g., Jackson et al., 1991; Milliken & Martins, 1996; Tsui et al.,1992). Combining (a) the importance of adherence within exercise settings with (b) therelevant work on similarity and sustained group membership conducted in the domains ofsocial and organizational psychology, the primary hypothesis examined in this study wasthat a participant’s perceived level of similarity between him or herself and the othergroup members would positively relate to that participant’s level of program adherence.This primary hypothesis was tested by relating the two ‘global’ measures ofperceived similarity (i.e., OPS and PPS variables) to measures of adherence. The OPS(overall perception of similarity) measure simply asked each respondent to rate howsimilar they felt they were to the other members of the group. This question was intendedto represent an overall assessment of the degree of perceived similarity between theparticipant and all other group members. In comparison, the second measure of global71similarity required participants to report the number of members they felt they were verysimilar to as well as the perceived size of their group. These two values (number of verysimilar others, reported group size) were used to calculate the PPS (perceivedproportional similarity) variable.In this study, the PPS variable was found to correlate significantly with theinstructor-mediated, and self-reported, adherence data. In comparison, the correlationsbetween OPS and these two measures of adherence did not reach the level of statisticalsignificance. In addition, when the instructor-mediated adherence data were regressedonto OPS and PPS only the latter independent variable significantly contributed to thepredictive value of the resulting equation. A similar pattern emerged when the self-reported adherence data were regressed onto OPS and PPS. Once again, only the PPSvariable significantly contributed to this equation. Thus, the PPS variable was a muchstronger predictor of adherence behaviours when compared to the OPS variable. Sinceone of the two global measures of perceived similarity (PPS) predicted adherence, partialsupport for the primary hypothesis proposed in this study was attained.The importance of considering the proportion of similar others in a group (asopposed to the overall level of perceived similarity) accords with a phenomenonhighlighted by Riordan and Shore (1997). These researchers brought attention to the factthat “research has indicated that as the proportion of individuals who possess a particularcharacteristic (e.g., female) grows smaller, people who possess the minority characteristicwill become increasingly self-aware of their social identity” (p. 343). It follows thatmembers who perceive a low proportion of similar others in the group may be markedlymore aware of their dissimilarity than members who perceive a medium or high degree of72PPS within their group. Consistent with the categorization perspective of diversity, thislack of identification with a significant portion of one’s group members may bedetrimental to adherence behaviours. That is, there may be a causal relationship betweenone’s level of PPS and adherence behaviours within group-based exercise settings. Thisstatement is made cautiously (i.e., may be as opposed to is) given the observational (i.e.,non-experimental) nature of the study design.4.2 Hypotheses 2 and 3The secondary aim of this study was to apply the taxonomy of intra-groupsimilarity proposed by Harrison et al., (1998) to the study of cohesion among exercisegroups. As discussed in the introduction, Harrison and colleagues conducted a study ofthe relationships between SLS, DLS, cohesion, and time in work groups. Theseresearchers reported that the degree of SES within groups (based on the observable traitsand attributes of group members) had a strong, positive relationship with cohesioninitially following group formation. As time progressed, the relationship between SESand cohesion decreased to the level of non-significance. The opposite was true for thegroup’s degree of DES (based on the non-observable or psychological traits and attributesof the group members; Harrison & Sin, 2005): initially, the group’s degree of DES had anon-significant relationship with cohesion. However, over time the level of DES withinthe group was found to positively relate to the group’s cohesion at an increasingly strongdegree.It was hypothesized that the relationships between SLS, DES, cohesion, and timeobserved in work groups by Harrison et al. (1998) would also be present among exercisegroups. In other words, it was hypothesized that perceptions of SLS would positively73relate to cohesion initially (Hypothesis 2A) and only initially (Hypothesis 3A) and thatperceptions of DLS would positively relate to cohesion during later data collectionperiods (Hypothesis 3B) and only during these later data collection periods (Hypothesis2B) among exercise group members.As highlighted in the results section, three of the four cohesion elements werepredicted by SLS at time 1. Also, two of the four cohesion elements were predicted byDLS at time 2. Thus, partial support for the relationships between SLS, DLS, cohesion,and time proposed in Hypotheses 2 and 3 was attained. These results speak to theimportance of considering perceptions of the surface-level and deep-level composition ofthe group and time when attempting to explain the level of cohesion present in a givenexercise class (and at a given time). Although support for the proposed relationships wasevident, this support was not absolute. Two findings observed ran contrary to thepredictions made in Hypotheses 2 and 3. These findings are discussed below.During time 1 (and contrary to Hypothesis 2B), two of the four elements ofcohesion were predicted by perceptions DLS. This unpredicted finding may be a result ofthe congruence assumption (see Hypothesis 6). Specifically, members may have instantlydeveloped a firm perception of the deep-level composition of the group (and acomparably strong perception of the degree of DLS in the group) based on the surfacelevel composition of the group. This perception (or, rather the similarity of perceptionsregarding the level of SES and DES in the group) may have influenced the resultsreported during time 1.The second unpredicted finding pertains to the data collected during time 2.During this time period (and contrary to Hypothesis 3A), all four of the cohesion74elements were predicted by a SLS variable. A possible explanation as to why this DLSvariable was not the primary predictor of cohesion at time 2 can be gleaned by returningto the Harrison et al. (1998) paper. It is important to note that, although ‘time’ wasmeasured by these researchers in this study, the theoretical variable of interest for theseresearchers was ‘information transmission’. As Harrison et al. (1998) stated,“Although time is the variable we examined, the fundamental medium isinformation. Demographic factors are often a poor surrogate for the deeper-levelinformation people need to make accurate judgments about similarity of attitudes amonggroup members. Time merely allows more information to be conveyed. Indeed, it mightbe more appropriate to think of the richness of interactions as the conduit of informationexchange” (p. 104).Indeed, if the participants in this study did not have sufficient time to interact witheach other accurate insight into the level of DLS in the group would be quite limited.Following Harrison et al.’s rationale for why DLS predicts cohesion (as a result of theacquisition of accurate knowledge relating to the deep-level composition of the group)this lack of information acquisition, due to the relatively short nature of the exercisecourses involved in this study, may have hindered the predictive power of DLS.In retrospect, perhaps eight weeks was not long enough to record the potentialrelationships between SLS, DLS, cohesion, and time. At the end of this eight week term,members may have had yet to discover the actual deep-level composition of the group.Although, studies that have look at this relationship in the workplace have employed asimilar time frame (e.g., Harrison et al., 1998), the groups explored in these studies havetypically met much more frequently than once a week. It remains possible that, if the75length of the current study was increased, with courses longer in duration, DLS wouldhave eventually usurped the predictive value of SLS.4.3 Hypothesis 4Due to the relationship between cohesion and adherence within exercise groupsreported in previous research (e.g., Annesi, 1999; Estabrooks & Carron, 1999a,b; 2000) itwas hypothesized that cohesion would positively relate to adherence behaviours in thissample. Although one of the cohesion factors (ATG-T) significantly correlated with theinstructor mediated adherence data, when this adherence data were regressed on all fourcohesion dimensions the resulting equation was non-significant. A similar result wasfound when the self-reported adherence data were regressed onto these four elements ofcohesion. That is, this equation was non-significant. Due to the non-significance of thesetwo equations, Hypothesis 4 was rejected.4.4 Hypothesis 5As Barron and Kenny (1986) state, one of the requisites for mediation is that“variations in the mediator significantly account for variations in the dependent variable”(p. 1176). Since, the hypothesized mediator (cohesion) failed to account for variations inthe dependent variable, Hypothesis 5 was rejected. The rejection of Hypotheses 4 and 5runs contrary to much of the published work exploring the relationship between cohesionand adherence among exercise group members (e.g. Anessi, 1999; Carron, et al., 2007;Carron et al., 1988; Estabrooks & Carron, 2000). This discrepancy is addressed in thegeneral discussion.764.5 Hypotheses 6 and 7People have a tendency to infer a group’s deep-level composition based on thesurface-level attributes of the group members when this deep-level information isunknown (Harrison et al., 1998; Phillips & Loyd, 2006). As a result, it was hypothesizedthat this assumption of congruence between SLS and DLS would be evident during thefirst data collection period (i.e., before a substantial amount of information regarding thedeep-level composition of the group would likely be known). At time 1, perceptions ofDLS were found to correlate (positively) with all three measures of SLS. In addition,when this omnibus perception of DLS was regressed onto these three SLS variables theequation was significant. Thus, full support of Hypothesis 6 was attained.The final hypothesis proposed in this study was that changes in perceptions ofDLS would be related to the scores of the four cohesion subscales collected during thelast data collection period. Partial support for this hypothesis was attained as cohesionaccounted for a significant portion of the variance in the DLS change score. Specifically,change in the perceived level of DLS was found to relate to scores derived from the twosocial cohesion subscales at time 2. However, only partial support can be given for thishypothesis as the DES change score did not relate to the associated scores derived fromthe two task cohesion subscales.It appears as though there is a relationship between changes in the perceived levelof DLS and cohesion (or certain elements of cohesion). However, this relationship is farfrom uniform. As mentioned in the introduction, realizing that one has overestimated thedegree of DES in the group is theorized to greatly increase the likelihood that one will77withdrawal from the group (Chen & Kenrick, 2002). The relationship between changes inDLS, cohesion, and withdrawal represents an appropriate focus of future study.4.6 General DiscussionThis study contributes to the existing literature in four ways. First, it reinforcesthe categorization perspective on intra-group similarity. Second, it provides addedsupport to the increasingly prevalent argument that diversity researchers should considerperceptual (or ‘subjective’) diversity in more depth. Third, it identifies a possibleantecedent of cohesion among exercise group participants. Finally, it contributes to theexercise psychology literature by linking perceptions of similarity to adherencebehaviours while simultaneously calling into question the appropriateness of exercisepsychologists’ continued emphasis on group cohesion. Each of these contributions arediscussed below.4.6.1 Is Intra-Group Similarity Beneficial to Member Functioning inExercise Groups?In this study the “primary thesis” (Harrison et al., 1998,p.96) proposed bydiversity researchers was tested within the population of exercise group members. In itssimplest and most direct form, this thesis purports that intra-group similarity is beneficialto group functioning. This proposition is largely informed by self-categorization theory(Turner, 1984; 1985; 1987) and the similarity-attraction hypothesis (Byrne, 1971). Thesetheories indicate that (a) we categorize or classify individuals as similar or dissimilarbased on salient qualities or attributes (e.g., gender, ethnicity, extraversion) and (b)similarity is attractive and dissimilarity repels. Taken together, this suggests that peopleshould be attracted to groups with similar others (or similar others within a group) and78driven away by groups with dissimilar others (or dissimilar others within the group). As aresult, this perspective has been referred to as the categorization perspective on diversity.In accordance with this perspective, the central thesis proposed in this study wasthat intra-group similarity would benefit member functioning within exercise groups.Functioning was operationalized based on (a) each participant’s adherence to the group-based exercise program and (b) the level of group cohesion reported by each participant.Hypotheses 1, 2A, 2B, 3A, and 3B, may be viewed as a test of this thesis as eachof these hypotheses proposed a link between perceptions of similarity and measures offunctioning (adherence and cohesion). When testing each of these hypotheses, positiverelationships between the degree of perceive similarity (whether global, surface-, or deep-level in nature) and the variables used to assess member functioning was found. This isevident as a result of the significant regression equations reported when testing the abovehypotheses. Given these results, general support for the central thesis of this study wasattained. It appears that subjective intra-group similarity is beneficial to the functioningof members within exercise groups.These results provide further support for the categorization perspective of intragroup similarity. As a result, the appropriateness of the application of the informationprocessing model (characterized by the belief that intra-group similarity is detrimental togroup functioning) to exercise groups must be questioned. None of the results reported inthis study provide support for the beliefthat disparate individuals will function at a highercapacity than more homogeneous members. This is not to say that the informationprocessing model is erroneous. Rather the applicability of each model seems inextricablylinked to the functional goals of the types of groups studied.79Diverse groups have been found to outperform more homogeneous groups ontasks that require a high level of creativity or imagination as well as tasks that require theconsideration of multiple perspectives (Dose, 1999). None of these processes appearparticularly applicable to exercise groups. Instead, processes such as cohesion, socialintegration, commitment to the group, enjoyment, and adherence seem much morerelevant. Indeed, a focus on cohesion and adherence is quite common among researchersstudying exercise groups (e.g., Anessi, 1999; Carron, et al., 2007; Carron et al., 1988;Estabrooks & Carron, 2000). Although limited research has explored the relationshipsbetween intra-group similarity, cohesion, and adherence, the vast majority of empiricalwork within social and organizational psychology exploring this relationship accordswith the categorization perspective of intra-group similarity (e.g., Knippenberg et al.,2007; Jackson et al., 1991; Milliken & Martins, 1996). In addition, previous work thathas explored the relationship between intra-group similarity and member functioning inexercise groups (i.e., Shapcott et al.’s 2006 study) is also consistent with this perspective.Given the theorized relationships between perceptions of intra-group similarity, cohesion,and adherence it may be concluded that the categorization perspective of intra-groupsimilarity is more applicable to exercise groups than the information processing model.4.6.2 The Importance of Considering Subjective Intra-Group SimilarityMontoya et al. (2008) recently published a meta-analysis exploring therelationships between perceived similarity, actual similarity, and attraction. In this paper,the studies reviewed were classified in one of three ways based on the duration of contactbetween the perceiver and the perceived. Studies classified as being ‘non-interactional’involved asking participants to rate the degree to which they were attracted to a80hypothetical target based on information provided by the researcher. ‘Short interaction’studies were those that involved individuals interacting for 10 minutes or less — mostoften in a lab setting. Finally, ‘existing relationship’ studies were those that assessedattraction in pre-existing relationships.Montoya and colleagues found that both subjective and objective measures ofsimilarity significantly related to attraction among non-interactional studies. However,they reported that, “there was a significant reduction in the effect size of actual similaritybeyond no-interaction studies and the effect of actual similarity in existing relationshipswas not significant”. (p. 889). In comparison, perceptual measures of similarity werefound to significantly relate to attraction in both ‘short interaction’ and ‘existing’relationships.The meta-analysis performed by Montoya et al., (2008) is both timely andappropriate. For decades (e.g., Curry & Emerson, 1970; Orpen, 1984; Turban & Jones,1988), perceptual measures of similarity have been found to relate quite strongly to theoutcome measures considered. In fact, in the majority of cases in which both subjectiveand objective measures of similarity were employed, subjective measures tended to sharea stronger relationship with the outcome measures considered (see Chapdelaine, Kenny,& LaFontana, 1994; Curry & Emerson, 1970; Harrison & Sin, 2005; Riordan & Wayne,2008; Strauss et al., 2001). Perceptual similarity has also been found to positively relateto such diverse outcomes as interpersonal liking (Chapdelaine et a!., 1994), identificationwith fictional characters (Johnson, 1995; Jose & Brewer, 1984), and messageeffectiveness (Andsager, Bemker, Choi, & Torwel, 2006).81It is interesting to note that calls for researchers to consider subjective measuresof similarity (e.g., Hartel & Fujimoto, 1999; Harrison et al., 2002; Riordan, 2000;Zellmer-Bruhn et a!., 2008) have, to a large degree, fallen on deaf ears. This is reflectedin the number of effect sizes reported in Montoya et al.’s (2008) meta-analysiscorresponding to objective (N= 406) and subjective (N 54) measurements of similarity.In order for the field of diversity research to progress a greater emphasis must be placedon perceived similarity (Riordan & Wayne, 2008). After all, we function largely on thebasis of perceptions, not objective reality (Ferris & Judge, 1991).While an emphasis has been placed on perceptual measures of similarity in theproceeding discussion, it should be stressed that the abandonment of objective similaritymeasures is not being purported here. Indeed, it would be foolish to suggest that oneshould be concerned solely with perceptions of similarity as these perceptions are likelyinfluenced by the objective composition of the group. It has even been suggested thatsubjective intra-group similarity may mediate the relationship between actual intra-groupsimilarity and outcome measures (Harrison et al., 2002). Indeed, Ashforth and Mae!(1989) have argued that the substantive effect of objective intra-group similarity iscarried out through perceptions. Instead, a call is placed for future research to attempt tomore appropriately balance a focus on both subjective and objective measures of intragroup similarity. This balance, couple with a more through investigation of therelationship between these two types of intra-group similarity (subjective and objective)will likely contribute to a greater understanding of group and individual functioningwithin teams.4.6.3 Perceptual Intra-Group Similarity, Cohesion, and Adherence82This is the first known study to relate perceptions of intra-group similarity tomeasures of cohesion among exercise groups. Although correlation does not equate tocausation, the results provide preliminary support for the theoretical argument thatsimilarity is an antecedent of cohesion. Future studies should attempt to establish a causallink between these two constructs through the implementation of experimental studies.This is also the first study to date to link perceptions of similarity with adherencebehaviours in exercise classes. In this study the PPS variable was positively linked to theinstructor-mediated and self-reported measure of adherence. This observed relationshipmay be due to the fact that individuals tend to be attracted to similar others (Byrne,1971), and that a group with a high proportion of very similar others will be moreattractive (and easier to identifr with; Hogg, 1992; Turner, 1987) than a group with a lowproportion of very similar individuals. This explanation is consistent with thecategorization perspective of diversity.In light of the relationship between global similarity and adherence behavioursreported in this thesis, the next logical step is to try to identify which specific attributes ortraits within the group (such as age, gender, personal values) contribute to this globalperception. Consistent with the argument put forth in the previous section, this step maybe most effectively completed by considering both subjective and objective conceptionsof similarity within the group. With this information in hand, a potential causal linkbetween these similarities (both subjective and objective) and adherence could beexplored through experimental investigation.In contrast to past research (e.g., Anessi, 1999; Carron et a!., 2007; Canon et a!.,1988; Estabrooks & Canon, 2000), the scores derived from the PAGEQ did not83significantly predict adherence behaviours within the present sample. At the very least,these results call into question Estabrooks (2007) bold claim that cohesion should be a“fundamental”(p.143) consideration in physical activity interventions. Given the resultsof this study, it is likely that other contextual determinants within the exerciseenvironment (such as the degree of perceived similarity between each member and therest of the group) may be more salient for adherence behaviors when compared tocohesion. It is also likely that non-interpersonal elements of the course (such as the mode,intensity, and duration of the activities performed) play an important role in adherencebehaviors. These non-interpersonal elements of courses were not explored in the currentstudy. In future, researchers are encouraged to explore the relationship betweencontextual determinants of the environment, non-interpersonal elements of the course,and adherence in tandem toidentifSrthe factors that share the strongest relationship withadherence to exercise programs. This will likely lead to a more thorough understandingof the primary reasons for why people do, or do not, adhere to exercise programs.4.6.4 LimitationsIn spite of the contributions made by this thesis to the existing exercisepsychology and diversity literatures several inherent limitations must be recognized. Tobegin, one of the adherence measures employed was self-reported. Although people wereasked to recall the number of classes they missed (as opposed to the number theyattended), and this measure was found to correlate strongly with the instructor-mediatedmeasure (r = .63,p< .001), it remains possible that the level of adherence reported by theparticipants was positively biased (McAuley et a!., 2007). As a result, conclusions drawnfrom these data should be interpreted with caution. That being said, both adherence84measures (including the instructor-mediated data) were predicted from the measures ofperceptual similarity. It follows that this limitation is somewhat innocuous.As a result of the data used in this thesis, there are two potential concernsregarding common method variance. First, participants were asked to report theperceptions of similarity and cohesion on the same Likert-type scale (9 point scaleranging from ‘strongly disagree’ [1] to ‘strongly agree’ [9]) and at the same time. Sincethe same scale (i.e., the 9 point Likert-type scale) was employed on different measureswithin a single data collection period, the similarity of responses across these measuresmay have been inflated. This may have led to an inflation in the reported strength of therelationships between the SLS and DLS variables and measures derived from the PAGEQ(see Hypothesis 2A, 2B, 3A, 3B, & 7). Second, the potential existence of percept-perceptinflation — an inflation in the strength of relationship between self-report measures as aresult of their self-reported nature (Crampton & Wagner, 1994) - must also berecognized. Aside from the instructor-mediated adherence data, all variables consideredin this study were self-reported in nature.It also goes without saying that the demographics of the sample collected restrictthe generalizability and external validity of the findings. Future researchers shouldexplore whether the results reported in this study hold in populations that differ markedlyfrom the one studied in this thesis (e.g., children, older adults, European or Asiancultures).The less than ideal response rate (i.e., 77.59% and 71.49% for the first and lastdata collection period respectively) is another potential limitation that should be dulynoted. As Allen, Stanley, Williams, and Ross (2007) suggest, “the lower the response85rate, the greater the variability in the observed correlations. . . lower response rates make itmore difficult to estimate accurately a true correlation”(p.1423). Steps were taken toensure that the response rates of this study were consistent with past research (e.g.,O’Reilly, Caldwell, & Barnett, 1989; Pelled, Ledford, & Mohrman, 1999; Pelled, Xin, &Weiss., 2001). Indeed, data from one of the three collection periods was not analyzedpartially due to the low response rate during this period. However, it remains possiblethat the relationships reported here may not hold within the entire population of group-based exercisers from which the current sample was drawn. That being said, notachieving a response rate of 100% would be a much greater limitation if similarity wasquantified at the group level. If this were done, the calculation of a ‘group’ score wouldbe based on incomplete information (Allen et al., 2007).Nevertheless, it remains possible that there was a demographic pattern in theresponse/non-response of group members. Potentially, this pattern could have been basedon the degree of similarity between the potential respondent and the other members oftheir group. Allen and colleagues (2007) investigated whether dissimilarity in age relatedto a participants’ propensity to participate in research involving their group. As theseresearchers reported, “it appears that responding to the survey was linked to theemployee’s age in relation to others in the workgroup, with those who were moredifferent being less likely to respond” (p. 1425). As a result, Allen et al. (2007) suggestthat similar group members (as compared to non-similar group members) are more likelyto participate in group-based research. Incorporating these findings into the presentdiscussion, it remains possible that the relationships reported in this thesis may havefallen prey to a form of range restriction. It follows that the relationships reported in this86study may have actually been smaller than they were in reality. In future, researchers maywish to explore the characteristics of responders/non-responders in exercise groups toaddress this issue.4.6.5 Summary and ConclusionThe present study contributes to the existing literature in four ways. First, itreinforces the validity of the categorization model of intra-group similarity. This modelmay be particularly appropriate for exercise groups given the established relationshipbetween similarity and outcome measures such as cohesion (e.g., Back, 1951; Mullen &Cooper, 1994; Shaw & Shaw, 1962; Wiersema & Bird, 1993), adherence (e.g., Jackson etal., 1991; Milliken & Martins, 1996; Tsui et al., 1992) social integration (e.g., Ely, 2004;Riordan & Shore, 1997), satisfaction and motivation to remain part of the group (e.g., Ely1995; Jackson et al., 1991; O’Reilly et al., 1989; Wharton & Baron, 1987; Wiersema &Bird, 1993) within social and organizational psychology.Second, the results of this study add credence to the increasingly prevalent call fordiversity researchers to consider objective and subjective measures of intra-groupsimilarity. An exclusive focus on either subjective or objective intra-group similarity maypreclude the advancement of a complete understanding of the relationship between groupcomposition and outcome measures. As a field we must begin to look at these two typesof diversities in relation to one another if we wish to continue to advance this area ofresearch. Given the prevailing use of objective measures of diversity, the necessary firststep in this process may be to allot a greater focus to perceptual data. Ideally, thisincreased focus will ultimately lead researchers to consider both types of diversity whenconducting research.87Third, this is the first known study to explore some of the potential antecedents ofcohesion within exercise groups. The results suggest that perceived similarity andcohesion share a positive relationship within this population. In future, researchers maywish to explore whether a causal link can be drawn between similarity and cohesionamong exercise group participants through the implementation of experimental trials.Finally, this study is the first known to explicitly link perceptions of similarity toadherence behaviours in exercise classes. Given the importance of adherence within thiscontext (Carron et al., 1996; Dishman & Buckworth, 1996), researchers should continueto explore (a) the relationship between similarity perceptions and adherence behaviours(b) the qualities or attributes that contribute to perceptions of similarity and (c) therelationship between a group’s composition and outcome measures such as adherence.Another potentially fruitful area of future research would be to explore whether a causallink between perceptions of similarity and adherence behaviours exists within these typesof groups.88REFERENCESAlcock, J.E., Carment, D.W., & Sadava, S.W. (1998). A Textbook ofSocial Psychology:Fourth Edition. Scarborough, ON: Prentice-Hall Canada.Allen, N.J., Stanley, D.J., Williams, H., & Ross, S.J. (2007). Assessing dissimilarityrelations under missing data conditions: Evidence from computer simulations.Journal ofApplied Psychology, 92, 1414-1426.Allen, V.L. & Wilder, D.A. (1979). Group categorization and attribution of beliefsimilarity. Small Group Behavior, 10, 73-80.Allport, G.W. (1924). The group fallacy in relation to social science. The Journal ofAbnormal Psychology and Social Psychology, 19, 208-219.Allport, G.W. (1954). The Nature ofPrejudice. Reading, MA: Addison-Wesley.Andsager, J.L., Bemker, V., Choi, H., & Torwel, V. (2006). Perceived similarity ofexemplar traits and behavior: Effects of message evaluation. CommunicationResearch, 33, 3-18.Anessi, J.J. (1999). Effects of minimal group promotion on cohesion and exerciseadherence. Small Group Research, 30, 542-557.Ashforth, B.E., & Mael, F. (1989). Social identity theory and the organization. AcademyofManagement Review, 14, 20-3 9.Back, K.W. (1951). Influence through social communication. The Journal ofAbnormaland Social Psychology, 46, 9-23.Banikiotes, P.G., & Neimeyer, G.J. (1981). Construct importance and rating similarity asdeterminants of interpersonal attraction. British Journal ofSocial Psychology, 20,259-263.89Baron, R., & Kenny, D.A. (1986). The moderator-mediator variable distinction in socialpsychological research: Conceptual, strategic, and statistical considerations.Journal ofPersonality and Social Psychology, 51, 1173-1182.Barry, B., & Stewart, G.L. (1997). Composition, process, and performance in self-managed groups: The role of personality. Journal ofApplied Psychology, 82, 62-78.Beauchamp, M.R., Carron, A.V., McCutcheon, S., & Harper, 0. (2007). Older adults’preferences for exercising alone versus in groups: Considering contextualcongruence. Annals ofBehavioral Medicine, 33, 200-206.Bickle, R. (2007). Multilevel Analysisfor Applied Research: It’s Just Regression!Guilford Press.Bleda. P.R. (1974). Towards a clarification of the role of cognitive and affectiveprocesses in the similarity-attraction relationship. Journal ofPersonality andSocial Psychology, 29, 3 68-373.Bond, C.F., & Titus, L.J. (1983). Social facilitation: A meta-analysis of 241 studies.Psychological Bulletin, 94, 265-292.Booth, A.O., Nowson, C.A, & Matters, H. (2008). Evaluation of an interactive, Internetbased weight loss program: A pilot study. Health Education Research, 23, 371-381.Bovard, E.W. (1951). Group structure and perception. Journal ofAbnormal and SocialPsychology, 46, 3 89-405.Brickson, 5. (2000). The impact of identity orientation on individual and organizational90outcomes in demographically diverse settings. Academy ofManagement Review,25, 82-101.Burke, S.M., Carron, A.V., & Eyes, M.A. (2006). Physical activity context: Preference ofuniversity students. Psychology ofSport and Exercise, 7, 1-13.Burke, S.M., Carron, A.V., Patterson, M.M., Estabrooks, P.A., Hill, J.L., Loughead,T.M., Rosenkranz, S.R., Spink, K.S. (2005). Cohesion as shared beliefs inexercise classes. Small Group Research, 36, 267-288.Byrne, D. (1971). The Attraction Paradigm. New York, NY: Academic Press.Byrne, D., Baskett, G.D., & Hodges, L. (1971). Behavioral indicators of interpersonalattraction. Journal ofApplied Social Psychology, 1, 137-149.Byrne, D., & Rhamey, R. (1965). Magnitude of positive and negative reinforcements as adeterminant of attraction. Journal ofPersonality and Social Psychology, 2, 884-889.Carels, R.A., Konrad, K, Young, K.M., Darby, L.A., Coit, C., Clayton, A.M., & Oemig,C.K. (2008). Taking control of your personal eating and exercise environment: Aweight maintenance program. Eating Behaviors, 9, 228-237.Carron, A.V., Hausenblas, H.A:, & Eys, M.A. (2005). Group Dynamics in Sport: ThirdEdition. Morgantown, WV: Fitness Information technology.Carron, A.V., Hausenbias, H.A., & Mack, D.M. (1996). Social influence and exercise: Ameta-analysis. Journal ofSport and Exercise Psychology, 18, 1-16.Carron, A.V., Shapcott, K.M., Burke, S.M. (2007). Group cohesion in sport and exercise:91Past, present and future. In M.R. Beauchamp & M.A. Eys (eds.) Group Dynamicsin Exercise and Sport Psychology: Contemporary Themes. (pp. 117-139).Abingdon, Oxon: Routledge.Carron, A.V., & Spink, K.S. (1993). Team building in an exercise setting. The SportPsychologist, 7, 8-18.Carron, A.V., Widmeyer, W.M., & Brawley, L.R. (1988). Group cohesion and individualadherence to physical activity. Journal ofSport and Exercise Psychology, 10,127-138.Caspersen, C.J., Christenson, G.M., & Pollard, R.A. (1986). Status of the 1990 physicalfitness and exercise objectives — Evidence from NHIS 1985. Public HealthReports, 101, 587-582.Castellani, W., lanni, L., Ricca, V., Mannucci, E., & Rotella, C.M. (2003). Adherence tostructured physical exercise in overweight and obese subjects: A review ofpsychological models. Eating and Weight Disorders, 8, 1-11.Chapdelaine, A., Kenny, D.A., & LaFontana, K.M. (1994). Matchmaker, matchmaker,can you make me a match? Predicting liking between two unacquainted persons.Journal ofPersonality and Social Psychology, 67, 83-91.Chatman, J.A., & Flynn, F.J. (2001). The influence of demographic heterogeneity on theemergence and consequences of cooperative norms in work teams. Academy ofManagement Journal, 44, 956-974.Chen, F.F. & Kenrick, D.T. (2002). Repulsion or attraction? Group membership andassumed attitude similarity. Journal ofPersonality and Social Psychology, 83,111-125.92Cohen, J., Cohen, P., West, S.G., Aiken, L.S. (2003). Applied MultipleRegression/Correlation Analysisfor the Behavioral Sciences: 3’” Edition.Mahwah, NJ: Lawrence Eribaum Associates.Colquitt, J.A., Noe, R.A., Jackson, C.L. (2002). Justice in teams: Antecedents andconsequences of procedural justice climate. Personnel Psychology, 55, 83-109.Cox, T.H. (1993). Cultural Diversity in Organizations: Theory, Research, and Practice.San Francisco: Berrett-Koehler.Cox, T.H., & Blake, S. (1991). Managing cultural diversity: Implications fororganizational competitiveness. Academy ofManagement Executives, 5, 45-56.Crampton, S.M., & Wagner, J.A. (1994). Percept-percept inflation in microorganizationalresearch: An investigation of prevalence and effect. Journal ofAppliedPsychology, 79, 67-76.Crutchfield, T.N., Spake, D.F., D’Souza, G., & Morgan, R.M. (2003). “Birds of a featherflock together”: Strategic implications for advertising agencies. Journal ofAdvertising Research, 43, 361-369.Curry, T.J., & Emerson, R.M. (1970). Balance theory: A theory of interpersonalattraction? Sociometry, 33, 2 16-238.Davis, J.M. (1984). Attraction to the group as a function of attitude similarity andgeographic distance. Social behavior andpersonality, 12, 1-5.Deep, S.D., Bass, B.M., & Vaughn, J.A. (1967). Some effects on business gaming ofprevious quasi-t group situations. Journal ofApplied Psychology, 51, 426-431.Dimock, H. (1941). Rediscovering the Adolescent. New York, NY: Associated Press.Dishman, R.K. (1988). Overview. In R.K. Dishman (ed.), Exercise Adherence: fts93impact on public health (pp. 1-9). Champaign, IL: Human Kinetics.Dishman, R.K., & Buchworth, J. (1996). Increasing physical activity: A quantitativesynthesis. Medicine and Science in Sport and Exercise, 28, 706-719.Dose, J.J. (1999). The relationship between work values similarity and team-member andleader-member exchange relationships. Group Dynamics: Theory, Research, andPractice, 3, 20-32.Dunlop, W. L., & Beauchamp, M. R. (2008, November). Preferencesfor gender-integrated versus gender-segregated exercise classes across the age spectrum: Asocial identity theory perpective. Paper presented at the annual conference of theCanadian Society for Psychomotor Learning and Sports Psychology (SCAPPS),Canmore, Alberta, Canada.Earley, P.C., Mosakowski, E. (2000). Creating hybrid team cultures: An empirical test oftransnational team functioning. Academy ofManagement Journal, 43, 26-49.Elsass, P.M., & Graves, L.M. (1997). Demographic diversity in decision-making groups:The experiences of women and people of color. Academy ofManagement Review,22, 946-973.Ely, R.J. (1995). The power of demography: Women’s social constructions of genderidentity at work. Academy ofManagement Journal, 38, 589-634.Ely, R.J. (2004). A field study of group diversity, participation in diversity educationprograms, and performance. Journal ofOrganizational Behavior, 25, 755-780.Ensher, E.A., & Murphy, S.E. (1997). Effects of race, gender, perceived similarity, andcontact on mentor relationships. Journal of Vocational Behavior, 50, 460-481.Estabrooks, P.A. (2007). Group integration interventions in exercise: Theory, practice94and future directions. In M.R. Beauchamp & M.A. Eys (eds.), Group Dynamics inExercise and Sport Psychology (pp. 141-156). New York, NY: Routledge.Estabrooks, P.A., & Carron, A.V. (1999a). Group cohesion in older adult exercisers:Prediction and intervention effects. Journal ofBehavioral Medicine, 22, 575-588.Estabrooks, P.A., & Carron, A.V. (1999b). The influence of the group with elderlyexercisers. Small Group Research, 30, 438-452.Estabrooks, P.A., & Canon, A.V. (2000). The physical activity group environmentquestionnaire: An instrument for the assessment of cohesion in exercise classes.Group Dynamics: Theory, Research, and Practice, 4, 230-243.Ferris, G.R., & Judge, T.A. (1991). Personnel/human resources management: A politicalinfluence perspective. Journal ofManagement, 17, 447-488.Fiedler, F. E. (1967). A theory ofleadership effectiveness. New York: McGraw-Hill.Fiske, S.T. (2000). Interdependence and the reduction of prejudice. In S. Oskamp (ed.),Reducing Prejudice and Discrimination. (pp. 115-135). Mahwah, NJ: Eribaum.Fiske, S.T., & Taylor, S.E. (1991). Social Cognition: 2” Edition. New York, NY:McGraw-Hill.Floyd, F.J., & Widaman, K.F. (1995). Factor analysis in the analysis in the developmentand refinement of clinical assessment instruments. Psychological Assessment, 7,286-299.Graves, L.M., & Powell, G.N. (1995). The effect of sex similarity on recruiters’evaluations of actual applicants: A test of the similarity-attraction paradigm.Personnel Psychology, 48, 85-98.Guerin, B. (1993). Social Facilitation. New York, NY: Cambridge University Press.95Harrison, D.A., Price, K.H., & Bell, M.P. (1998). Beyond relational demography: Timeand the effect of surface- and deep-level diversity on work group cohesion.Academy ofManagement Journal, 41, 96-107.Harrison, D.A., Price, K.H., Gavin, J.H., & Florey, A.T. (2002). Time, teams, and taskperformance: Changing effects of surface- and deep-level diversity on groupfunctioning. Academy ofManagement Journal, 45, 1029-1045.Harrison, D.A., & Sin, H. (2005). What is diversity and how should it be measured? InA.M. Konrad, P. Prasad, & J.K. Pringle (eds.) Handbook of Workplace Diversity.(pp. 191- 216). Thousand Oaks, CA. Sage Publications Ltd.Hartel, C.E.J., & Fujimoto, Y. (1999). Explaining why Diversity Sometimes has PositiveEffects in Organizations and Sometimes has Negative Effects in Organizations:The Perceived Dissimilarity Openness Moderator Model. Paper presented at theannual meeting of the59thAcademy of Management Conference, Chicago.Harwood, C., & Beauchamp, M.R. (2007). Group functioning through optimalachievement goals. In M.R. Beauchamp & M.A. Eys (eds.) Group Dynamics inExercise and Sport Psychology: Contemporary Themes. (pp. 201-219). Abingdon,Oxon: Routledge.Heinzelmann, F., & Bagley, R.W. (1970). Response to physical activity programs andtheir effects on health behavior. Public Health Reports, 85, 905-911.Hensley, W.E. (1981). The effect of attire, location, and sex on aiding behavior: Asimilarity explanation. Journal ofNonverbal Behavior, 6, 3-11.Hobman, E.V., Bordia, P., & Gallois, C. (2003). Consequences of feeling dissimilar fromothers in a work team. Journal ofBusiness and Psychology, 17, 301- 325.96Hobman, E.V., Bordia, P., & Gallois, C. (2004). Perceived dissimilarity and work groupinvolvement: The moderating effects of group openness to diversity. Group &Organizational Management, 29, 560-587.Hogg, M.A. (1992). The Social Psychology ofGroup Cohesiveness: From Attraction toSocial Identity. England: Harvester, Wheatsheaf.Hogg, M.A. (1993). Group cohesiveness: A critical review and some new directions.European Review ofSocial Psychology, 4, 85-111.Hogg, M.A., Abrams, D., Otten, S., & Hinkle, S. (1994). The social identity perspective:Intergroup relations, self-conception, and small groups. Small Group Research,35, 246-276.Holtz, R. & Miller, N. (1985). Assumed similarity and opinion certainty. Journal ofPersonality and Social Psychology, 48, 890-898.Hong, S.Y., Hughes, S., & Prohaska, T. (2008). Factors affecting exercise attendance andcompletion in sedentary older adults: A meta-analytic approach. Journal ofPhysical Activity & Health, 5, 385-397.Husain, A., & Kureshi, A. (1983). Value similarity and friendship: A study ofinterpersonal attraction. Psychologia: An International Journal ofPsychology inthe Orient., 26, 167-174.Jackson, S.E., Brett, J.F., Sessa, V.L., Cooper, D.M., Julian, J.A., & Peyronnin, K.(1991). Some differences make a difference: Interpersonal dissimilarity and groupheterogeneity as correlates of recruitment, promotion, and turnover. Journal ofApplied Psychology, 76, 675-689.Johnson, K.K.P. (1995). Attributions about date rape: Impact of clothing, sex, money97spent, date type, and perceived similarity. Family and Consumer SciencesResearch Journal, 23, 292-310.Jose, P.E., & Brewer, W.F. (1984). Development of story liking: Character identification,suspense, and outcome resolution. Developmental Psychology, 20, 911-924.Kenrick, D.T., Neuberg, S.L., & Cialdini, R.B. (2005). Social Psychology: Unravelingthe Mystery:3rdEdition. Boston, MA: Pearson Education Ltd.King, A.C., Ahn, D.K., Oliveira, B.M., Atienza, A.A., Castro, C.M., Gardner, C.D.(2008). Promoting physical activity through hand-held computer technology.American Journal ofPreventative Medicine, 34, 138-142.Klein, M. & Christiansen, G. (1969). Group composition, group structure, and groupeffectiveness of basketball teams. In J.W. Loy & G.S. Kenyon (eds.), Sport,Culture, and Society (pp. 3 97-408). New York, NY: Macmillan.Klein, K.J., Conn, A.B., Smith, D.B., & Sorra, J.S. (2001). Is everyone in agreement? Anexploration of within-group agreement in employee perceptions of the workenvironment. Journal ofApplied Psychology, 86, 3-16.Klein, K.J., & Kozlowski, S.W. (2000). From micro to meso: Critical steps inconceptualizing and conducting multilevel research. Organizational ResearchMethods, 3, 211-236.Knippenberg, D.V., De Dreu, C.K.W., & Homan, A.C. (2004). Work group diversity andgroup performance: An integrative model and research agenda. Journal ofApplied Psychology, 89, 1008-1022.Knippenberg, D.V., Haslam, S.A., & Platlow, M.J. (2007). Unity through diversity:98Value-in-diversity beliefs, work group diversity, and group identification. GroupDynamics: Theory, Research, and Practice, 11, 207-222.Knippenberg, D.V., & Schippers, M.C. (2007). Work group diversity. Annual Review ofPsychology, 58, 5 15-41.Kramer, R.M. (1991). Intergroup relations and organization dilemmas: The role ofcategorization processes. Research in Organizational Behavior, 13, 19 1-228.Lawrence, B. S. (1997). The black box of organizational demography. OrganizationalScience, 8, 1-22.Lee, M., & Duck, S. (1982). A model for the role of similarity of values in friendshipdevelopment British Journal ofSocial Psychology, 21, 301-3 10.Levinger, G., & Breedlove, J. (1966). Interpersonal Attraction and agreement: A study ofmarriage partners. Journal ofPersonality and Social Psychology, 3, 367-372.Lincoln, J.R., & Miller, L. (1979). Work and Friendship ties in organizations : Acomparative analysis. Administrative Science Quarterly, 24, 181-199.Lott, A.J., & Loft, B.E. (1965). Group cohesiveness as interpersonal attraction: A reviewof relationships with antecedents and consequent variables. PsychologicalBulletin, 64, 259-309.Mannix, E., & Neale, M.A. (2005). What differences make a difference: The promise andreality of diverse teams in organizations. Psychological Science in the PublicInterest, 6, 3 1-55.Marcus, B.H., Napolitano, M.A., King, A.C., Lewis, B.A., Whiteley, J.A., Albrecht, A.,Parisi, A., Bock, B., Pinto, B., Sciamanna, C., Jakicic, J., & Papandonatos, G.D.(2007). Telephone versus print delivery of an individualized motivationally99tailored physical activity intervention: Project STRIDE. Health Psychology, 26,401-409.McAuley, E., Morris, K.S., Motl, R.W., Hu, L., Konopack, J.F., & Elavsky, S. (2007).Long-term follow-up of physical activity behavior in older adults. HealthPsychology, 26, 375-380.Meglino, B.M., Ravlin, E.C., Adkins, C.L. (1989). A work values approach to corporateculture: A field test of the value congruence process and its relationship toindividual outcomes. Journal ofApplied Psychology, 74, 424-432.Milliken, F.J., & Martins, J.L. (1996). Searching for common threads: Understanding themultiple effects of diversity in organizational groups. Academy ofManagementReview, 21, 402-433.Mohammed, S., & Angell, L.C. (2004). Surface- and deep-level diversity in workgroups:Examining the moderating effects of team orientation and team process inrelationship conflict. Journal ofOrganizational Behavior, 25, 1015-1039.Molleman, E. (2005). Diversity in demographic characteristics, abilities and personalitytraits: Do faultlines affect team functioning? Group Decision and Negotiation, 14,173-193.Montoya, R.M., Horton, R.S., & Kirchner, J. (2008). Is actual similarity necessary forattraction? A meta-analysis of actual and perceived similarity. Journal ofSocialand Personal Relationships, 25, 889-922.Mullen, B., & Cooper, C. (1994). The relation between group cohesiveness andperformance: An integration. Psychological Bulletin, 115, 210-227.Napolitano, M.A., Papandonatos, G.D., Lewis, B.A., Whiteley, J.A., Williams, D.M.,100King, A.C., Bock, B.C., Pinto, B., & Marcus, B.H. (2008). Mediators of physicalactivity behavioral change: A multivariate approach. Health Psychology, 27, 409-418.O’Reilly, C.A., Caidwell, D.F., & Barnett, W.P. (1989). Work group demography, socialintegration, and turnover. Administrative Science Quarterly, 34, 21-37.Orpen, C. (1984). Attitude similarity, attraction, and decision-making in the employmentinterview. The Journal ofPsychology, 117, 111-120.Pelled, L.H. (1996). Relational demography and perceptions of group conflict andperformance: A field investigation. International Journal ofConflictManagement, 7, 23 0-246.Pelled, L.H., Ledford, G.E., & Mohrman, S.A. (1999). Demographic dissimilarity andworkplace inclusion. Journal ofManagement Studies, 36, 1013-1031.Pelled, L.H., Xin, K.R., & Weiss, A.M. (2001). No es come mi: Relational demographyand conflict in a Mexican production facility. Journal ofOccupational andOrganizational Psychology, 74, 63-84.Pfeffer, J. (1983). Organizational demography. Research in Organizational Behavior, 5,299-357.Phillips, K.W. & Loyd, D.L. (2006). When surface and deep-level diversity collide: Theeffects on dissenting group members. Organizational Behavior and HumanDecision Processes, 99, 143-160.Phillips, K.W., Northcraft, G., & Neale, M. (2006). Surface-level diversity andinformation sharing: When does deep-level similarity help? Group Processes andIntergroup Relations, 9, 467-482.101Piasentin, K.A., & Chapman, D.S. (2007). Perceived similarity and complementarity aspredictors of subjective person-organization fit. Journal ofOccupational andOrganizational Psychology, 80, 341-354.Pillcington, N.W., & Lydon, J.E. (1997). The relative effect of attitude similarity andattitude dissimilarity on interpersonal attraction: Investigating the moderatingroles of prejudice and group membership. Personality and Social PsychologyBulletin, 23, 107-122.Peterson, J.L., & Miller, C. (1980). Physical attractiveness and marriage adjustment inolder American couples. Journal ofPsychology: Interdisciplinary and Applied,105, 247-252.Poizer, J.T., Milton, L.P., Swann, W.B. (2002). Capitalizing on diversity: Interpersonalcongruence in small work groups. Administrative Science Quarterly, 47, 296-324.Priem, R.L., Lyon, D.W., & Dess, G.G. (1999). Inherent limitations of demographicproxies in top management team heterogeneity research. Journal ofManagement,25, 935-953.Randel, A.E., & Earley, P.C. (2009). Organizational culture and similarity among teammembers’ salience of multiple diversity characteristics. Journal ofApplied SocialPsychology, 39, 804-833.Riordan, C.M., & Shore, L.M. (1997). Demographic diversity and employee attitudes: Anempirical examination of relational demography within work units. Journal ofApplied Psychology, 82, 342-358.Riordan, C.M., & Wayne, J.H. (2008). A review and examination of demographic102similarity measures used to assess relational demography within groups.Organizational Research Methods, 11, 562-592.Robbins, S.P., & Langton, N. (1999). Organizational Behaviour: Concepts,Controversies, Applications. Scarborough, ON: Prentice Hall Canada Inc.Rokeach, M. (1968). Beliefs, Attitudes, and Values. Josey-Bass Inc. San Francisco, CA.Royal, E.G., & Golden, S.B. (1981). Attitude similarity and attraction to an employeegroup. Psychological Reports, 48, 25 1-254.Sachs, D.H. (1975). Belief similarity and attitude similarity as determinants ofinterpersonal attraction. Journal ofResearch in Personality, 9, 57-65.Scheffer, J.L. (1997). The Analysis ofIncomplete Multivariate Data. Chapman & Hall.Scheffer, J. (2002). Dealing with missing data. Res. Lett. Math. Sd., 3, 153-160.Shapcott, K.M., Carron, A.V., Burke, S.M., Bradshaw, M.H., & Estabrooks, P.A. (2006).Member diversity and cohesion and performance in walking groups. Small GroupResearch, 37, 701-730.Shaver, K.G. (1975). An Introduction to Attribution Processes. Oxford: Winthrop.Shaw, M.E. (1981). Group Dynamics: The Psychology ofSmall Group Behavior: 3’’Edition. New York, NY: McGraw-Hill.Shaw, M.E., & Shaw, L.M. (1962). Some effects of sociometric grouping upon learningin a second grade classroom. Journal ofSocial Psychology, 57, 453-458.Singh, R., Ho, L.J., Tan, H.L., Bell, P.A. (2007). Attitudes, personal evaluations,cognitive evaluation and interpersonal attraction: On the direct, indirect andreverse-causal effects. British Journal ofSocial Psychology, 46, 19-42.Singh, R., Ng, R., Ong, E.L., Lin, P.K.F. (2008). Different mediators for the age, sex,103and attitude similarity effects in interpersonal attraction. Basic and Applied SocialPsychology, 30, 1-17.Spence, K.W. (1956). Behavioral Theory and Conditioning. New Haven, CT: YaleUniversity Press.Spink, KS., & Carron, A.V. (1992). Group cohesion and adherence in exercise-classes.Journal ofSport and Exercise Psychology, 14, 78-86.Spink, K.S., & Carron, A.V. (1994). Group cohesion effects in exercise classes. SmallGroup Research, 25, 26-42.Spink, K.S., & Carron, A.V. (1995). The group size-cohesion relationship in minimalgroups. Small Group Research, 26, 86-105.Stephens, T., & Craig, C. (1990). The Well-Being ofCanadians. Ottawa: CanadianFitness and Lifestyle Research Institute.Stevens, G., Owens, D., & Schaefer, E.C. (1990). Education and attractiveness inmarriage choices. Social Psychology Quarterly, 53, 62-70.Stokes, J.P. (1983). Components of group cohesion: Intermember attraction, instrumentalvalue, and risk taking. Small Group Behavior, 14, 163-173.Stangor, C., Lynch, L., Duan, C., & Clas, B. (1992). Categorization of individuals on thebasis of multiple social features. Journal ofPersonality and Social Psychology,62, 207-218.Strauss, J.P., Barrick, M.R., & Connerley, M.L. (2001). An investigation of personalitysimilarity effects (relational and perceived) on peer and supervisor ratings and therole of familiarity and liking. Journal ofOccupational and OrganizationalPsychology, 74, 637-657.104Tabachnick, B.C., & Fidell, L.S. (2007). Using Multivariate Statistics: Fifth Edition.Pearson Education Inc.Tan, D.T.Y., Singh, R. (1995). Attitudes and attraction: A developmental study of thesimilarity-attraction and dissimilarity-repulsion hypotheses. Personality andSocial Psychology Bulletin, 21, 975-086.Tajfel, H. (1981). Human Groups and Social Categories. Cambridge: CambridgeUniversity Press.Tajfel, H., & Turner, J.C. (1979). An integrative theory of intergroup conflict. In W.G.Austin & S. Worchel (eds.) Psychology ofIntergroup Relations. Monterey, CA:Brooks/Cole.Terborg, J.R., Castore, C., & DeNinno, J.A. (1976). A longitudinal field investigation ofthe impact of group composition on group performance and cohesion. Journal ofPersonality and Social Psychology, 34, 782-790.Triandis, H.C., Kurowski, L.L., & Gelfand, M.J. (1994). Workplace diversity. In H.C.Triandis, M.D. Dunnette, & L.M. Hough (eds.) Handbook ofIndustrialOrganizational Psychology. (pp. 769-827). Palo Alto, CA: Psychological Press.Triplett, N. (1897). The dynamogenic factors in pacemaking and competition. AmericanJournal ofPsychology, 9, 507-533.Turban, D.B., Dougherty, T.W., & Lee, F.K. (2002). Gender, race, and perceivedsimilarity effects in developing relationships: The moderating role of relationshipduration. Journal of Vocational Behavior, 61, 240-262.Turban, D.B., & Jones, A.P. (1988). Supervisor-subordinate similarity: Types, effects,and mechanisms. Motivation and Emotion, 14, 2 15-233.105Turner, J.C. (1984). Social identification and psychological group formation. In H. Tajfel(ed.) The Social Dimension: European Developments in Social Psychology.Cambridge: Cambridge University Press.Turner, J.C. (1985). Social categorization and the self-concept: A social cognitive theoryof group behavior. In E.J. Lawler (Ed.) Advances in Group Processes: Theory andResearch Volume 2. Greenwich, CT: JAI Press.Turner, J.C. (1987). Rediscovering the Social Group: A SeCategorization Theory. NewYork, NY: Basil Blackford Ltd.Tsui, A.S., Egan, T.D., & O’Reilly, C.A. (1992). Being different: Relational demographyand organization attachment. Administrative Science Quarterly, 37, 547-579.Tsui, A.S., Egan, T.D., & Xix, K.R. (1995). Diversity in organizations: Lessons fromdiversity research. In M. Chemers, S. Oskamp, & M. Costanzo (eds.) Diversity inOrganizations: New Perspectivesfor a Changing Workforce. Thousand Oaks,CA: Sage Publications.Tsui, A.S., & O’Reilly, C.A. (1989). Beyond simple demographic effects: Theimportance of supervisor-subordinate dyads. Academy ofManagement Journal,32, 402-423.Tsui, A.S., Porter, L.W., & Egan, T.D. (2002). When both similarities and dissimilaritiesmatter: Extending the concept of relational demography . Human Relations, 55,899-929.Vroom, V.H., & Pahl, B. (1971). Relationship between age and risk taking amongmanagers. Journal ofApplied Psychology, 55, 399-405.Wharton, A.S., & Baron, J.N. (1987). So happy together? The impact of gender106segregation on men at work. American Sociological Review, 52, 574-587.Wiersema, M.F., & Bird, A. (1993). Organizational demography in Japanese firms:Group heterogeneity, individual dissimilarity, and top management team turnover.Academy ofManagement Journal, 36, 996-1025.Wilcox, S., King, A.C., Brassington, G.S., & Ahn, D.K. (1999). Physical activitypreferences of middle-aged and older adults: A community analysis. Journal ofAging and Physical Activity, 7, 386-399.Wilder, D.A. (1984). Predictions of belief homogeneity and similarity following socialcategorization. British Journal ofSocial Psychology, 23, 323-33 3.Wilkenson, D. (2007). The multidimensional nature of social cohesion: Psychologicalsense of community, attraction, and neighboring. American Journal ofCommunityPsychology, 40, 2 14-229.Young, D.R., King, A.C., & Oka, R.K. (1995). Determinants of exercise level in thesedentary versus underactive older adult — implications for physical activityprogram development. Journal ofAging and Physical Activity, 3, 4-25.Zajonc, R.B. (1965). Social Facilitation. Science, 149, 269-274.Zellmer-Bruhn, M.E., Maloney, M.M., Bhappu, A.D., Salvador, R. (2008). When andhow do differences matter? An exploration of perceived similarity in teams.Organizational Behavior and Human Decision Processes, 107, 4 1-59.107Appendix AfJBCThe University of British ColumbiaOffice of Research ServicesBehavioural Research Ethics BoardSuite 102, 6190 Agronomy RoacJ Vancouver, B.C. V6T 1Z3CERTIFICATE OF APPROVAL - MINIMAL RISKPRINCIPAL INVESTIGATOR: INSTITUTION I DEPARTMENT: UBC BREB NUMBER:Mark R. Beauchamp UBC/Education/Human Kinetics H08-02850INSTITUTION(S) WHERE RESEARCH WILL BE CARRIED OUT:InstitutionISiteUBC Vancouver (excludes UBC Hospital)Other locations where the research will be conducted:Community/fitness centres in the Lower Mainland of British ColumbiaCO-INVESTIGATOR(S):Nilliam DunlopSPONSORING AGENCIES:UBC Dean of EducationPROJECT TITLE:Group Diversity and Exercise AdherenceCERTIFICATE EXPIRY DATE: December 3, 2009DOCUMENTS INCLUDED IN THIS APPROVAL: DATE APPROVED:December 3, 2008Document NameIVersionIDateProtocol:Protocol N/A November 21, 2008Questionnaire, Questionnaire Cover Letter. Tests:Questionnaire N/A November21, 2008Letter of Initial Contact:Program Coordinator Contact N/A November 21,2008Participant Contact and Consent Letter N/A November21, 2008Instructor Contact Letter N/A November21, 2008Other Documents:Log Sheet N/A November 21, 2008Ehe application for ethical review and the document(s) listed above havebeen reviewed and the procedures wereound to be acceptable on ethical grounds for research involving human subjects.Approval is issued on behalf of the Behavioural Research Ethics Boardand signed electronically by one of the following:Dr. M. Judith Lynam, ChairDr. Ken Craig, ChairDr. Jim Rupert, Associate ChairDr. Laurie Ford, Associate ChairDr. Daniel Salhani, Associate ChairDr. Anita Ho, Associate Chair108UL3CAppendix BThe University ofBritish ColumbiaOffice of Research ServicesBehavioural ResearchEthics BoardSuite 102, 6190Agronomy Road,Vancouver, B.C. V6T1Z3CERTIFICATEOF APPROVAL- M1NIMALRISK AMENDMENTPRINCIPAL INVESTIGATOR:DEPARTMENT:UBC BREB NUMBER:Mark R. BeauchampUBC/Education/HumanKinetics H08-02850INSTITUTION(S) WHERERESEARCH WILLBE CARRIED OUT:Institution ISiteUBCVancouver (excludesUBC Hospital)Other locations where the researchwill be conducted:ommunity/fltness centresin the LowerMainland of British ColumbiaDO-INVESTIGATOR(S):Miliam DunlopSPONSORING AGENCIES:UBC Dean of EducationPROJECT TITLE:Group Diversity and ExerciseAdherenceExpiry Date - Approval ofan amendment does notchange the expiry dateon the current UBC BREBapproval of this study. Anapplication for renewalis required on or before:December 3, 2009MENDMENT(S):IAMENDMENTAPPROVAL DATE:1January 29, 2009EoCument NameIVersion Dateuestionnaire, QuestionnaireCover Letter. Tests:Questionnaire - Time 3N/A January 22, 2009rhe amendment(s) and the document(s)listed above have been reviewedand the procedures were foundto be3cceptable on ethical groundsfor research involving humansubjects.Approval is issued on behalfof the Behavioural ResearchEthics Boardand signed electronically by oneof the following:Dr. M. Judith Lynam, ChairDr. Ken Craig, ChairDr. Jim Rupert, Associate ChairDr. Laurie Ford, Associate ChairDr. Daniel Salhani, Associate ChairDr. Anita Ho, Associate Chair109Appendix CExercise in Group ContextsPrincipal Investigator: Primary Contact:Mark R. Beauchamp, Ph.D. William Dunlop, M.A. StudentMichael Smith Foundation for Health Research Michael Smith Foundation forHealth Research (Scholar) Health Research (Trainee)School of Human Kinetics School of Human KineticsUniversity of British Columbia University of British ColumbiaContact Number: 604-822 4864 Contact Number: 604-822-0219mark. wdunlop2tinterchanqe.ubc.caDear Program CoordinatoiWe are researchers from the University of British Columbia (UBC) who are currently involved in a long-termprogram of research that is designed to better understand the factors that influence adherence to group-basedexercise programs. The reason we are writing to you is because we would like to inviteyou to take part in astudy that we are currently undertaking. This study would involve having class members at your centercompleting a very short (5-minute) questionnaire, about their perceptions of the exercise class they areenrolled in, on three occasions over the length of the exercise course. The questions do not deal with anyissues of a sensitive nature, and it is anticipated that the results of this research will be able to help thoseconcerned with health promotion develop better group-based exercise programs.If you decide to take part, all information that your members provide will be kept private/confidential, and willnot be shared with ANYONE else. This means that their responses will be combined with all others and so noparticipants will know how any other members have responded to the survey. All questionnaires will be kept ina locked cabinet in the office of the principal investigator (at UBC), and will not be made available to anyoneother than the researchers involved in this study. In this study, your members will be asked to provide the firstthree letters of their first and last name on the top of each questionnaire to allow us to match up thequestionnaires over the three time points. Once we have matched up each of the questionnaires, we willremove this information (all information that they have provided will remain anonymous). Our database will bepassword protected and will be stored on a secure computer in the office of principal investigator.There are no known risks associated with participation in this study. It is hoped that your center’s involvementwill help advance our understanding of the factors that influence adherence to group-based exercise programs.These results may be used to enhance future programs your center may pursue. If you have any questionsabout what is involved please contact Dr. Mark Beauchamp by email or phone (his contact details are at thetop of this page). You can also contact the Office of Research Services at UBC. Their phone number is 604-822-8598.Over the next week a researcher from our lab will be in touch via telephone to see if you would like to haveyour center participate.Many thanks in advance for your help,Mark Beauchamp, PhDWilliam Dunlop, M.A. Student110Appendix DExercise in Group ContextsPrincipal Investigator: Primary Contact:Mark R. Beauchamp, Ph.D. William Dunlop, M.A. StudentMichael Smith Foundation Michael Smith Foundationfor Health Research Scholar Research TraineeSchool of Human Kinetics School of Human KineticsUniversity of British Columbia University of British ColumbiaContact Number: 604-822 4864 Contact Number: 604-822-0219mark. wdunlop2interchange.ubc.caDear Fitness Instructor,We are researchers from the University of British Columbia (UBC) currently involved in a long-term program ofresearch that is designed to better understand the factors that influence adherence to group-based exerciseprograms. The reason we are writing to you is because we would like to invite you to take part in a study that weare currently undertaking. This study would involve having your class members completing a very short (5-minute) questionnaire, about their perceptions of the exercise class, on three occasions over the length of theexercise course. The questions do not deal with any issues of a sensitive nature, and it is anticipated that theresults of this research will be able to help those concerned with health promotion develop better group-basedexercise programs.To help us better understand how participants’ perceptions of the class relate to their involvement in that classwe would like to collect data about their adherence behaviours. Specifically, and only if you agree to do so, wewould like to provide you with an attendance log sheet that will allow us to track the attendance of members ofyour class. After each class, all we ask is that you record the class members that were in attendance on thesheet provided. If you allow us to involve your exercise class in this study, we will provide you with a$30honorarium regardless of whether you choose to record the attendance of your class members.All information that you and your members provide will be kept private/confidential, and will not be shared withANYONE else. This means that your class members’ responses will be combined with those from others thatparticipate in the study (across centres in the Lower Mainland of BC) and so no participants will know howanyone else has responded to the surveys. All questionnaires will be kept in a locked cabinet in the office of theprincipal investigator (at U BC), and will not be made available to anyone other than the researchers involved inthis study. Our database will be password protected and will be stored on a secure computer in the office ofprincipal investigator.There are no known risks associated with participation in this study. It is hoped that your involvement will helpadvance our understanding of the factors that influence adherence to group-based exercise programs. Theseresults may be used to enhance future intervention-based programs. If you have any questions about what isinvolved please contact Dr. Mark Beauchamp by email or phone (his contact details are at the top of this page).You can also contact the Office of Research Services at UBC. Their phone number is 604-822-8598. If, at theend of this research (after June 2009), you would like to see a summary report of the study findings pleasecontact either of us by e-mail, phone, or mail (see above contact details).Many thanks in advance for your help,Mark Beauchamp, PhDWilliam Dunlop, M.A. Student111Instructor Name:Class Location:Class Description:Appendix EAttendance Log SheetPlease provide each class members’ name and record (by providing a checkmark) eachtime they attended the class during each week listed.Member’s Name Wk. 1 Wk. 2 Wk. 3 Wk. 4 Wk. 5 Wk. 6 Wk. 7 Wk.8112Appendix FExercise Class StudyPrincipal Investigator: Primary Contact:Mark R. Beauchamp, Ph.D. William Dunlop, M.A. StudentMichael Smith Foundation Michael Smith Foundationfor Health Research Scholar Research TraineeSchool of Human Kinetics School of Human KineticsUniversity of British Columbia University of British ColumbiaContact Number: 604-822 4864 Contact Number: 604-822-0219mark. wdunIop2interchange.ubc.caDear class participant,We are researchers from the University of British Columbia (UBC) currently involved in a long-term program ofresearch that is designed to better understand the factors that influence adherence to group-based exerciseprograms. The reason we are writing to you is because you have recently enrolled in an exercise class, and weare interested in your perceptions about that class. Your participation in this study would involve completing avery short (5-minute) questionnaire on three occasions over the length of the exercise course. The questions donot deal with any issues of a sensitive nature, and it is anticipated that the results of this research will be able tohelp those concerned with health promotion develop better group-based exercise programs. At the end of yourfirst class a research assistant will invite you to take part in this research.Please know that your involvement in this study is completely voluntary. It’s up to you if you want to take part ornot. If for ANY reason, you do not want to take part in this study that’s fine, you don’t have to. If you decide totake part, you will also be free to withdraw at any time without having to give any reason. If you drop out you willnot experience ANY negative consequences at all. If you would like to take part in this study all you have to do iscomplete the questionnaires described above (by completing the questionnaires you have consented to take partin this research). We recommend that you keep a copy of this letter for your records.If you decide to take part, all information you provide will be kept private/confidential, and will not be shared withANYONE else. This means that your responses will be combined with those of other participants and so no-onewill know how you will have answered the surveys except you. All questionnaires will be kept in a locked cabinetin the office of the principal investigator (at UBC), and will not be made available to anyone other than theresearchers involved in this study. Our database will be password protected and will be stored on a securecomputer in the office of principal investigator.There are no known risks associated with participation in this study. If you have any questions about what isinvolved please contact Dr. Mark Beauchamp by email or phone (his contact details are at the top of this page).You can also contact the Office of Research Services at UBC if you have any concerns about your rights ortreatment as a research subject. Their phone number is 604-822-8598. If, at the end of this research (after June2009), you would like to see a summary report of the study findings please contact either of us by e-mail, phone,or mail (see above contact details).Many thanks in advance for your help,Mark Beauchamp, PhDWilliam Dunlop, M.A. Student113Appendix GExercise and Group Contexts Questionnaire First Name (first three letters only):Last Name (first 3 letters only):Postal Code (first 3 letters only):1) Gender (please circle appropriate response): Male / Female2) Your Age:___________yrs3) What is your occupation?__________________________________________6) Education Level:High school education College or technical training Undergraduate degreeGraduate degree (e.g., MA, M.D., Ph.D.) Other____________________7) How do you describe yourself in terms of your ethnic origin? PLEASE CHECK THAT APPLY.V V VCanadian East Indian American (USA)Native/Aboriginal Dutch NorwegianChinese Persian ItalianBritish Polish KoreanIrish Hispanic FilipinoGerman Portuguese JewishFrench Vietnamese JapaneseOther__________8) Height____________ cms (or_____________ft)9) Weight____________ kgs (or___________Ibs)10) Which of the following activities have you taken part in during the last 2 weeks?ActivityVHow For how long Experienced (a) no, (b) small, (c) moderate, or (d)many — total large increases in heart and breathing rates whiletimes? (mins)? participatingWalkingRunningGardeningYogaAerobics/Exercise Classes —SwimmingCyclingRacquet Sports —(Others)12) In my exercise class, I believe that group members are similar to me in terms of:Strongly StronglyDisagree______Age 1 2 3 4 5 6 7 8 9Gender 1 2 3 4 5 6 7 8 9Attitudes 1 2 3 4 5 6 7 8 9Education 1 2 3 4 5 6 7 8 9Personal values 1 2 3 4 5 6 7 8 9Physical appearance 1 2 3 4 5 6 7 8 9Personal beliefs 1 2 3 4 5 6 7 8 9Life experiences 1 2 3 4 5 6 7 8 9Physical condition 1 2 3 4 5 6 7 8 9Ethnicity 1 2 3 45 6 7 8 911413)Strongly StronglyDisagree AgreeOverall, I feel that I am similar to othermembers of my exercise class:2 3 4 5 6 7 8 9114___________out of group members.your exercise class, as well as your perceptions aboutStrongly StronglyDisagree — — — A1. I like the amount of physical activity I get in this program.1 2 3 4 5 6 7 8 92. This physical activity group provides me with a good opportunity to2 3 4 6 7 8 9improve in areas of fitness I consider important. — — — — — —3. I am happy with the intensity of this physical activity program.1 2 3 4 5 6 7 8 94. I like the program of physical activities done in this group.2 3 4 5 6 7 8 95. I enjoy new exercises done in this physical activity group.2 3 4 5 6 7 8 96. This physical activity group provides me with good opportunities to2 3 4 5 6 7 8 9improve my personal fitness. — — — —7. This physical activity group is an important social unit for me.1 2 3 4 5 6 7 8 98. 1 enjoy my social interactions within this physical activity group.2 3 4 6 7 8 99. I like meeting the people who come to this physical activity group.2 3 4 5 6 7 8 910. If this program was to end, I would miss my contact with the other2 3 4 6 7 8 9participants. — — — — —11. In terms of the social experiences in my life, this physical activity1 2 3 4 5 6 7 8 9group is very important to me. — — —12. The social interactions I have in this physical activity group are1 2 3 4 6 7 8 9important to me. — — — —13. Our group is united in its beliefs about the benefits of the physical1 2 3 4 5 6 7 8 9activities offered in this program. — — — —14. Our group is in agreement about the program of physical activities1 2 3 4 5 6 7 8 9that should be offered. — — — —15. Members of our group are satisfied with the intensity of the physical1 2 3 4 6 7 8 9activity in this program. — — — — —16. Members of our group enjoy helping if work needs to be done to1 2 3 4 5 6 7 8 9prepare for the activity sessions. — — — — — —17. We encourage each other in order to get the most out of the program.1 2 3 4 6 7 8 918. Members of our physical activity group often socialize during2 3 4 5 6 7 8 9exercise time. — — — — — — —19. Members of our physical activity group would likely spend time1 2 3 4 5 6 7 8 9together if the program were to end. — — — — — — —20. Members of our group sometimes socialize together outside of1 2 3 4 5 6 7 8 9activity time. — — — — — — —21. We spend time socializing with each other before and after our1 2 3 4 6 7 8 9activity sessions. — — — — — — —How many members of your exercise class do you feel that you are very similar to?15) The following questions correspond to your personal involvement withthe class as a whole. Please circle the appropriate number from I (Strongly Disagree) to 9 (StronglyAgree).115Exercise and Group Contexts Questionnaire1) Gender (please circle appropriate response): Male / Female2) Your Age:___________yrs3) What is your occupation?__________________________________________6) Education Level:High school education College or technical trainingGraduate degree (e.g., MA, M.D., Ph.D.) Other_____8) Height____________ cms (or_____________ft)9) Weight____________ kgs (or___________Ibs)10) Which of the following activities have you taken part in during the last 2 weeks?YESYESYESYESYESNONONONONOAppendix HFirst Name (first three letters only): — — —Last Name (first 3 letters only): —Postal Code (first 3 letters only): — —Undergraduate degreeActivity “ How For how long Experienced (a) no, (b) small, (c) moderate, or (d)many — total large increases in heart and breathing rates whiletimes? (mins)? participatingWalkingRunningGardeningYogaAerobics/Exercise Classes —SwimmingCyclingRacquet Sports(Others)classes.11 A) During the first eight weeks of this course how many classes have you missed?11 B) How many classes have you attended?_________classes.12) Prior to this course, had you previously taken a course at this center?13) Prior to this course, had you previously taken a course with this instructor?14) Prior to this course, did you know any of the other members in this group?15) Have you signed up for a future course at this center?16) If you have not yet signed up for a future course at this center, do you plan to do so?17) In my exercise class, I believe that group members are similar to me in terms of:Strongly StronglyDisagree______ — —Age 1 2 3 4 5 6 7 8 9Gender 1 2 3 4 5 6 7 8 9Attitudes 1 2 3 4 5 6 7 8 9Education 1 2 3 4 5 6 7 8 9Personal values 1 2 3 4 5 6 7 8 9Physical appearance 1 2 3 4 5 6 7 8 9Personal beliefs 1 2 3 4 5 6 7 8 9Life experiences 1 2 3 4 5 6 7 8 9Physical condition 1 2 3 4 5 6 7 8 9Ethnicity1 2 3 4 5 6 7 8 911618)19) How many members of your exercise class do you feel that you are very similar to?____________out of groupmembers.20) The following questions correspond to your personal involvement with your exercise class, as well as your perceptions aboutthe class as a whole. Please circle the appropriate number from 1 (Strongly Disagree) to 9 (Strongly Agree).Strongly Strongly— — — —1. I like the amount of physical activity I get in this program.1 2 3 4 5 6 7 8 92. This physical activity group provides me with a good opportunity to2 3 4 5 6 7 8 9improve in areas of fitness I consider important. — — — — — — —3. I am happy with the intensity of this physical activity program.1 2 3 4 5 6 7 8 94. I like the program of physical activities done in this group.2 3 4 5 6 7 8 95. I enjoy new exercises done in this physical activity group.2 3 4 5 6 7 8 96. This physical activity group provides me with good opportunities to2 3 4 5 6 7 8 9improve my personal fitness. — — — — — — — —7. This physical activity group is an important social unit for me.1 2 3 4 5 6 7 8 98. I enjoy my social interactions within this physical activity group.2 3 4 5 6 7 8 99. 1 like meeting the people who come to this physical activity group.2 3 4 5 6 7 8 910. If this program was to end, I would miss my contact with the other2 3 4 6 7 8 9participants. — — — — — — — —11. In terms of the social experiences in my life, this physical activity1 2 3 4 5 6 7 8 9group is very important to me. — — — — — — — — —12. The social interactions I have in this physical activity group are1 2 3 4 6 7 8 9important to me. — — — — — — — — —13. Our group is united in its beliefs about the benefits of the physical1 2 3 4 5 6 7 8 9activities offered in this program. — — — — — — — — —14. Our group is in agreement about the program of physical activities that1 2 3 4 5 6 7 8 9should be offered. — — — — — — — — —15. Members of our group are satisfied with the intensity of the physical1 2 3 4 5 6 7 8 9activity in this program. — — — — — — — — —16. Members of our group enjoy helping if work needs to be done to1 2 3 4 5 6 7 8 9prepare for the activity sessions. — — — — — — — — —17. We encourage each other in order to get the most out of the program.1 2 3 4 5 6 7 8 918. Members of our physical activity group often socialize during exercise2 3 4 6 7 8 9time. — — — — — — — — —19. Members of our physical activity group would likely spend time1 2 3 4 6 7 8 9together if the program were to end. — — — — —20. Members of our group sometimes socialize together outside of activity1 2 3 4 6 7 8 9time. — — — — — — —21. We spend time socializing with each other before and after our activity1 2 3 4 6 7 8 9sessions. — — —117


Citation Scheme:


Usage Statistics

Country Views Downloads
United States 7 2
China 4 27
Canada 3 4
Romania 2 0
France 1 0
United Kingdom 1 0
Ukraine 1 0
India 1 0
Italy 1 0
City Views Downloads
Shenzhen 4 27
Unknown 3 20
Sunnyvale 3 0
Craiova 2 0
Ashburn 2 0
Vancouver 2 1
Hamilton 1 2
London 1 0
Chicago 1 0
Florence 1 0
Seattle 1 0

{[{ mDataHeader[type] }]} {[{ month[type] }]} {[{ tData[type] }]}
Download Stats



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items