adolescent hierarchy formation and the social competition theory of depression

30
Journal of Soc¡al and Cl¡n¡cal Psychology, Vol. 28, No. 9, 2009, pp. 1144-1172 ADOLESCENT HIERARCHY FORMATION AND THE SOCIAL COMPETITION THEORY OE DEPRESSION MARCA. FOURNIER University of Toronto Scarborough Social competition theorists propose: (a) social hierarchies form spontaneously when resources are scarce; (b) social rank in the human hierarchy depends part- ly on attention-holding; and (c) depression constitutes an involuntary response to low social rank, entrapment, anci defeat. These propositions were tested in a sample of 121 students in grades 7-11 at an all-boys school. Students complet- ed sociometric evaluations of their classmates' reputation and likability, as well as self-report surveys assessing attachment security, social support, self-esteem, and depression. Empirical support for all three propositions was obtained. First, adolescents perceived hierarchy in their social ecology, agreed on where others stood in the social hierarchy, and knew their own place in the social hierarchy. Second, adolescents obtained prominence, respect, and influence by being lik- able; this was evident both from the perceiver's point of view, in the impressions of reputation and likability that each adolescent provided, and from the target's point of view, in the impressions of reputation and likability that each adolescent received. Third, over and above the variance explained by attachment security, social support, and peer-rated likability, sociometric determinations of an ado- lescent's place in the hierarchy contributed significantly to the prediction of his self-esteem and his depression. This research was supported by a grant from the Social Sciences and Humanities Research Council of Canada. Portions of this research were presented at the tenth annual meeting of the Society for Interpersonal Theory and Research (SITAR), Madison, Wisconsin. I wish to thank Debbie Moskowitz, David Zuroff, Kate McLean, Joey Cheng, and Nina Dhir for their thoughtful comments on earlier drafts of this article. Correspondence concerning this article should be addressed to Marc A. Fournier, Department of Psychology, University of Toronto Scarborough, 1265 Military Trail, Toronto, ON, Canada MIC 1A4. E-mail: [email protected]. 1144

Upload: claudia-borges-oliveira

Post on 26-Jan-2016

220 views

Category:

Documents


1 download

DESCRIPTION

Adolescent Hierarchy Formation and the Social Competition Theory of Depression

TRANSCRIPT

Page 1: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

Journal of Soc¡al and Cl¡n¡cal Psychology, Vol. 28, No. 9, 2009, pp. 1144-1172

ADOLESCENT HIERARCHY FORMATION ANDTHE SOCIAL COMPETITION THEORY OEDEPRESSION

MARCA. FOURNIERUniversity of Toronto Scarborough

Social competition theorists propose: (a) social hierarchies form spontaneouslywhen resources are scarce; (b) social rank in the human hierarchy depends part-ly on attention-holding; and (c) depression constitutes an involuntary responseto low social rank, entrapment, anci defeat. These propositions were tested in asample of 121 students in grades 7-11 at an all-boys school. Students complet-ed sociometric evaluations of their classmates' reputation and likability, as wellas self-report surveys assessing attachment security, social support, self-esteem,and depression. Empirical support for all three propositions was obtained. First,adolescents perceived hierarchy in their social ecology, agreed on where othersstood in the social hierarchy, and knew their own place in the social hierarchy.Second, adolescents obtained prominence, respect, and influence by being lik-able; this was evident both from the perceiver's point of view, in the impressionsof reputation and likability that each adolescent provided, and from the target'spoint of view, in the impressions of reputation and likability that each adolescentreceived. Third, over and above the variance explained by attachment security,social support, and peer-rated likability, sociometric determinations of an ado-lescent's place in the hierarchy contributed significantly to the prediction of hisself-esteem and his depression.

This research was supported by a grant from the Social Sciences and HumanitiesResearch Council of Canada. Portions of this research were presented at the tenthannual meeting of the Society for Interpersonal Theory and Research (SITAR), Madison,Wisconsin.

I wish to thank Debbie Moskowitz, David Zuroff, Kate McLean, Joey Cheng, andNina Dhir for their thoughtful comments on earlier drafts of this article.

Correspondence concerning this article should be addressed to Marc A. Fournier,Department of Psychology, University of Toronto Scarborough, 1265 Military Trail,Toronto, ON, Canada MIC 1A4. E-mail: [email protected].

1144

Page 2: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

ADOLESCENT HIERARCHY EORMATION 1145

Adolescence is a time when reputation and likability seem to haveparticular significance (Rosenberg, 1965), during which concernsabout low social standing in the eyes of one's peers contributesignificantly to the experience of depression. Several studies havefound that rejected children report particularly high levels of de-pression (see Newcomb, Bukowski, & Fattee, 1993). In the presentresearch, adolescent reputation and vulnerability to depression areconceptualized within the framework of social competition theory(Gilbert, 2000; Price, 2000; Sloman, 2000). The purpose of this re-search is to provide, with data obtained from a sample of adoles-cent boys, support for the claims of the social competition theoriststhat hierarchy formation is natural, predictable, and consequential:natural, in that group hierarchies tend to emerge spontaneously innaturalistic settings wherever people congregate; predictable, inthat each person's position in the group is predicated in part on theability to hold positive social attention; and consequenfial, in thatlow social rank prompts an involuntary response called the invol-untary defeat strategy that can, imder the proper conditions, elicitthe experience of depression.

SPONTANEOUS HIERARCHY FORMATION

In the 1970s, a body of evidence emerged attesting to the spontane-ous formation of hierarchies in the social ecologies of young chil-dren and early adolescents. Strayer and Strayer (Strayer & Strayer,1976; Strayer & Strayer, 1977) found evidence of stable dominancehierarchies in the free play of preschool children. Savin-Williams(1976, 1977, 1979) found evidence of stable dominance hierarchiesamong early adolescents away at summer camp. These dominancerelations were found to reduce intragroup confiict, to facilitate thedistribution of scarce resources, and to help organize the division oflabor. Social competition theorists (Fournier, Moskowitz, & Zuroff,2002) assume that reproductively relevant resources, such as food,territory, and mating opportunities, were often in scarce supply overthe course of human evolutionary history. Social hierarchies servedto organize priority of access to these resources, and so humans ac-quired, through evolutionary selection pressures, a repertoire of be-havioral strategies with which to assess and pursue standing in thesocial hierarchy.

Page 3: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

1146 FOURNIER

COMPETITION THROUGH ATTRACTION

Social hierarchies in effect summarize the history of competitiveinteractions between group members, whereby those who wonrank contests now hold higher positions in the social hierarchy thanthose who lost (Bernstein, 1981; Hawley & Little, 1999; Strayer &Strayer, 1976). However, social competition in humans is in manyways different from the kinds of social competition observed inother species. Although humans continue in some contexts to de-termine dominance relations through face-to-face aggression andintimidation, social competition in humans is largely based on theability to attract others' attention and interest, their admiration andinvestment. Therefore, in comparison to species where standing inthe dominance hierarchy is predicated upon one's resource-holdingpotential (RHP; Parker, 1974), defined as one's size, strength, andfighting ability, standing in the human hierarchy requires one tohave positive social attention-holding power (SAHP; Gilbert, 1992,1997, 2000), defined as the ability to attract and hold the positiveattention and interest of others. Humans must therefore competeto be valued, chosen, and liked—in essence, to stimulate feelings ofpositive affect in others—through the display of their skills, talents,and competencies.

The importance of social attractiveness can be seen in the socialdevelopment literature. Before the age of five, disputes over so-cially valued but limited resources (toys, play partners) are mostoften resolved by aggressive-coercive behavior, whereby the moredominant child simply takes the resource in question from theless dominant child. Through their aggressive interpersonal style,domineering toddlers and preschoolers rise to the top of the hierar-chy, where they are the objects of attention and imitation from theirpeers (Abramovitch, 1976; Abramovitch & Grusec, 1978; Hawley &Little, 1999; LaFreniere & Charlesworth, 1983; Mischel & Grusec,1966; Strayer & Trudel, 1984; Vaughn & Waters, 1981). Around theage of five, however, agonistic encounters begin to decrease infrequency. Aggressive children from this point on become targetsof peer disapproval and rejection as their aggressive behavior be-comes tolerated less and less frequently (Coie & Dodge, 1993; Coie,Dodge, & Cappotelli, 1982; Newcomb et al, 1993). From this pointon, disputes over resources are more often settled by compromis-

Page 4: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

ADOLESCENT HIERARCHY FORMATION 1147

ing, turn-taking, and sharing. Standing in the hierarchy is thereafterpredicated upon one's degree of centrality to the social network,one's capacity to establish cooperative alliances, and one's displayof socially desirable skills, talents, and competencies.

DEFEAT AND DEPRESSION

In the pursuit of social standing some competitors will lose, andsocial competition theorists assert that humans come equippedwith an extensive repertoire of defensive strategies to assist themwhen they find themselves facing defeat (Gilbert, 1992,2000,2001a,2001b). These defenses include ïg i (reinvesting efforts to win thecontest or to attain the goal), flight (redirecting efforts to escape thecontest), help-seeking (enlisting the aid of allies), and disengagement(relinquishing the goal and accepting one's lower social rank). How-ever, should a competitor feel unable to prevail, unable to escape thecompetitive encounter, unable to enlist the aid of allies, and unableto let go and admit defeat, a highly stressful situation of entrappeddefeat may ensue and elicit what social competition theorists referto as the involuntary defeat strategy (IDS; Sloman & Gilbert, 2000).The IDS discourages losing competitors from persisting in the pur-suit of an unobtainable goal and assists them in accommodating theloss of a rank contest.

The IDS is a defensive behavioral strategy that involves: (a) theinhibition of dominance displays that might provoke harmful coun-ter-attacks from powerful others; (b) the display of submissive be-havioral cues that signal "no contest" to potential rivals and adver-saries; and (c) a subjective sense of "personal failure, inferiority, in-ability, powerlessness, and hopelessness" (Sloman, Price, Gilbert, &Gardner, 1994, p. 405) that discourages the losing competitor frompersisting in the face of inevitable defeat, thus limiting both the pos-sibility of injury or death and the loss of time and energy in thepursuit of unattainable goals. Usually, the IDS is activated for onlybrief time periods; the individual accepts that the contest cannot bewon or the goal attained, opts instead to reconcile with his or hercompetitors, and then moves on to pursue other contests or goals.Brief activation of the IDS would appear as mild depression or dys-phoria. However, should the state of entrapped defeat persist, thenthe IDS may become intensified and prolonged and a severe, poten-tially debilitating clinical depression may develop. Furthermore, asthe individual's subsequent social comparisons with those higher

Page 5: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

1148 FOURNIER

up in the hierarchy confirm his or her low RHP and/or SAHP, theindividual's level of self-esteem, an indicator and regulator of chal-lenge and confidence (Gilbert, Price, & Allan, 1995), may fall. There-fore, from the vantage point of the social competition theorist, lowsocial rank in the hierarchy is expected to predict higher levels ofdepression and lower levels of self-esteem.

Empirical research has linked several of the h5^othesized com-ponents of the IDS to the experience of depression. Depression cor-relates with feelings of inferiority (Allan & Gilbert, 1997; Cheung,Gilbert, & Irons, 2004; Gilbert & Allan, 1998; Zuroff, Fournier, &Moskowitz, 2007), with high levels of submissive behavior (Allan& Gilbert, 1997; Gilbert & Allan, 1998; Irons & Gilbert, 2005; Zuroffet al, 2007), with low levels of dominance (Zuroff et al., 2007) andinfluence (Nezleck, Hampton, & Shean, 2000), as well as with fear-ing one's own anger and suppressing overt hostility (Allan & Gil-bert, 2002; Gilbert, Gilbert, & Irons, 2004). Furthermore, empiricalresearch has linked depression to entrapped defeat. Depression cor-relates with feelings of humiliation and defeat (Brown, Harris, &Hepworth, 1995; Gilbert & Allan, 1998; Kendler, Hettema, Butera,Gardner, & Prescott, 2003), with wanting to escape and yet feelingtrapped (Brown et al., 1995, Gilbert & Allan, 1998; Gilbert et al.,2004), and with feeling that help from others is not available (Bill-ings, Cronkite, & Moos, 1983). However, evidence supporting thesocial competition theory of depression is limited by researchers' al-most exclusive reliance on self-reports of the principal constructs.

THE PRESENT STUDY

A more comprehensive test of the social competition theory wouldrequire more than the recruitment of unacquainted individuals whocomplete a series of self-report questiormaires concerning their feel-ings and behavior in their respective social ecologies. One strategywould be to recruit entire social ecologies, so that every group mem-ber not only self-reports on their own feelings and behavior, but alsoshares their impressions and evaluations of all other individuals inthe same social ecology; such is the case with round-robin designs,in which each participant rates every other participant and data arecollected from both members of all possible dyadic combinations ofparticipants (Kenny, 1994). This painstaking, labor-intensive tech-nique would provide a unique test of the social competition theoryand its proposition that the successes and failures of individuals to

Page 6: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

ADOLESCENT HIERARCHY FORMATION 1149

obtain standing in the social hierarchy render them differentiallyvulnerable to experiences of low self-esteem and depression.

The purpose of the present research is to test the social competitiontheory of depression in a sample of adolescent boys in grades 7-11.The following propositions will be examined: first, that hierarchiesemerge spontaneously in the social ecologies of adolescents; second,that hierarchies emerge as a function of the differences between ad-olescents in their respective capacities to socially attract others tothem; and third, that the emergence of hierarchy has consequencesfor adolescents' socio-emotional functioning, such that standing inthe hierarchy is predictive of self-esteem and depression. Further-more, the utility and incremental validity of the social competitiontheory's key construct, social rank (assessed here as peer-rated rep-utation), will be demonstrated by its ability to predict self-esteemand depression over and above the variance in these outcomes thatcan be explained by the well-established vulnerability factors ofattachment insecurity, low social support, and interpersonal rejec-tion. Adolescent reputation will be assessed using sociometric rat-ing procedures both to overcome past researchers' over-reliance onself-reports and to ensure that the validity evidence obtained couldnot be attributed to shared method variance between predictor andcriterion.

Adolescents in grades 7-11 at a private all-boys school participatedin the present study. Same-sex environments are well suited to testthe present hypotheses, given the complex interpersonal dynam-ics that emerge between adolescent boys and girls in coed institu-tions. Adolescents completed sociometric ratings of all the studentsin their grade. Students rated the extent to which they saw each oftheir classmates as having prominence, respect, and influence; thesepeer-ratings served as indicators of reputation. Students also ratedthe extent to which they liked each of their classmates and enjoyedhanging out with them; these peer-ratings served as indicators oflikability. Finally, students completed a battery of self-report ques-tionnaires to assess their levels of attachment security, social sup-port, self-esteem, and depression.

HYPOTHESES

Although it was expected that reputation and likability, the key con-cepts of this research, would be strongly correlated, it was also ex-

Page 7: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

1150 EOURNIER

pected that they could be successfully differentiated. Likability wasframed at the level of dyadic interaction (i.e., "Does i like hangingout with/?"), and so it was expected that students' ratings of likingwould display some level of reciprocity (i.e., that if / likes /, then ;will like /)• Kenny (1994) foimd that dyadic correlations (reciprocity)are larger for ratings of liking than for trait ratings, and so largerreciprocity effects were expected for likability than for reputation.In contrast, reputation was framed at the level of group interaction(i.e., "Does i see ; as prominent, respected, and influential in thegrade?"), and so it was expected that students' ratings of reputationwould display some level of consensus (i.e., that if / sees / as promi-nent, respected, and influenfial, then so will k). Kenny (1994) foundthat target effects (consensus) are larger for leadership ratings thanfor ratings of liking, and so larger consensus effects were expectedfor reputafion than for likability. These hypotheses were tested with-in Kenny's (1994) social relations model, making use of round-robinratings in which students both provide evaluations of others andreceive evaluations from them. Differences in the processes under-lying the perception of reputation and likability were demonstratedto provide sufficient evidence that these two constructs are indeedseparable. Following this demonstration, three sets of hypothesesconcerning adolescent hierarchy formation were tested.

The first set of hypotheses concerned the nature and spontane-ity of hierarchy formation. The study of spontaneous social rank-ordering has an extensive history, and the phenomenon has beenwell documented both inside the laboratory using leaderless groupdiscussions (Bass, 1954) and outside the laboratory using naturallyoccurring groups, such as fraternities and sororities (Anderson,John, Keltner, & Kring, 2001; Harms, Roberts, & Wood, 2007). Thisresearch suggests that social hierarchies often form quite rapidly,possibly at first glance (Kalma, 1991), in both contrived and natu-ralistic settings. Three sources of evidence were expected to attestto the spontaneous formation of hierarchies in the adolescent socialecology. First, it was expected that adolescents would recognize thehierarchical features of the social ecology and would use a rangeof values to characterize the prominence, respect, and influence ofthose around them. Second, given that these adolescents were tooccupy the same social ecology, it was expected that they woulddemonstrate a high level of agreement in their assessments of theirclassmates' prominence, respect, and influence. Third, given thatcompefitors must know their social rank (Fournier et al., 2002) andtheir social attractiveness (Gilbert et al., 1995) in order to decide

Page 8: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

ADOLESCENT HIERARCHY FORMATION 1151

whether or not to compete, it was expected that adolescents wouldbe able to gauge their own standing in the social ecology.

The second set of hypotheses concerned the prediction of adoles-cent hierarchy formation from individual differences in the ability toattract and hold the positive social attention of others. Adolescents'SAHP was operationalized in terms of their likability, a conceptwith a long history in the field of sociometry. Sociometric research-ers have converged upon a two-dimensional system for character-izing the social status of children and adolescents (Coie et al., 1982;Newcomb & Bukowski, 1983) that differentiates social preferencefrom social impact. Sociometric procedures ask children to nominatethe peers that they most like and most dislike; the nominations thateach child receives are summed and standardized to produce in-dices of acceptance and rejection. Social preference, the differencebetween acceptance and rejection, quantifies the likability of eachchild. Social impact, the total of acceptance and rejection, quantifiesthe visibility of each child. It was expected that adolescents wouldobtain prominence, respect, and influence by being likable (i.e., thatreputation and likability would be correlated), and that this wouldbe evident both from the perceiver's point of view, in the evalua-tions of reputation and likability that each adolescent provided, andfrom the target's point of view, in the evaluations of reputation andlikability that each adolescent received.

The third set of hypotheses concerned the impact of adolescenthierarchy formation on individual differences in self-esteem anddepression, and the incremental validity of adolescent peer-ratingsof reputation in predicting rates of self-esteem and depression. Itseemed likely that peer-based indicators of social rank would tosome extent correlate with other indices of social competence; how-ever, over and above the variance accounted for by other socialcompetence indicators like attachment security, social support, andpeer-based likability (or its counterpart, interpersonal rejection), itwas still expected that adolescent peer-ratings of reputation wouldcontribute significantly to the prediction of both self-esteem and de-pression. Therefore, given that low social rank presents an increasedopportunity for the IDS to become activated, it was expected thatpeer-based indicators of adolescent social rank would contribute tothe prediction of both self-esteem and depression, even after con-trolling for other established indicators of social competence. Suchfindings would attest to the incremental validity of social rank, thesocial compefition theory's key construct.

Page 9: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

1152 FOURNIER

The present study thus provided a comprehensive test of thethree basic propositions underlying the social competition theoryof depression—i.e., that hierarchy formation is natural, predictable,and consequential. It was the first study to incorporate both self-and peer-reports, thereby providing a test of the hypotheses thatwas perhaps more stringent than previous studies, and it was thefirst study to examine the social competition theory in the naturallyoccurring social ecologies of adolescents. The present study thusprovided an important opportunity to examine hierarchy formationand adolescent vulnerability to depression.

METHOD

PARTICIPANTS

Students in grades 6-12 at a private all-boys school participated inthe present study. Most of the students enrolled at the school comefrom affluent families. Of the 531 students enrolled at the time oftesting, 265 (50%) were boarding students.

PROCEDURE

Letters were sent to all parents explaining the purpose of the studyand requesting their consent for their sons' participation. The lettersstated explicitly that their sons' participation would be voluntary,and that both they and their sons would have the choice to decidewhether or not the son would participate. Hard copies of the letterwere mailed to parents in the summer, and follow-up copies weree-mailed to parents in the fall. Of the 531 boys enrolled at the school,parental consent was obtained for 181 (34%). In the weeks prior todata collection in the spring, letters were sent to students explainingthe purpose of the study and requesting their consent to participate.Of the 181 boys for whom parental consent was obtained, 148 (82%)agreed to participate. Students were tested in two 90-minute ses-sions in the last week of April and the first week of May and werecompensated with pizza following each testing session.

Only seven students from grade 6 provided usable data; althoughthese seven students constituted a third of the grade, this samplewas considered too small to provide stable estimates. Fourteen stu-dents from grade 12 provided usable data; although this sample

Page 10: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

ADOLESCENT HIERARCHY FORMATION 1153

was comparable to those obtained in other grades, these fourteenstudents constituted only one seventh of the grade. Consequently,statistical analyses were restricted to the data provided by studentsin grades 7-11. The final sample upon which the following analyseswere performed consisted of 121 boys, or almost one-third of thestudent population enrolled in grades 7-11 at the time of testing.^

The final sample consisted of 16 students in grade 7 (28% of thegrade), 18 students in grade 8 (32% of the grade), 34 students in grade9 (36% of the grade), 21 students in grade 10 (24% of the grade), and32 students in grade 11 (26% of the grade). With respect to ethnic-ity, 78 students (64%) described themselves as White, 14 students(12%) described themselves as Chinese, 10 students (8%) describedthemselves as Korean, 10 students (8%) described themselves as La-tino, and 2 students (2%) described themselves as Black; the remain-ing ethnic categories were only endorsed by one student (1%) each.Eighty-nine students (74%) reported having been born in Canada.Fifty-three students (44%) reported that they board at the school.

MEASURES

Sociometric Ratings. Adolescents completed a series of sociometricrating tasks. Research assistants were on hand during both testingsessions to address any questions concerning these tasks. One taskconcerned students' social status and prestige, and included the fol-lowing set of instructions: "Thinking of the students in your grade,you might see how some of them have prominence, respect, and in-fluence. Other students pay attention to them; in other words, theyare prominent. Other students look up to them; in other words, theyare respected. Other students go along with them; in other words,

1. The participation rates obtained in the present study raise two important topicsof concern. First, those who participated may have differed systematically from thosewho did not. Indeed, peer-ratings of reputation were significantly higher among thosewho participated (M = 2.68) than among those who did not (M = 2.47), F = 9.07, p < .01,and peer-ratings of likability were significantly higher among those who participated(M = 3.08) than among those who did not (M = 2.97), F = 4.14, p < .05. However, thesedifferences are similar to those already reported in the sociometric literature (Noll,Zeller, Vannatta, Bukowski, & Davies, 1997) and the size of these differences wouldappear to be rather modest. Second, peer-ratings of reputation and likability may havelimited validity given that only about one third of the sampling population contributedto these evaluations. Past research suggests that the validity of the sociometricprocedure with younger children may be compromised unless >70-75% of all classroom

(continued)

Page 11: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

1154 FOURNIER

they are influential. Below is a list of the students in your grade.Please rate each of your classmates in terms of his prominence,respect, and influence on a scale ranging from 1 (Not at All) to 5(A Lot) and then circle the appropriate number." A separate taskconcerned students' social preferences, and included the followingset of instructions: "Below is a list of the students in your grade.Please think about how much you like each of your classmates. Ona scale ranging from 1 (Not at All) to 5 (A Lot), how much do youlike hanging out with each of them? Please circle the appropriatenumber." The ratings that each student received on these two so-ciometric rating tasks were aggregated to produce peer-ratings ofreputation and likability.

As part of each rating task, all students eventually came upontheir own name in the list of included names. Respondents weregiven the following instructions concerning how they should han-dle this element of the task: "You will also find your own name inthe list of names below. When you get to your own name, pleasetry to imagine how the other students in your grade are answeringthis question about you, and then circle the appropriate number."The answer that students provided on this item as it appeared inthe sociometric task concerning social status and prestige was takenas their self-rated reputation. The answer that students providedon this item as it appeared in the sociometric task concerning socialpreferences was taken as their self-rated likability.

Although not every student enrolled in the school participatedin the present study, those students who did participate were pre-sented with an unabridged list of the students in their grade. Conse-quently, the round-robin rating matrix obtained was asymmetrical;students provided more ratings to peers than they received fromthem. To restore symmetry to the round-robin ratings, only those

peers participate as nominators in the sociometric procedure (Crick & Ladd, 1990).Unfortunately, logistical issues render it essentially impossible to obtain comparablyhigh participation rates among adolescents (for a detailed discussion, see Prinstein,2007). Terry, Coie, Lochman, and Cillessen (1998) posited that a randomly selectedsubsample of adolescent nominators would be able to provide a reliable estimateof peer status even when data from as much as 75% of the sampling populationwere unavailable. Consistent with this proposal, Prinstein (2007) recently found thatsubsamples of adolescent nominators representing only 10% of the full sample, selectedeither randomly by the researchers or specifically by teachers for their popularity andsocial expertise, were able to provide valid estimates of their peers' popularity. Thesefindings suggest that it is acceptable to proceed with approximately one third of thesampling population, even if the recruitment strategy has produced a relatively popularsubsample of adolescents.

Page 12: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

ADOLESCENT HIERARCHY EORMATION 1155

students who served as both rater and target were retained in theanalyses reported.

Attachment. Students' levels of attachment security were as-sessed through the 75-item Inventory of Parent and Peer Attach-ment (IPPA; Armsden & Greenberg, 1987). The inventory consists ofthree 25-item subscales that respectively assess the extent to whichrespondents feel that they can rely upon their mothers, their fathers,and their close friends. The IPPA has proven itself capable of dis-tinguishing depressed from nondepressed adolescents (Armsden,McCauley, Greenberg, Burke, & Mitchell, 1990). Coefficient alphasfor the mother, father, and peer subscales equaled .91, .92, and .90,respectively.

Social Support. Students' levels of social support were assessedthrough the 40-item Perceived Social Support Scale (PSSS; Procida-no & Heller, 1983). The scale consists of two 20-item subscales thatassess respondents' perceived level of social support from friendsand family. The PSSS has been shown to predict depressive symp-tomatology across disparate samples (Lyons, Perrotta, & Hancher-Kvam, 1988). Coefficient alphas for the family and friends subscalesequaled .84 and .79, respectively.

Self-Esteem. Students' levels of self-esteem were assessed throughthe Rosenberg Self-Esteem Scale (Rosenberg, 1965). The scale asksrespondents to indicate the extent to which they agree with 10 state-ments (e.g., I feel that I am a person of worth") on a four-point scalefrom "strongly agree" to "strongly disagree." Researchers have con-sistently demonstrated self-esteem to be significantly and inverselycorrelated with depression (Rosenberg, Schooler, & Schoenbach,1989). Coefficient alpha for the scale equaled .86.

Depression. Students' levels of depressive symptomatology (e.g.,feelings of sadness, guilt, and sleep disturbance) were assessedthrough the Beck Youth Depression Inventory (BDI-Y; Beck, Beck,Jolly, & Steer, 2005). The BDI-Y presents respondents with 20 state-ments, and asks them to indicate on a four-point scale from "never"to "always" how often the statement has been true for them dur-ing the past two-week period. Coefficient alpha for the inventoryequaled .91.

The BDI-Y was developed and normed using standardizationsamples of American youth stratified to match the U.S. census. Rawscores are first converted to T scores that can then be compared to

Page 13: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

1156 FOURNIER

TABLE 1. Descriptive Statistics

IPPA Attachment Security (Mother)

IPPA Attachment Security (Father)

IPPA Attachment Security (Peers)

PSSS Social Support (Family)

PSSS Social Support (Friends)

Peer-Rated Reputation

Self-Rated Reputation

Peer-Rated Likability

Self-Rated Likability

Self-Esteem

Depression

M

3,88

3,73

3,72

13.49

12,02

2,71

3,54

3.14

4,07

3,22

49,56

SD

0,60

0,68

0,55

4,41

4,22

0,67

0,95

0,52

0.90

0,48

7,97

Range

2,33-5,00

2,28-5,00

2,08-5,00

1,00-20,0

0,00-20,0

1,48-4.29

1,00-5,00

1,80-4,15

1,00-5,00

1,90 - 4,00

35,0-77,0

Note. IPPA = Inventory of Parent and Peer Attachment; PSSS = Perceived Social Support Scale,

the responses characteristic of the U.S. population of children andadolescents. The BDI-Y was thus scored by first calculating a rawscore for each student from the sum of his scores on each of the 20statements; raw scores could range from 0 to 60. Raw scores werethen converted to their T score equivalents using the sex- and age-appropriate conversion tables provided in the scoring manual.

RESULTS

DESCRIPTIVE STATISTICS

Table 1 provides the descriptive statistics for all variables. Mean lev-els of depression on the Beck Youth Depression Inventory equaled49.56 {SD = 7.97); these levels were thus quite comparable to thoseobtained from the normative sample (M = 50, SD = 10). As is oftenthe case when self-reports and peer-reports are compared (Taylor& Brown, 1988), adolescents appraised their own reputation (M -3.54) and likability (M - 4.07) to be higher than others on averageappraised it to be (reputation: M = 2.71; likability, M = 3.14).

SOCIAL RELATIONS ANALYSIS

The variance in ratings of reputation and likability were partitionedaccording to Kenny's social relations model (1994) to examine basic

Page 14: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

ADOLESCENT HIERARCHY FORMATION 1157

TABLE 2. A Social Relations Analysis of Reputation and Likability

Variance Component Reputation Likability

Perceiver .21 .31

Target .29 .19

Residual .50 .50

Dyadic Reciprocity

Correlation .10 .49

Note. The perceiver, target, and residual variances are relative and therefore sum to unity. Residual =relationship (perceiver X target interaction) variance plus error variance. Dyadic reciprocity = residualcovariance divided by residual variance.

perceptual processes (i.e., consensus, reciprocity) in students' rat-ings. These analyses were intended to demonstrate that reputationand likability, though empirically correlated, nevertheless behavelike distinct social psychological constructs.

Kenny (1994) has shown that ¿'s impression of; can be understoodas a function of a perceiver effect, reflecting how i generally seesothers, a target effect, reflecting how others generally see/, and a re-lationship effect, reflecting i's unique or idiosyncratic impression of; (that aspect of ¿'s impression of; that remains after controlling forperceiver and target effects). To separate relationship effects fromerror, multiple observations of / must be obtained from /; so, in thepresent analyses, error cannot be separated from relationship.

A social relations analysis of round-robin data allows the research-er to examine the interpersonal processes of assimilation, consen-sus, and reciprocity. Assimilation addresses the question, "Does ¿see /' and k in the same way?" The proportion of the total varianceattributable to perceiver variance indexes assimilation. Consensusaddresses the question, "Is i seen by / and k in the same way?" Theproportion of the total variance attributable to target variance in-dexes consensus. Reciprocity addresses the question, "Does i see; inthe same way that;' sees ¿?" Kenny (2006) has outlined the statisticalprocedures for indexing dyadic reciprocity.

Following Kenny (2006), a multilevel analysis of the sociomet-ric ratings was conducted. Models were fit using the SAS MIXEDprocedure. Ratings within grades were simultaneously nested bothwithin perceivers and within targets, and so the cross-classifiedmultilevel structure of the data was represented by way of twoRANDOM statements; one allowed intercepts to vary across per-ceivers (nested within grades), and another allowed intercepts tovary across targets (nested within grades). The DDFM = SATTERTH

Page 15: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

1158 FOURNIER

option was used to specify a general Satterthwaite approximationfor the denominator degrees of freedom. The dyadic covariance(nested within grades) was modeled via the REPEATED statement,and the TYPE = CS (compound symmetry) option was used for theerror covariance structure.

The results of these analyses for both rating tasks can be foundin Table 2. Perceiver effects accounted for a larger portion of the to-tal variance in likability (31%) than in reputation (21%), indicatinghigher levels of assimilation (the tendency for one to see others sim-ilarly), unexpectedly, for likability than for reputation. Target effectsaccoiinted for a larger portion of the total variance in reputation(29%) than in likability (19%), indicating higher levels of consen-sus (the tendency for others to see one similarly), as expected, forreputation than for likability. Following the statistical proceduresoutlined by Kermy (2006) to obtain an index of the degree of dyadiccorrelation (the extent to which i's unique impression of; correlateswith;'s uruque impression of i), we see that reciprocity effects werefive times larger for likability (r = .49) than for reputation (r = .10);ratings of liking were, as expected, reciprocated far more frequentlythan were ratings of prominence, respect, and influence.

PROPOSITION I. HIERARCHY FORMATION IS NATURAL

It was expected that students would be capable of recognizing thehierarchical features of their social ecology, and would differentiallyrate their peers in terms of their prominence, respect, and influence.As expected, students displayed considerable variability in theirpeer-ratings of classmates' reputation and likability. For each of thesociometric rating tasks, the range (maximum score - minimumscore) was calculated for each student's ratings across classmates(the potential range of each scale was 4). Students' peer-ratings oftheir classmates' reputation yielded a mean within-perceiver rangeof 3.37, SD = .73, range = 2 to 4, f = 49.14, p < .0001. Students' peer-ratings of their classmates' likability yielded a mean within-perceiv-er range of 3.42, SD = .80, range = 0 to 4, f = 45.83, p < .0001. Studentsthus saw sizable variation in their social ecology. As can be seen inTable 3, this effect was evident in all five grades.

It was expected that the hierarchical features of the adolescent hi-erarchy would be consensually accepted among students; in otherwords, that students would agree on those who hold positions of

Page 16: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

ADOLESCENT HIERARCHY FORMATION 1159

TABLE 3. Within-Perceiver Ranges by Grade for Reputation and Likability

Grade

Reputation

7

8

9

10

11

Likability

7

8

9

10

11

Mean Range

3.53

3.29

3.59

3.19

3.21

3.47

3.41

3.47

3.29

3.45

SD

.64

.77

.67

.81

.73

.74

.94

.88

.85

.63

Range

2 to 4

2 to 4

2 to 4

2 to 4

2 to 4

2 to 4

1 to 4

Oto 4

1 to 4

2 to 4

f

21.38*"

17.60***

30.56***

17.97***

23.79***

18.07*"

14.98"*

22.32***

17.82***

29.40***

Note, The ¡-test indicates whether or not the mean differs significantly from zero. The potential range ofthe reputation and likability scales was four. *p < ,05; " p < .01 ; »"p < .001.

prominence, respect, and influence in their grade. Given that socialrank is acquired by appearing socially attractive to one's peers, itwas expected that raters would also agree on those that they see aslikable in their grade. Agreement was estimated using coefficientalpha, calculated separately for each grade. In contrast to Kenny'sconsensus coefficient, which is analogous to the expected correlationbetween two individual students rating the same target, coefficientalpha is analogous to the expected correlation between two groupsof students (i.e., two halves of one grade) rating the same target. Ascan be seen in Table 4, students' peer-ratings of both reputation andlikability demonstrated a considerable level of consistency acrossraters. Students strongly agreed upon those who hold positions ofprominence, respect, and influence in their grade; coefficient alphasfor peer-ratings of reputation ranged from .82 to .96 across grades.Students also tended to converge in their social preferences; coef-flcient alphas for peer-ratings of likability ranged from .65 to .88across grades.

It was expected that students would be able to gauge their ownstanding in the social ecology. Social competition theorists pre-sume that competitors are typically able to estimate their socialrank (Fournier et al., 2002) and their social attractiveness (Gilbertet al., 1995), and that such abilities are necessary if competitors areto make appropriate decisions about whether to escalate or de-es-

Page 17: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

1160 EOURNIER

TABLE 4. Between-Target Reliabilities by Grade for Reputation and Likability

Grade Reputation Likability

7 .87 .75

8 .87 .80

9 .96 .88

10 .82 .65

11 .95 .88

Note. Reliabilities were calculated using Cronbach's coefficient alpha.

caíate in rank contests. This question was examined by correlatingstudents' aggregated peer-ratings with their own estimates of howother students were rating them along these evaluative dimensions.Students' estimates of their own likability were modestly but signif-icantly correlated with the aggregated peer-ratings they received,r = .21, p < .05, suggesting that students are indeed aware of howwell liked they are by their peers, albeit only to a limited extent. Incontrast, there was a much stronger correlation between students'estimates of their own reputafion and the aggregated peer-ratingsthey received, r = .51, p < .0001, suggesting that students are, asthe social compefifion theorists would expect, highly attuned to theprominence, respect, and inñuence they hold in their social ecology.A test of the difference between these two correlafions confirmedthat students were significantly more capable of estimating theirprominence, respect, and infiuence than they were of estimatingtheir likability in the adolescent social ecology, z = 2.54, p < .01.

PROPOSITION IL HIERARCHY FORMATION IS PREDICTABLE

The second set of hypotheses concerned the proposition that stand-ing in the human hierarchy is predicated on the ability to appear so-cially attractive to others. It was expected that students' evaluationsof reputafion and likability would be significantly and posifivelycorrelated, and that this relation would be evident both in terms ofthe evaluations that each student provided (i.e., the student as per-ceiver) and in terms of the evaluations that each student received(i.e., the student as target). As expected, the average within-perceiv-er correlation between reputafion and likability was significantlygreater than zero, mean within-perceiver r = .56 (95% CI = .50, .61),range =:= -.63 to .91, í = 14.65, p < .0001. As can be seen in Table 5, thiseffect was evident in all five grades. Also as expected, the aggregat-

Page 18: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

ADOLESCENT HIERARCHY FORMATION 1161

TABLE 5. Correlations by Grade Between Reputation and Likability

Grade

7

8

9

10

11

Mean r

,61

,52

,55

,46

,61

Witbin-Perceiver

95% C.I.

,48, ,72

,28, ,71

,44, ,65

,28, ,60

,54, ,68

Range

-,11 to ,88

-,63 to .86

-,62 to ,91

-,35 to ,88

-,09 to ,87

f

6,93***

3,62**

7,77***

4,83***

11,52***

Between-Target

r

,76***

,68**

,74***

,60**

,85***

Note. Within-perceiver correlations were subjected to Fisher's r-to-z transformation before being aver-aged together by grade; (-tests indicate whether the z-transformed means differ significantly from zero.Mean rs and 95% confidence intervals (C,l,) were back-transformed, z-to-r, before being reported here.*p<,05 ; * *p<.01 ; * * *p<,001.

ed peer-ratings of reputation and likability were significantly andpositively correlated, r = .76, p < .0001. As can be seen in Table 5, thiseffect was evident in all five grades. The between-target correlationappears to be somewhat stronger than the average within-perceivercorrelation, suggesting that the consensus among students concern-ing a classmate's likability is a better predictor of his prominence,respect, and influence than is the liking score obtained from onestudent alone.

It is important to note that not all students were well character-ized by the tendency to relate prominence and influence to likabil-ity. Although most adolescents did have strong positive relationsbetween their ratings of classmates' reputation and likability, oth-ers demonstrated a substantial negative relation between these twodimensions of evaluation. It was of interest to know whether thetendency to see high-ranking classmates as socially attractive wasin some way related to students' social standing. Perhaps thosestudents who hold the least social rank in the eyes of their peers,and/or those students who are found by their peers to be the leastsocially attractive, would in turn be those students whose ratingsof classmates' reputation and likability are the least positively cor-related.

This was indeed the case. Multilevel analyses confirmed thataggregated peer-ratings moderated the within-perceiver relationbetween reputation and likability. Models were fit using the SASMIXED procedure. Students' ratings of their classmates' likability

Page 19: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

1162 FOURNIER

5 1

¿.4-

3 •

2 -

Low High

Ratings of Classmates' Reputation

I Reputation Low Reputation High |

FIGURE 1. Plot showing how the tendency to report liking for thoseclassmates who are prominent, respected, and influential varies asa function of one's own peer-rated reputation. "Low" and "High"indicate 1 SD below and above the sample mean, respectively.

were treated as the DV, students' ratings of their classmates' rep-utation were treated as a Level-1 (within-person) predictor, andstudents' peer-rated reputation and likability scores were treatedas Level-2 (between-person) predictors. Level-1 reputation scoreswere centered within-persons, and a random intercept was includ-ed. When Level-2 predictors were entered into the model separate-ly, significant cross-level interaction effects with Level-1 reputationwere found for aggregated peer-ratings of reputation, b - .10, F(l,2573) = 34.40, p < .0001, and for aggregated peer-ratings of likabil-ity, b = .13, F(l, 2573) = 60.83, p < .0001. As can be seen in Figure1 and Figure 2, those students who were seen by their classmatesas the least influential and the least liked were in return less likelyto express strong social preferences for their most prominent class-mates.

Page 20: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

ADOLESCENT HIERARCHY FORMATION 1163

PROPOSITION III. HIERARCHY FORMATIONIS CONSEQUENTIAL

The third set of hypotheses concerned the incremental validity ofthe adolescent reputation construct. Although adolescent reputa-tion is likely to correlate with other individual difference charac-teristics (attachment security, social support) related to adjustmentoutcomes like self-esteem and depression, these correlations shouldnot be so large as to suggest that adolescent reputation can be ex-plained in terms of these other forms of individual difference. Fur-thermore, over and above the variance explained by empirically es-tablished risk factors like attachment insecurity, low social support,and interpersonal rejection, adolescent reputation should contrib-ute significantly to the prediction of self-esteem and depression forthis construct to demonstrate incremental validity.

Adolescent reputation evidenced appropriate discriminant valid-ity from attachment security and social support. Although correlat-ed with two of the three subscales of the IPPA (mother: r = .11, p = ns;father: r = .22, p < .05; peers: r = .20, p < .05) and with both subscalesof the PSSS (family: r = .16, p < .10; friends: r = .24, p < .01), thesefive subscales together explained only 10% of the variance in peer-rated reputation. Furthermore, adolescent reputation evidenced ap-propriate incremental validity. Stepwise multilevel analyses wereconducted. Both self-esteem and depression were predicted fromthe three subscales of the IPPA, from the two subscales of the PSSS,and from peer-rated likability, all variables known to be correlatedwith socio-emotional adjustment. Self-esteem and depression weremodeled as nested within grades, and all predictors were centeredwithin grades. Over and above the variance explained by these pre-dictors, peer-rated reputation still contributed significantly to theprediction of self-esteem, b = .31, f = 9,76, p < .01, and depression, b= -3.50, F = 4.14, p < .05. There was thus evidence for the discrimi-nant validity of adolescent reputation as well as for its incrementalvalidity in predicting self-esteem and depression over and abovethe variance explained by other empirically established correlates.

DISCUSSION

Social competition theorists state that (a) social hierarchies formspontaneously when individuals are required to compete over

Page 21: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

1164 FOURNIER

5 1

¿•4 -

3 -

Low High

Ratings of Classmates' Reputation

I Likability Low Likability High |

FIGURE 2. Plot showing how the tendency to report liking for thoseclassmates who are prominent, respected, and influential varies as afunction of one's own peer-rated likability. "Low" and "High" indicate1 SD below and above the sample mean, respectively.

scarce resources, (b) standing in the human social hierarchy is pred-icated in part on the ability to attract and hold positive social atten-tion, and (c) positions of low social rank can contribute to feelingsof entrapped defeat that can render the individual more vulner-able to low self-esteem and depression. These three propositionswere examined in a sample of adolescent boys. Support for all threepropositions was obtained here, in the first study to test the socialcompetition theory of depression in an adolescent sample with bothself- and peer-reports of the key constructs (social rank, social at-tractiveness).

Page 22: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

ADOLESCENT HIERARCHY EORMATION 1165

SOCIAL RELATIONS ANALYSIS

Round-robin ratings of reputation and likability were collected andanalyzed in accordance with the social relations model (Kenny,1994). Even though reputation and likability were strongly corre-lated both within perceivers and between targets, the two ratingsscales were found to behave quite differently, perhaps because lik-ability was framed at the level of the dyad (i.e., "Does / like hang-ing out with /'?"), whereas reputation was framed at the level of thegroup (i.e., "Does / see ; as prominent, respected, and infiuential inthe grade?"). Consistency across perceivers was greater for repu-tation than for likability, indicating higher levels of consensus forcharacteristics relevant to group interaction than for characteristicsrelevant to dyadic interaction. In contrast, consistency within dyadswas five times greater for likability than for reputation, suggestinghigher levels of reciprocity for characteristics relevant to dyadicinteraction than for characteristics relevant to group interaction.Adolescents were thus clearly not using the two rating scales in-terchangeably, suggesting that the strong positive correlation ob-served between reputation and likability was not due to the adop-tion of response sets.

SPONTANEOUS HIERARCHY FORMATION

Hierarchy formation was observed in the adolescent social ecology.Three sources of evidence corroborated this. First, adolescents werecapable of indicating who held positions of prominence, respect,and inñuence in the hierarchy. Second, adolescents demonstratedan impressive level of agreement concerning who held such posi-tions in the hierarchy. Third, adolescents knew where they them-selves stood in the hierarchy. Hierarchy was thus clearly evident,both in adolescents' private social cognition and in their public so-cial reality.

It is interesting to note that, in comparison to the strong positivecorrelation obtained between self-rated and peer-rated reputation,the correlation between self-rated and peer-rated likability was con-siderably more modest. While adolescents appear to be, as the so-cial competition theorist would expect, highly attimed to the promi-nence, respect, and influence they hold in their social ecology, ado-

Page 23: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

1166 FOURNIER

lescents appear to be only partially aware of how well liked they areby their peers. This raises the question of whether there is perhapsmore adaptive value in knowing one's social rank in the hierarchythan one's likability and general level of inclusion (c.f. Leary, Haupt,Strausser, & Chokel, 1998; Leary, Tambor, Terdal, & Downs, 1995).However, it is also possible that when adolescents are asked to en-gage in reflected-appraisal processes, particular friendships and/orenemyships become unusually salient and either inflate or deflatetheir self-estimated likability. If adolescents had been asked to ratenot only how much they like each of their classmates, but also howmuch they think each of their classmates likes them, then perhapsthe aggregated self-ratings obtained from the second rating taskwould have more closely approximated the aggregated peer-ratingsobtained from the first. Kenny (1994) found that individuals can in-dicate how well liked they are, occasionally with substantial accu-racy. Future research should continue to examine this question.

COMPETITION THROUGH ATTRACTION

Hierarchy formation was expected to emerge on the basis of indi-vidual differences in the capacity to attract and hold positive socialattention from others, assessed in terms of likability. Reputation (so-cial rank) and likability (social attractiveness) were sigrüficantly andpositively correlated, and this was evident both from the perceiver'spoint of view, in the impressions of reputation and likability that eachadolescent provided, and from the target's point of view, in the im-pressions of reputation and likability that each adolescent received.

Given the correlational design utilized, it is not possible to teaseapart the causal direction in these findings. Perhaps, as predictedby social competition theory, adolescents indeed achieved promi-nence, respect, and influence by being likable. However, we alsoknow that it is the dominant child who is the object of attention andtarget for imitation with other children (e.g., Abramovitch, 1976;Abramovitch & Grusec, 1978; Abramovitch & Strayer, 1978); con-sequently, perhaps those adolescents who achieved prominence,respect, and influence thereafter became the preferred interactionpartners among the "wannabes" and "outcasts" (Adler & Adler,1998) around them. However, had this alternative been the case, wewould have expected that those adolescents who were the least in-fluential and the least liked would have been those to express the

Page 24: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

ADOLESCENT HIERARCHY EORMATION 1167

strongest social preferences for their higher-ranking classmates; in-stead, the opposite was in fact the case. Those adolescents who werethe least influential and the least liked were less likely to expressstrong social preferences for their higher-ranking classmates. In fu-ture research, perhaps social network analysis (e.g., Wasserman &Faust, 1994) could be used to distinguish between those high likingscores that come from close friends within the high-ranking adoles-cent's social circle, and those that come from "warmabes" and "out-casts" (Adler & Adler, 1998) who fall outside the popular cliques.

DEFEAT AND DEPRESSION

Hierarchy formation was expected to help explain the variationacross adolescents in levels of self-esteem and depression. Socialcompetition theorists conceptualize self-esteem as an indicator andregulator of challenge and confidence, and depression as the outputof the IDS, triggered by the appraisal that one has been trapped anddefeated. As expected, low social rank was found to predict bothlow levels of self-esteem (indicating the individual's low RHP/SAHP) and high levels of depression (indicating the triggering ofthe involuntary subordinate response) in the social ecologies towhich most adolescents are confined (i.e., high school).

These findings encourage an alternative evolutionary account ofself-esteem with which to complement the sociometer hypothesis(Leary et al., 1995). Leary and colleagues have argued that self-esteem is the output of an evolved psychological apparatus thatserves to monitor cues of acceptance and inclusion vs. rejection andexclusion: high levels of self-esteem indicate feeling accepted andincluded; low levels of self-esteem indicate feeling rejected and ex-cluded. In the present sample, however, levels of depression andself-esteem were found to correlate with peer-ratings of reputation,not likability. These findings suggest that there is more to adolescentself-esteem than feelings of belonging, and that once adolescentsbecome part of a group, concerns for status and esteem emerge thatbear directly on their own self-worth. It may be that the sociometerand social competition hypotheses speak to the different concernsthat preoccupy the individual at different stages of intra-group rela-tions: that when adolescents are attempting to join some group andtheir membership status is still uncertain, adolescent self-esteem isprimarily contingent on cues of belonging; and that later, once their

Page 25: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

1168 EOURNIER

group membership has been established and the matter of "fittingin" is no longer of great concern, adolescent self-esteem becomesincreasingly contingent on cues of prominence and influence.

LIMITATIONS AND FUTURE DIRECTIONS

The present investigation was the first to test the social competitiontheory of depression with both self-report and round-robin ratings.Although support for the basic propositions of the theory was ob-tained, future research should address the three key limitations tothe present study: its gender composition, its research design, andits recruitment strategy.

The present investigation benefited from using a same-sex sam-ple, given the complex interpersonal d)mamics that emerge be-tween adolescent boys and girls in coed institutions. Still, it remainsto be seen whether the same effects would be obtained within anall-girls school. Subsequent research with both sex-segregated andcoed samples will be needed to determine the generalizability of thepresent findings.

The present investigation utilized a cross-sectional design fromwhich we may infer correlation, not causation. For instance, thecorrelation obtained between reputation and likability could reflectthe impact being likable has on one's reputation, the impact reputa-tion has on one's social desirability, the influence of both processessimultaneously, or some unknown process. Longitudinal researchwill be needed to isolate and identify distinct causal processes.

One final limitation concerns the proportion of students who par-ticipated (i.e., one third). Higher rates of participation would allowone to identify clique structures in the social ecology. Cliques allhave their own hierarchical structure, with leaders and followers;in turn, cliques are organized hierarchically, such that some cliquesare more prominent and influential than others. There may well bestudents at the periphery of the popular cliques whose risk for de-pression goes underestimated when the organization of studentsinto cliques is not taken into account. Future research in which alarger proportion of the student population is recruited and sur-veyed would be able to identify these social substructures and theirsocio-emotional implications.

Page 26: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

ADOLESCENT HIERARCHY FORMATION 1169

CONCLUSION

Scarcity breeds competition. However, rather than competing forsocial rank and status through intimidation, the objective of whichis to produce states of negative affect in others (fear), humans moreoften seem to compete for social rank and status through attraction,the objective of which is to produce states of positive affect in oth-ers. Social competition theory thus introduces the critical distinc-tion between the physical capacity to hold one's resources (RHP)and the social capacity to hold others' positive attention (SAHP).Among adolescent boys, standing in the social hierarchy would ap-pear to depend on the ability to present oneself as likable; i.e., assomeone to be admired, valued, and chosen as a friend or ally. Thecosts of failing to be seen as likable and socially attractive (valuable)are emotionally steep for the adolescent boy, and can be seen in hislowered levels of self-esteem and increased vulnerability to depres-sion.

REFERENCES

Abramovitch, R. (1976). The relation of attention and proximity to rank in preschoolchildren. In M. R. A. Chance & R. R. Larsen (Eds.), The social structure of atten-tion (pp. 153-176). London: Wiley.

Abramovitch, R., & Grusec, J. E. (1978). Peer imitation in a natural setting. ChildDevelopment, 49, 60-65.

Abramovitch, R., & Strayer, F. (1978). Preschool social organization: Agonistic, spac-ing, and attentional behaviors. In L. Krames, P. Planer, & T. Alloway (Eds.),Aggression, dominance, and individual spacing (pp. 107-128). New York: Ple-num.

Adler, P. A., & Adler, P. (1998). Peer power: Préadolescent culture and identity. NewBrunswick, NJ: Rutgers University Press.

Allan, S., & Gilbert, P. (1997). Submissive behaviour and psychopathology. BritishJournal of Clinical Psychology, 36, 467-488.

Allan, S., & Gilbert, P. (2002). Anger and anger expression in relation to perceptionsof social rank, entrapment and depressive symptoms. Personality and Indi-vidual Differences, 32, 551-565.

Anderson, C , John, O. P., Keltner, D., & Kring, A. M. (2001). Who attains social sta-tus? Effects of personality and physical attractiveness in social groups. Journalof Personality and Social Psychology, 81,116-132.

Armsden, G. C , & Greenberg, M. T. (1987). The inventory of parent and peer attach-ment: Relationships to well-being in adolescence. Journal of Youth and Adoles-cence, 16,427-454.

Page 27: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

1170 EOURNIER

Armsden, G. C , McCauley, E., Greenberg, M. T., Burke, P. M., & Mitchell, J. R.(1990). Parent and peer attachment in early adolescent depression. Journal ofAbnormal Child Psychology, 18, 683-697.

Bass, B. M. (1954). The leaderless group discussion. Psychological Bulletin, 51,465-492.

Beck, J., Beck, A., Jolly, J., & Steer, R. A. (2005). Beck youth inventories for children andadolescents (2nd ed.). San Antonio, TX: Psychological Corporation.

Bernstein, I. S. (1981). Dominance: The baby and the bathwater. The Behavioral andBrain Sciences, 4,419-457.

Billings, A. G., Cronkite, R. C , & Moos, R. H. (1983). Social-environmental factors inunipolar depression: Comparisons of depressed patients and nondepressedcontrols. Journal of Abnormal Psychology, 92,119-133.

Brown, G. W., Harris, T., & Hepworth, C. (1995). Loss, humiliation, and entrapmentamong women developing depression: A patient and non-patient compari-son. Psychological Medicine, 25, 7-21.

Cheung, M. S.-P, Gilbert, P., & Irons, C. (2004). An exploration of shame, social rankand rumination in relation to depression. Personality and Individual Differences,36,1143-1153.

Coie, J. D., & Dodge, K. A. (1993). Continuities and changes in childrenis social sta-tus: A five-year longitudinal study. Merrill-Palmer Quarterly, 29, 261-282.

Coie, J. D., Dodge, K. A., & Cappotelli, H. (1982). Dimensions and types of socialstatus: A cross-age perspective. Developmental Psychology, 18, 557-570.

Crick, N. R., & Ladd, G. W. (1990). Nominator attrition: Does it affect the accuracy ofchildrenis sociometric classifications? Merrill-Palmer Quarterly, 35,197-207.

Fournier, M. A., Moskowitz, D. S., & Zuroff, D. C. (2002). Social rank strategiesin hierarchical relationships. Journal of Personality and Social Psychology, 83,425-433.

Gilbert, P. (1992). Depression: The evolution of powerlessness. Hillsdale, NJ: Erlbaum.Gilbert, P. (1997). The evolution of social attractiveness and its role in shame, hu-

miliation, guilt and therapy. British Journal of Medical Psychology, 70,113-147.Gilbert, P. (2000). Varieties of submissive behavior as forms of social defense: Their

evolution and role in depression. In L. Sloman & P. Gilbert (Eds.), Subordina-tion and defeat: An evolutionary approach to mood disorders and their therapy (pp.3-45). Mahwah, NJ: Erlbaum.

Gilbert, P. (2001a). Depression and stress: A biopsychosocial exploration of evolvedfunctions and mechanisms. Stress, 4,121-135.

Gilbert, P. (2001b). Evolutionary approaches to psychopathology: The role of natu-ral defences. Australian and New Zealand Journal of Psychiatry, 35,17-27.

Gilbert, P, & Allan, S. (1998). The role of defeat and entrapment (arrested flight) indepression: An exploration of an evolutionary view. Psychological Medicine,28, 585-598.

Gilbert, P., Gilbert, J., & Irons, C. (2004). Life events, entrapments and arrested angerin depression. Journal of Affective Disorders, 79,149-160.

Gilbert, P., Price, J. S., & Allan, S. (1995). Social comparison, social attractiveness andevolution: How might they be related? New Ideas in Psychology, 13,149-165.

Harms, P D., Roberts, B. W., & Wood, D. (2007). Who shall lead? An integrative per-sonality approach to the study of the antecedents of status in informal socialorganizations. Journal of Research in Personality, 41, 689-699.

Page 28: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

ADOLESCENT HIERARCHY FORMATION 1171

Hawley, P, H,, & Little, T, D, (1999), On winning some and losing some: A socialrelations approach to social dominance in toddlers, Merrill-Palmer Quarterly,45,185-214,

Irons, C , & Gilbert, P, (2005), Evolved mechanisms in adolescent anxiety and de-pression symptoms: The role of the attachment and social rank systems. Jour-nal of Adolescence, 28,325-341,

Kalma, A, (1991), Hiérarchisation and dominance assessment at first glance, Euro-pean Journal of Social Psychology, 21,165-181,

Kendler, K, S,, Hettema, J, M,, Butera, F,, Gardner, C, O,, & Prescott, C, A, (2003), Lifeevent dimensions of loss, humiliation, entrapment, and danger in the predic-tion of onsets of major depression and generalized anxiety. Archives of GeneralPsychiatry, 60, 789-796,

Kenny, D, A, (1994), Interpersonal perception: A social relations analysis. New York:Guilford,

Kenny, D. A, (2006), Estimation of the SRM not using specialized software. RetrievedJuly 1, 2007, from http://davidakenny,net/srm/srm,htm,

LaFreniere, P, & Charlesworth, W, R, (1983), Dominance, attention, and affiliationin a preschool group: A nine-month longitudinal study. Ethology and Sodobiol-ogy, 4,55-67.

Leary, M, R,, Haupt, A, L,, Strausser, K, S,, & Chokel, J, T, (1998), Calibrating thesociometer: The relationship between interpersonal appraisals and state self-esteem. Journal of Personality and Social Psychology, 74,1290-1299,

Leary, M, R,, Tambor, E, S,, Terdal, S, K,, & Downs, D, L, (1995), Self-esteem as aninterpersonal monitor: The sociometer hypothesis. Journal of Personality andSocial Psychology, 68, 518-530,

Lyons, J, S,, Perrotta, P, & Hancher-Kvam, S, (1988), Perceived social support fromfamily and friends: Measurement across disparate samples. Journal of Person-ality Assessment, 52, 42-47,

Mischel, W,, & Grusec, J, (1966), Determinants of the rehearsal and transmission ofneutral and aversive behaviors. Journal of Personality and Social Psychology, 3,197-205,

Newcomb, A, F,, & Bukowski, W, M, (1983), Social impact and social preference asdeterminants of children's peer group status. Developmental Psychology, 19,856-867.

Newcomb, A, F,, Bukowski, W, M,, & Pattee, L, (1993), Children's peer relafions: Ameta-analytic review of popular, rejected, neglected, controversial, and aver-age sociometric status. Psychological Bulletin, 113, 99-128,

Nezlek, J, B,, Hampton, C, P, & Shean, G, D, (2000), Clinical depression and day-to-day social interaction in a community sample. Journal of Abnormal Psychology,109,11-19,

Noll, R, B,, Zeller, M, H,, Vannatta, K,, Bukowski, W, M,, & Davies, W, H, (1997),Potential bias in classroom research: Comparison of children with permissionand those who do not receive permission to participate. Journal of ClinicalChild Psychology, 26, 36-42,

Parker, G, A, (1974), Assessment strategy and the evolution of fighting behavior.Journal of Theoretical Biology, 47, 223-243,

Price, J, (2000), Subordination, self-esteem, and depression. In L, Sloman & P Gilbert(Eds,), Subordination and defeat: An evolutionary approach to mood disorders andtheir therapy (pp, 165-177), Mahwah, NJ: Erlbaum,

Page 29: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

1172 EOURNIER

Prinstein, M. J. (2007). Assessment of adolescents' preference- and reputation-basedpeer status using sociometric experts. Merrill-Palmer Quarterly, 53,243-261.

Procidano, M. E., & Heller, K. (1983). Measures of perceived social support fromfriends and from family: Three validation studies. American Journal of Com-munity Psychology, 11,1-24.

Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: PrincetonUniversity Press.

Rosenberg, M., Schooler, C , & Schoenbach, C. (1989). Self-esteem and adoles-cent problems: Modeling reciprocal effects. American Sociological Review, 54,1004-1018.

Savin-Williams, R. C. (1976). An ethological study of dominance formation and main-tenance in a group of human adolescents. Child Development, 47, 972-979.

Savin-Williams, R. C. (1977). Dominance in a human adolescent group. Animal Be-havior, 25,400-406.

Savin-Williams, R. C. (1979). Dominance hierarchies in groups of early adolescents.Child Development, 50, 923-935.

Sloman, L. (2000). How the involuntary defeat strategy relates to depression. In L.Sloman & P. Gilbert (Eds.), Subordination and defeat: An evolutionary approach tomood disorders and their therapy (pp. 47-67). Mahwah, NJ: Erlbaum.

Sloman, L., & Gilbert, P. (2000). Subordination and defeat: An evolutionary approach tomood disorders and their therapy. Mahwah, NJ: Erlbaum.

Sloman, L., Price, J., Gilbert, P., & Gardner, R. (1994). Adaptive function of depres-sion: Psychotherapeutic implications. American Journal of Psychotherapy, 48,401-416.

Strayer, F. R, & Strayer, J. (1976). An ethological analysis of social agonism and dom-inance relations among preschool children. Child Development, 47,980-989.

Strayer, J., & Strayer, R R (1977). Social aggression and power relations among pre-school children. Aggressive Behavior, 4,173-182.

Strayer, R R, & Trudel, M. (1984). Developmental changes in the nature and func-tion of social dominance among young children. Ethology and Sociobiology, 5,279-295.

Terry, R., Coie, J. D., Lochman, J., & Cillessen, A. H. N. (1998). Sampling issues insociometric assessment. Unpublished manuscript. Duke University, Durham,NC.

Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social psychologicalperspective on mental health. Psychological Bulletin, 103,193-210.

Vaughn, B. E., & Waters, E. (1981). Attention structure, sociometric status, and dom-inance: Interrelations, behavioral correlates, and relationships to social com-petence. Developmental Psychology, 17, 275-288.

Wasserman, S., & Faust, K. (1994). Social network analysis: Methods and applications.Cambridge: Cambridge University Press.

Zuroff, D. C , Fournier, M. A., & Moskowitz, D. S. (2007). Depression, perceivedinferiority, and interpersonal behavior: Evidence for the involuntary defeatstrategy. Journal of Social and Clinical Psychology, 26, 751-778.

Page 30: Adolescent Hierarchy Formation and the Social Competition Theory of Depression

Copyright of Journal of Social & Clinical Psychology is the property of Guilford Publications Inc. and its

content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's

express written permission. However, users may print, download, or email articles for individual use.