Aggressive and Prosocial Behaviors within Early Adolescent Friendship Cliques:What's Status Got to Do with It?
This study involved an examination of the role of perceived popularity and social dominance in the social behaviors used within early adolescents' (N = 387) friendship cliques. A status hierarchy between cliques within each grade (based on peer-rated perceived popularity) and a status hierarchy between individuals within each clique (based on friend-rated social dominance) were delineated. Results revealed that adolescents' within-clique dominance rank and their clique's collective perceived popularity status were independently related to their social behaviors used within their cliques and their likability by friends and peers. Levels of aggression were highest within perceived popular cliques and perceived unpopular cliques, whereas levels of prosocial behavior and friend/peer likability were lowest in perceived unpopular cliques. Findings also indicated that aggression toward clique members was associated with social dominance within the clique. However, dominant adolescents using both aggressive and prosocial behaviors within their cliques may be afforded the most social rewards.
Traditionally, developmental researchers have conceptualized high-status youths as liked, sociable, and nonaggressive and low-status youths as disliked, unsociable, and aggressive or withdrawn (Newcomb, Bukowski, & Pattee, 1993). However, recent work has begun to show that behavioral characteristics and outcomes associated with social status are not as clearcut as previously thought. For example, when considering measures of social status based in sociology (i.e., perceived popularity) and ethology [End Page 406] (i.e., social dominance), high status is often associated with aggressive behavior (Hawley, 2003; Prinstein & Cillessen, 2003; Rodkin, Farmer, Pearl, & Van Acker, 2000). In fact, researchers have argued that some aggression may be normative and beneficial to social adjustment (Hawley, 2003; Little, Brauner, Jones, Nock, & Hawley, 2003), especially if aggression is carefully balanced by the use of prosocial behavior (Hawley, 2003). Thus far, the social behaviors and experiences of such high-status children and adolescents have been quantitatively studied within the larger peer group context (e.g., Hawley, 2003; Lease, Kennedy, & Axelrod, 2002; Parkhurst & Hopmeyer, 1998; Prinstein & Cillessen, 2003; Rodkin et al., 2000; Savin-Williams, 1979) and, less often, within dyadic friendships (Hawley, Little, & Card, 2007).
Friendship Cliques and Social Status
Except for a few ethnographic studies (Adler & Adler, 1998; Merten, 1997), researchers have yet to explore associations between status and behavior used within friendship cliques. Unlike larger peer groups such as classmates or grademates at school, cliques are comprised of youths who voluntarily and frequently associate with one another and identify one another as friends (Rubin, Bukowski, & Parker, 1998). By late childhood/early adolescence, nearly all youths report being members of cliques (Rubin et al., 1998; Thompson, O'Neill Grace, & Cohen, 2001). Since the prevalence and social influence of cliques has been reported to peak during early adolescence (Thompson et al., 2001), the associations between social status and the dynamics within cliques may be particularly important to study at this stage of development.
Social status is a central aspect of daily life for early adolescents, as inclusion and acceptance by peers become strongly desired (Merten, 2004). During this time, social status exists within the larger peer network (e.g., classroom or grade) as well as within smaller social groups (e.g., cliques) and can reflect one's own social reputation or one's reputation based on with whom one affiliates (Cairns, Perrin, & Cairns, 1985). Both individuals within cliques and cliques within the larger peer network may be organized hierarchically (Adler & Adler, 1998). For purposes of this study, the hierarchy of cliques within the larger peer network (i.e., grade) was determined based on the collective status (i.e., peer-rated perceived popularity) of those who comprise each clique, and the hierarchy within each clique was determined based on the status (i.e., friend-rated social dominance) of each individual clique member. The primary aim of the present study was to examine the relationships between two measures of [End Page 407] social status (i.e., perceived popularity and social dominance) and early adolescents' experiences within their cliques.
Likability and Social Impact
Developmental psychologists have traditionally assessed social status using measures of peer liking and disliking (i.e., sociometric status). Sociometric measures allow individuals to be classified based on social preference (nominations of like most minus like least), and social impact (nominations of like most plus like least) (Coie, Dodge, & Coppotelli, 1982). Social preference is a measure of relative likability in the peer group, whereas social impact is a measure of visibility in that peers regard the child with strong opinions (Terry, 2000). Thus, those who score high on social impact are highly liked, highly disliked, or a combination of both. During early adolescence, both peer likability (i.e., social preference) and social impact have been positively associated with supportiveness, assertiveness, and physical attractiveness; however, social impact has also been positively associated with disruptiveness and aggression (Coie et al., 1982). Prior to the present study, research examining whether peer likability and social impact relate to measures of status within and between friendship cliques has been lacking.
Over the past decade, researchers have been turning their attention to assessments of status using measures of peer-rated social reputation (i.e., perceived popularity). Perceived popularity is often assessed by asking participants to nominate who is most and least popular (e.g., LaFontana & Cillessen, 1999; Lease et al., 2002) and is said to measure social prominence in the peer group (Parkhurst & Hopmeyer, 1998; Rodkin et al., 2000). Across Grades 4-10, perceived popularity has been positively associated with the use of prosocial behavior toward peers (Gorman et al., 2002; LaFontana & Cillessen, 2002; Rodkin et al., 2000). In comparison, positive associations between perceived popularity and both overt and relational forms of aggression toward peers have been most commonly reported across Grades 6-10 (LaFontana & Cillessen, 2002; Prinstein & Cillessen, 2003; Rose, Swenson, & Waller, 2004). It has been argued that the ability to maintain a position of high social status within the peer group using aggressive means requires keen interpersonal skills that develop with age (Rose, Swenson, & Waller, 2004). Furthermore, perceived popular youths are often described by peers as cool, nice, and funny yet also conceited, exclusionary, and mean (Adler & Adler, 1998; Closson, 2009; Eder, Evans, & Parker, 1995). Given their mixed profile, it is not surprising that perceived popular early adolescents [End Page 408] are not necessarily well liked by all peers (Farmer, Estell, Bishop, O'Neal, & Cairns, 2003; Parkhurst & Hopmeyer, 1998). Certain peers may admire their prosocial characteristics, whereas others may resent their social power and disapprove of their aggressiveness. Indeed, the prominence of perceived popular youths enables peers to notice their behaviors and form strong opinions of them. Therefore, perceived popular youths are likely to score high on social impact (Parkhurst & Hopmeyer, 1998).
Research findings in regard to gender differences in associations between peer-directed aggression and perceived popularity have been mixed. Using an early adolescent sample, one study demonstrated that overt aggression was more strongly linked with perceived popularity for boys than for girls (Parkhurst & Hopmeyer, 1998). Other studies have indicated that overt and relational aggression may be positively associated with perceived popularity for both boys and girls; however, the association involving relational aggression may be stronger for girls than for boys in early adolescence (Cillessen & Mayeux, 2004; Rose, Swenson, & Waller, 2004). Whether the association between perceived popularity and overt aggression within cliques is stronger for boys than for girls and the association between perceived popularity and relational aggression within cliques is stronger for girls than for boys remain empirical questions to be explored in the current study.
Although researchers are beginning to uncover a great deal about perceived popular youths, relatively little is known about those who are perceived as comparatively less popular. Using a preadolescent sample, Adler and Adler (1998) demonstrated that subordinate to the popular cliques there may be other groups such as middle friendship circles and social isolates. In Adler and Adler's qualitative study, children in middle friendship circles were described as nice to peers, with primarily positive relationships with friends. Comparatively, social isolates were those at the bottom of the peer status hierarchy, as they were generally rejected by the larger peer group but would sometimes befriend peers at school who were of similar low status. Similarly, Prinstein and Cillessen (2003) found that in Grade 10, peer-directed aggression was quantitatively associated with low levels of perceived popularity but to a lesser degree than the association with high levels of perceived popularity. Taken together, these findings suggest that youths in perceived average cliques may be the least aggressive individuals in the peer group.
Not only may a status hierarchy exist between cliques within the larger peer network, but a status hierarchy between individuals within cliques may also be present (Adler & Adler, 1998). According to Adler and Adler, at the top [End Page 409] of the hierarchy within cliques are leaders followed by those who are second tier and finally those who are followers. During preadolescence and early adolescence, the leaders within popular cliques have been found to use aggression toward socially threatening subordinate clique members to secure their own dominant position within the clique (Adler & Adler, 1998; Merten, 1997). The current study sought to determine whether these ethnographic findings can be replicated through quantitative measures and assess whether such behavior occurs within lower status cliques.
Using an ethology based conception of social dominance, researchers have studied the development and influence of status hierarchies by assessing the relative social dominance of individuals in a group (e.g., Hawley, 1999). According to this perspective, group members use aggression to compete for resources, thereby establishing and maintaining dominance hierarchies (Savin-Williams, 1979). Through competition, socially dominant individuals gain greater access than others to social resources (e.g., being deemed the leader of the group) or material resources (e.g., the preferred spot in the lunch cafeteria) that are highly desired by group members (Hawley, 1999). Relative to their subordinate peers, socially dominant youths have been characterized as more influential, daring, physically attractive, and socially appropriate; thus, they are often perceived as attractive social partners and are afforded high status (Hawley, 1999; Savin-Williams, 1979; Savin-Williams & Freedman, 1977).
The traditional view of social dominance suggests that dominant individuals use aggressive strategies to attain their goals. However, Hawley (1999) proposed that social dominance may be obtained by acquiring resources, regardless of the means by which they are attained. Hawley's resource control theory suggests that one's ability to compete for and control social or material resources determines each group member's status within the dominance hierarchy. According to Hawley and her colleagues (e.g., Hawley, 2003; Hawley et al., 2007; Hawley, Little, & Pasupathi, 2002), dominant children and adolescents tend to use one of three strategies of resource control: prosocial (e.g., reciprocity, cooperation, persuasion, and helping behaviors), coercive (e.g., aggression, manipulation, deception, insults, and threats), or bistrategic (i.e., the use of either prosocial or coercive strategies depending on what would bring the most success given the context of the situation).
During late childhood and early adolescence, prosocial dominance is associated with positive social characteristics and outcomes (e.g., perceived as popular, liked by peers, agreeable, conscientious, sensitive to social cues, socially connected, and positive affect) (Hawley, 2003; Hawley et al., 2002). Comparatively, coercive dominance is associated with aggression, [End Page 410] hostility, willingness to cheat, negative affect, and loneliness (Hawley, 2003; Hawley et al., 2002). Bistrategic dominants are socially competent, sharing characteristics and outcomes of prosocial dominants (e.g., perceived as popular, reasonably liked by peers, agreeable, conscientious, socially skilled, and socially integrated) as well as coercive dominants (e.g., high levels of hostility, loneliness, cheating, and aggression) (Hawley, 2003; Hawley et al., 2002). Bistrategic individuals report the highest need for recognition and highest influence over their peers (Hawley et al., 2002). Hawley and colleagues (2007) found that Grades 7-10 peer-rated bistrategic dominants were aggressive within their dyadic friendships. Despite the presence of aggression and conflict in their friendships, they were rated as intimate and fun. Interestingly, the study also showed that the resource control strategy employed with peers may not necessarily be the same strategy used within their friendship. For example, peer-rated prosocial dominants did not differ from coercive dominants or bistrategic dominants in the level of coercive behaviors used within their friendships. Therefore, dominance strategies may differ depending on the social context. As a study of early adolescents' experiences within cliques, the current study extends the extant literature by exploring dominance processes in a context not previously examined through quantitative measures.
Some research suggests that there may be gender differences in the peer-directed social behaviors of socially dominant early adolescents. For instance, in a study with youths at a summer camp, Savin-Williams (1979) found that verbal ridicule (i.e., verbal and relational aggression) was the most frequent expression of social dominance for both boys and girls. However, socially dominant boys used more physical aggression, whereas socially dominant girls tended to use more prosocial strategies. Similarly, Hawley (2003) found that more boys than girls used coercive dominance strategies and that more girls than boys used prosocial dominance strategies, whereas boys and girls were equally likely to be bistrategic. Although previous research suggests that there may be gender differences in dominance strategies used within the larger peer group, it is presently unknown whether these gender differences exist within the context of a friendship clique.
The Present Study
Qualitative researchers have demonstrated that status hierarchies may exist at two levels: between cliques within the larger peer network and between individuals within each clique (Adler & Adler, 1998). Drawing upon the extant literature to inform the methodology of the current study, I used quantitative measures of perceived popularity and social dominance to [End Page 411] delineate these two hierarchies. It was decided that within the larger peer network, cliques may be organized hierarchically based on the collective perceived popularity of the individuals who comprise each clique as rated by peers. It was also decided that within each clique, individuals may be organized hierarchically based on the level of social dominance of each individual within their clique as rated by clique members. The overarching aim of the current study was to identify these two types of hierarchies and quantitatively examine how social status may be associated with social behaviors and experiences within cliques.
Little is known about gender differences in the use of aggressive and prosocial behaviors within cliques. Merten (1997) described a popular clique of early adolescent girls as often being mean within the clique, yet they were also nice toward clique members as well as toward outsiders. Unfortunately, Merten's study did not involve a comparable clique of early adolescent boys. Adler and Adler's (1998) ethnography examining preadolescent clique dynamics suggested that boys and girls in popular cliques may behave in similar ways toward clique members, involving a balance of both aggressive and prosocial behaviors. In addition, there are no apparent differences in Adler and Adler's descriptions of the within-clique behaviors of boys and girls in lower status cliques. Therefore, in the present study it was expected that the relationship between social status and behaviors within cliques would be similar for boys and girls.
Peer-directed prosocial behavior has previously been linked with high perceived popularity (e.g., Gorman et al., 2002; LaFontana & Cillessen, 2002; Rodkin et al., 2000), whereas peer-directed aggression has previously been associated with high and low levels of perceived popularity (Prinstein & Cillessen, 2003). Furthermore, Adler and Adler (1998) described those in middle friendship circles as having positive relationships with friends. Therefore, it was expected that members of perceived popular cliques and perceived unpopular cliques would demonstrate more within-clique aggression than members of perceived average cliques. In addition, it was expected that members of perceived popular cliques and perceived average cliques would demonstrate more within-clique prosocial behavior than members of perceived unpopular cliques.
Adler and Adler (1998) found that individuals in middle friendship circles were nice to all peers, and their friendships were less exclusive and more intimate than those in perceived popular cliques. Therefore, those in perceived average cliques were expected to be more liked by peers and friends than those in perceived popular and perceived unpopular cliques. Furthermore, previous research has found that an individual's perceived popularity has been positively associated with peer social impact [End Page 412] during early adolescence (Parkhurst & Hopmeyer, 1998). Therefore, it was expected that individuals in perceived popular cliques would score higher on peer social impact than individuals in perceived average cliques and perceived unpopular cliques.
Previous research has associated peer-rated social dominance with peer-directed aggression (e.g., Savin-Williams, 1979) that is used for instrumental purposes (Hawley, 1999). Therefore, this study tested the hypothesis that high dominants would use more aggression, particularly instrumental aggression, within their clique than subordinate clique members. In addition, the least dominant adolescent has been found to be the least liked individual in a peer group (Savin-Williams, 1979; Savin-Williams & Freedman, 1977). Thus, this finding was expected to be replicated within the friendship clique context.
According to Hawley and colleagues (Hawley, 2003; Hawley et al., 2002), youths who employ both prosocial and coercive resource control strategies (i.e., bistrategic) are often rated high on perceived popularity. Based on this research, it was expected that high dominants who were bistrategic would be members of cliques with a high level of perceived popularity among members. In addition, Hawley (2003) has shown that peer-rated social dominants who are bistrategic or prosocial may be more liked by peers than those who are coercive. Therefore, it was expected that bistrategic dominants and prosocial dominants would be more liked by peers and friends than aggressive dominants.
Participants were 387 early adolescents (53% girls) from three middle schools comprised of Grades 6-8 in a small city in western Canada. The local population consists primarily of Caucasian families (97%) of middle socioeconomic status (Statistics Canada, 2006). The participation rate for the entire sample was 75%, exceeding the recommended minimum participation rate of 60% when utilizing unlimited sociometric nomination procedures (Nukulkij, Cillessen, Bellmore, Whitcomb, & Burke, 2002, as cited in Sandstrom & Cillessen, 2003).
Students who received parental permission to participate completed a series of questionnaires that were group administered in their classrooms during a [End Page 413] 45-minute session. As described below, students completed a number of self-report and peer-report measures.
Consistent with previous research (e.g., LaFontana & Cillessen, 2002), participants nominated an unlimited number of participating grademates they like most ("students who you personally like the most") and an unlimited number of peers they like least ("students who you personally like the least"). Nominations from other-gender peers and grademates outside a participants' classroom were allowed, as such procedures have been shown to improve the predictive validity of sociometric measures (Poulin & Dishion, 2008). Numbers of like-most and like-least nominations received for each participant were separately summed and standardized within grade and school. Like-most z-scores minus like-least z-scores were calculated, and the resulting score was restandardized within grade and school to create a continuous measure of peer likability for each participant. Like-most z-scores plus like-least z-scores were calculated, and the resulting score was restandardized within grade and school to create a continuous measure of peer social impact for each participant (Coie et al., 1982).
Participants nominated an unlimited number of participating grademates as most popular ("students who you think are most popular") and as least popular ("students who you think are least popular") (e.g., Cillessen & Mayeux, 2004; Lease et al., 2002; Prinstein & Cillessen, 2003). Participants were allowed to nominate other-gender peers and grademates outside their classroom. The number of most-popular and least-popular nominations received by each participant were each summed and standardized within grade and school. A continuous measure of perceived popularity for each participant was created by subtracting least-popular z-scores from most-popular z-scores, and the resulting score was restandardized within grade and school.
Classification of clique perceived popularity
Adapting the procedures described by LaFontana and Cillessen (1999), cliques were classified as perceived popular, perceived average, or perceived unpopular by assessing the collective perceived popularity of all members in a clique. The likelihood of individuals within each clique being similar in terms of perceived popularity is high, given that youths who are dyadic friends as well as those who hang around together have similar levels of perceived popularity (Farmer et al., 2003; Rose, Swenson, & Carlson, 2004). A criterion cutoff score of 1 standard deviation was used to tentatively classify cliques. Specifically, cliques whose members' mean perceived popularity score was [End Page 414] equal to or greater than 1.00 were classified as perceived popular, cliques whose members' mean perceived popularity score was equal to or less than -1.00 were classified as perceived unpopular, and cliques whose members' mean perceived popularity score was greater than -1.00 and less than 1.00 were classified as perceived average. Since it was possible that the scores of one or two clique members could skew a clique's mean perceived popularity score, it was important to examine the homogeneity of individual scores within each clique. Therefore, cliques whose members' mean perceived popularity score was near either criterion cutoff score (i.e., a score between .75 and 1.25 or a score between -.75 and -1.25) were reassessed to ensure that the cliques were appropriately classified. If the majority (i.e., more than 50%) of the clique members' individual perceived popularity scores were greater than or equal to 1.00, the clique was classified as perceived popular. If the majority of the clique members' individual perceived popularity scores were less than or equal to -1.00, the clique was classified as perceived unpopular. Finally, if the majority of the clique members' individual perceived popularity scores were greater than -1.00 and less than 1.00, the clique were classified as perceived average. As a result, there were 23 perceived popular cliques, 34 perceived average cliques, and 17 perceived unpopular cliques.
Cliques were identified using a modification of the procedures described by Paxton, Schutz, Wertheim, and Muir (1999). Using a list of participating grademates, participants were instructed to identify the "students who are in the group of friends you hang around with most often." Because cliques generally range in size from 3 to 10 members (Ennett, Bauman, & Koch, 1994; Paxton et al., 1999; Rubin et al., 1998), participants were instructed to list not more than 9 friends. If their clique consisted of more than 10 friends (including themselves), participants were to choose 9 of their friends in the clique with whom they spend the most time. Friendship clique nomination data were analyzed using the UCINET 6.0 statistical package (Borgatti, Everett, & Freeman, 2002). By identifying clusters around mutual friendship ties, the program generated a list of all possible cliques. Since the program allowed individuals to be members of more than one clique, it was necessary to consider all possible cliques that any one individual may have membership in and assign individuals to the clique in which they had the highest number of friendship links. To assess the strength of the friendship ties among members in each clique, friendship cohesion was calculated as the observed number of within-clique links divided by the possible number of within-clique links (i.e., observed links/ni × [ni - 1], where ni = the number of members in clique i) (Ennett et al., 1994; Paxton et al., 1999). A unilateral friendship nomination (i.e., A nominates B as a friend, but B does [End Page 415] not nominate A) was counted as one within-clique link, and a reciprocated friendship nomination (i.e., A nominates B as a friend, and B nominates A as a friend) was counted as two within-clique links. A maximum of two links was possible between any two clique members. In previous studies, some researchers have not set a minimum criterion cutoff score for friendship cohesion (e.g., Ennett et al., 1994). Such studies are limited because it is unknown whether cliques consist of valid friendship relations among members. Obtaining cliques with strong friendship ties among members was of greater importance than the exhaustive assignment of participants to cliques. Paxton et al. (1999) proposed that a cutoff score of .40 should be stringent enough to allow for the inclusion of cliques with valid and meaningful friendship ties between friends. Of the 387 participants, 11 participants (6 boys and 5 girls) were not members of a clique (i.e., they were either members of a dyad or were not identified as a friend). Following the procedures outlined above, 74 cliques were identified. Cliques ranged in size from 3 to 9 members (M = 5.56, SD = 1.62). Thirty-four cliques were composed of all girl members, 29 were all boy members, and 11 were mixed-gender cliques. Friendship-cohesion scores calculated for each clique ranged between .40 and 1.00 (M = .72, SD = .18). Since no clique fell below the minimum required friendship-cohesion score of .40, all cliques were included in subsequent analyses.
Friend regard, within-clique social behavior, and social dominance
These measures were a compilation and modification of several measures that have previously been developed by researchers (Asher, Singleton, Tinsley, & Hymel, 1979; Lease et al., 2002; Little, Jones, Henrich, & Hawley, 2003; Prinstein, Boergers, & Vernberg, 2001). These original measures were developed to assess the larger peer group context; therefore, it was necessary to modify the wording of items to apply exclusively to the friendship clique context. Participants completed a questionnaire regarding members of their clique by rating to what extent each item described each friend using a 4-point scale, from 1 (not at all) to 4 (a whole lot). Constructs assessed included friend likability ("I like this friend," "I like to spend time with this friend"; α = .75; adapted from Asher et al., 1979), prosocial behavior ("This friend helps someone in our group when they are having problems," "This friend is nice to someone in our group when they need help"; α = .70; adapted from Prinstein et al., 2001), instrumental relational aggression ("To get what he/she wants, this friend tells others in our group to stop liking someone in our group," "To get what he/she wants, this friend gossips or spreads rumors about someone in our group"; α = .75; adapted from Little, Jones et al., 2003), instrumental overt aggression ("This friend hits, kicks, or punches someone in our group to get what he/she wants," "This friend says mean things to someone in our group to get what he/she [End Page 416] wants"; α = .62; adapted from Little, Jones et al., 2003), reactive relational aggression ("If someone in our group upsets or hurts this friend, he/she tells others in our group to stop liking the person that hurt him/her," "If this friend is mad at someone in our group, he/she gossips or spreads rumors about them"; α = .75; adapted from Little, Jones et al., 2003), reactive overt aggression ("This friend hits, kicks, or punches someone in our group if they have angered him/her," "This friend says mean things to someone in our group if he/she is hurt by them"; α = .61; adapted from Little, Jones et al., 2003), and social dominance ("This friend has influence and power over our group of friends," "This friend gets people in our group to do what he/she wants"; α = .68; adapted from Lease et al., 2002).
Classification of individual dominance rank within cliques
Using the mean score as rated by clique members from the social dominance items, individuals within cliques were classified as high dominant, moderate dominant, or low dominant. Savin-Williams and Freedman (1977) suggested that the most striking individual differences occur between the most dominant individual and the least dominant individual of a group. Therefore, only one clique member was classified as high dominant (i.e., the individual with the highest mean social dominance score as rated by his/her friends), only one clique member was classified as low dominant (i.e., the individual with the lowest mean social dominance score as rated by his/her friends), and all other clique members were classified as moderate dominant. Following this procedure, 74 individuals were classified as high dominant, 228 individuals were classified as moderate dominant, and 74 individuals were classified as low dominant.
Correlational analyses were conducted between all variables of interest in the present study to determine whether any significant associations existed between variables (Table 1). Noteworthy positive correlations were found between clique perceived popularity and peer likability, clique perceived popularity and peer social impact, individual social dominance and peer social impact, and individual social dominance and all types of aggression. Correlational analyses were also conducted between all variables for boys and girls separately (Table 2). In general, a similar pattern of correlations was found for boys and girls. A particularly noteworthy finding was that the correlations between social status variables (i.e., clique perceived popularity and social dominance) and behavior within cliques were similar for boys and girls. Fisher's z-tests (Cohen, 1988) were used to test for significant [End Page 417] differences in the correlations for boys and girls. With the exception of the correlation between instrumental relational aggression and instrumental overt aggression, all of the correlations between the different types of aggression were significantly stronger for girls than for boys. In addition, the correlation between reactive overt aggression and peer likability was significantly different between boys (r = .09, p > .05) and girls (r = -.20, p < .01).
Initially, the gender composition of each clique (i.e., all boy, all girl, or mixed gender) was included as an independent variable in all multivariate analyses. There were no significant main effects or interaction effects found for clique gender composition. Therefore, the data were reanalyzed with the clique gender composition variable removed. These results are presented below.
Social Behavior as a Function of Clique Perceived Popularity and Individual Dominance Rank
It was of interest to determine if an individual's use of social behavior within her or his clique varied as a function of that individual's dominance rank within the clique and/or the clique's level of perceived popularity. To examine potential differences in social behavior, a 2 (gender) × 3 (clique perceived popularity) × 3 (dominance rank) MANOVA was conducted whereby gender (i.e., boy, girl), perceived popularity of clique (i.e., perceived popular, perceived average, and perceived unpopular), and dominance rank (i.e., high dominant, moderate dominant, low dominant) were used as the independent variables, and instrumental overt aggression, instrumental relational aggression, reactive overt aggression, reactive relational aggression, and prosocial behavior were used as the dependent variables.
Results revealed a significant main effect of gender (F[5, 354] = 20.36, p < .001). Examination of subsequent univariate analyses indicated a significant effect of gender for prosocial behavior (F[1, 376] = 58.02, p < .001), instrumental relational aggression (F[1, 376] = 6.68, p < .01), and reactive relational aggression (F[1, 376] = 12.94, p < .001). Girls engaged in more prosocial behavior (M = .41, SD = .83) than boys (M = -.47, SD = .98), more instrumental relational aggression (M = .15, SD = 1.12) than boys (M = -.17, SD = .81), and more reactive relational aggression (M = .21, SD = 1.14) than boys (M = -.24, SD = .75). Girls and boys did not differ on their level of instrumental overt aggression (F[1, 376] = .86, p > .05 [Ms = .07 and -.07]) or reactive overt aggression (F[1, 376] = 1.20, p > .05 [Ms = -.03 and .04]).
Results also indicated a significant main effect of clique perceived popularity (F[10, 708] = 3.46 p < .001). Examination of subsequent univariate analyses indicated a significant effect of clique perceived popularity for [End Page 418]
[End Page 419]
[End Page 420]
prosocial behavior, instrumental relational aggression, instrumental overt aggression, and reactive overt aggression (Table 3). Bonferroni post hoc tests revealed that adolescents in perceived average cliques engaged in more prosocial behavior than adolescents in perceived unpopular cliques. Adolescents in perceived popular cliques engaged in more instrumental relational aggression, instrumental overt aggression, and reactive overt aggression than adolescents in perceived average cliques. In addition, adolescents in perceived unpopular cliques engaged in more instrumental overt aggression than adolescents in perceived average cliques.
Results also indicated a significant main effect of dominance rank (F[10, 708] = 3.91, p < .001). Examination of subsequent univariate analyses revealed a significant effect of dominance rank for all types of aggression (Table 4). Bonferroni post hoc tests indicated that high dominants engaged in more instrumental relational aggression and instrumental overt aggression than all other clique members and also engaged in more reactive relational aggression and reactive overt aggression than low dominants. Moderate dominants engaged in more aggression, regardless of form or function, than low dominants. No significant interactions were found.
Peer/Friend Regard as a Function of Clique Perceived Popularity and Individual Dominance Rank
It was of interest to determine if an individual's ratings of peer and friend regard were associated with her or his dominance rank and/or the clique's level of perceived popularity. To examine potential differences in peer/friend regard, a 2 (gender) × 3 (clique perceived popularity) × 3 (dominance rank) MANOVA was conducted whereby gender (i.e., boy, girl), perceived popularity of clique (i.e., perceived popular, perceived average, and perceived unpopular) and dominance rank (i.e., high dominant, moderate dominant, low dominant) were used as the independent variables, and scores on peer likability, peer social impact, and friend likability were used as the dependent variables.
A significant main effect of gender was not found (F[3, 355] = .57, p > .05). Boys and girls were rated as equally liked by friends (F[1, 376] = .03, p > .05 [Ms = -.06 and .05]) and by peers (F[1, 376] = 1.63, p > .05 [Ms = -.08 and .14]) and were rated similarly on social impact (F[1, 376] = .02, p > .05 [Ms = -.01 and .01]).
Results indicated a significant main effect of clique perceived popularity (F[6, 710] = 27.71, p < .001). Examination of subsequent univariate analyses indicated a significant effect of clique perceived popularity for [End Page 421]
[End Page 422]
[End Page 423]
peer likability, peer social impact, and friend likability (see Table 3). Bonferroni post hoc tests revealed that adolescents in perceived popular cliques were more liked by peers than adolescents in perceived average cliques and perceived unpopular cliques. In addition, adolescents in perceived average cliques were more liked by peers than adolescents in perceived unpopular cliques. In terms of peer social impact, adolescents in perceived popular cliques scored higher on peer social impact than adolescents in perceived average cliques and perceived unpopular cliques. In addition, adolescents in perceived unpopular cliques scored higher on peer social impact than adolescents in perceived average cliques. In terms of friend likability, adolescents in perceived popular cliques and perceived average cliques were more liked by friends than adolescents in perceived unpopular cliques.
Results also revealed a significant main effect of dominance rank (F[6, 710] = 3.20, p < .01). Examination of subsequent univariate analyses indicated a significant effect of dominance rank for peer social impact (see Table 4). Bonferroni post hoc tests revealed that high dominants scored higher on peer social impact than moderate dominants and low dominants. Also, moderate dominants scored higher on peer social impact than low dominants. No significant interactions were found.
Identification of Dominance Strategies
High dominants (n = 74) were assigned to one of three types of dominance strategies (i.e., prosocial dominant, aggressive dominant, bistrategic dominant). For purposes of this analysis, each high dominant's mean level of aggression (i.e., mean score across the four types of aggression) was calculated to determine the individual's overall tendency toward engaging in aggressive behavior regardless of the form and function of aggression.1 [End Page 424] High dominants were classified as prosocial dominant if their score for prosocial behavior was higher than the mean of their clique and their mean level of aggression was lower than the mean of their clique. Similarly, high dominants were classified as aggressive dominant if their mean level of aggression was higher than the mean of their clique and their score for prosocial behavior was lower than the mean of their clique. High dominants were classified as bistrategic dominant if their scores for both prosocial behavior and mean level of aggression were equal to or higher than the means of their clique. Following this procedure, 25 high dominants were classified as prosocial dominant (10 boys, 15 girls), 29 were classified as aggressive dominant (14 boys, 15 girls), and 20 were classified as bistrategic dominant (8 boys, 12 girls). Chi-square analyses indicated no significant gender differences in dominance strategy (χ2 [2, n = 74] = .49, p > .05).
Comparison between Dominance Strategies: Individual Social Dominance, Clique Perceived Popularity, and Peer/Friend Regard
It was of interest to determine whether dominance strategy was associated with clique perceived popularity and peer/friend regard. It was also of interest to determine whether high dominants employing one of the three types of strategies were perceived as equally socially dominant by clique members. To examine potential differences in social dominance, clique perceived popularity, and peer/friend regard, I conducted a MANOVA whereby gender (i.e., boy, girl) and dominance strategy (i.e., prosocial dominant, aggressive dominant, bistrategic dominant) served as the independent variables, and scores on social dominance, clique perceived popularity, peer likability, peer social impact, and friend likability were used as the dependent variables.
A main effect of gender was not found (F[5, 64] = 1.03, p > .05). High dominant boys and girls were rated as equally socially dominant (F[1, 73] = 1.52, p > .05 [Ms = 1.28 and 1.18]) and equally liked by friends (F[1, 73] = .35, p > .05 [Ms = -.13 and -.16]) and by peers (F[1, 73] = .01, p > .05 [Ms = .02 and .17]). In addition, high dominant boys and girls did not differ on [End Page 425]
[End Page 426]
social impact (F[1, 376] = 2.80, p > .05 [Ms = .49 and .11]), and their clique's level of perceived popularity did not differ (F[1, 376] = .25, p > .05 [Ms = -.01 and -.02]).
Multivariate results revealed a significant main effect of dominance strategy (F[10, 128] = 3.85, p < .001). Univariate results for social dominance were nonsignificant (F[2, 73] = 2.28, p > .05), indicating that high dominants were considered to be equally dominant regardless of the strategy they tended to employ within their cliques. Results also revealed a significant univariate effect of dominance strategy for clique perceived popularity, peer likability, and friend likability (Table 5). Bonferroni post hoc analyses indicated that bistrategic dominants were more likely to be members of cliques with higher perceived popularity than were prosocial dominants. In addition, relative to aggressive dominants, prosocial dominants and bistrategic dominants were more liked by friends, and bistrategic dominants were more liked by peers. A significant interaction between gender and dominance strategy was not found.
Inspired by qualitative work on clique dynamics in late childhood and early adolescence (Adler & Adler, 1998; Eder, 1985; Merten, 1997), the present study quantitatively delineated two types of status hierarchies within early adolescents' social worlds and examined variations in within-clique behavior and experiences based on relative status differences. The results from this study suggest that the perceived popularity of cliques and the social dominance of individuals within each clique are associated with how one is regarded by friends and peers and also are associated with one's behavior within the clique.
As expected, relatively few gender differences emerged. Nevertheless, results revealed significant negative associations between most types of aggression within cliques and peer likability for girls but not for boys. In particular, the correlation between reactive overt aggression and peer likability differed significantly for boys and girls: reactive overt aggression was negatively associated with peer likability for girls, whereas the correlation was nonsignificant for boys. These findings allude to girls' heightened social awareness (LaFontana & Cillessen, 1999). Relative to boys, girls may be more aware of peers who are aggressive toward friends in their clique and may dislike them for engaging in such behavior. Correlational [End Page 427] analyses also revealed significant gender differences in the associations between the various types of aggression whereby these positive correlations were stronger for girls than for boys. These findings suggest that compared to boys, girls may be less selective in the type of aggression they use toward the friends in their clique. Consistent with previous research suggesting girls are highly prosocial toward peers (Hawley, 2003; Lease et al., 2002), the present results showed that girls were also more prosocial toward the friends in their clique than were boys. In addition, girls were more relationally aggressive toward friends in their clique than were boys, yet girls did not differ from boys in terms of overt aggression, consistent with some previous peer level research (e.g., Prinstein & Cillessen, 2003). Beyond these differences, gender did not affect the associations between social status and social behavior within cliques or peer/friend regard. Thus, it appears that the processes involved in clique dynamics as they relate to social status within and between cliques may be similar for boys and girls.
Social Behavior and Peer/Friend Regard as a Function of Clique Perceived Popularity
The results demonstrate that membership in perceived popular cliques may be associated with drawbacks as well as benefits. The finding that perceived popular clique members were aggressive toward the friends in their clique contributes to the growing body of research showing that there may be negative components to being perceived as popular during early adolescence (e.g., Merten, 2004; Sandstrom & Cillessen, 2006). Compared to members of lower status cliques, those in perceived popular cliques used more instrumental relational aggression. Some researchers have suggested that peer-directed relational aggression may be used intentionally to damage another peer's social reputation, thereby maintaining an individual's own high status (e.g., Rose, Swenson, & Waller, 2004). Thus, instrumental relational aggression used toward other popular clique members may be an effective method for maintaining or increasing one's status within the clique by decreasing the social position of others.
Despite the existence of aggression within their cliques, members of perceived popular cliques were reasonably well liked by their friends and were prosocial toward their friends. Receiving prosocial acts may provide sufficient incentive for popular youths to maintain amiable friendships. However, Adler and Adler (1998) have suggested that niceness within popular cliques may sometimes be disingenuous and that liking and disliking among members may be influenced by the preferences of clique leaders. Nevertheless, the benefits associated with membership in these cliques may outweigh negative [End Page 428] aspects, as the desire for popularity may be stronger than the desire for a stable, high-quality friendship in early adolescence (Merten, 2004).
Youths in perceived popular cliques scored the highest on peer social impact and peer likability. Thus, the behaviors of those in high-status positions in the social milieu may be widely noticed, allowing for their peers to form strong opinions of them. Although many peers may personally like those in perceived popular cliques, these positive feelings are potentially an indication of admiration, since those with high social status are looked up to and most often emulated by others (Lease et al., 2002). Despite their use of aggression, perceived popular early adolescents are still held in high esteem by many of their peers.
In contrast to perceived popular cliques, membership in perceived unpopular cliques may be associated with the fewest positive aspects. Individuals in these cliques were aggressive toward their friends and, compared to those in perceived popular and perceived average cliques, were less liked not only by their peers but also by their friends. Interestingly, individuals in perceived unpopular cliques used more instrumental overt aggression toward the friends in their clique than did members of perceived average cliques. Adler and Adler (1998) have found ethnographic evidence suggesting that low-status youths are often discontented with their social position in the peer group. If this is the case, the present findings may reflect their desire to increase their status; however, instrumental overt aggression may not prove as effective a method as using relational or reputational forms of instrumental aggression (Prinstein & Cillessen, 2003). Although popular youths may use aggression to their advantage, unpopular youths may not possess the savoir faire required to produce their desired effect (Adler & Adler, 1998), as evidenced by their lower levels of prosocial behavior toward clique members. This inability to balance aggression with prosocial behavior may prevent unpopular adolescents from ascending the hierarchy and contributes to the negative perceptions held by their friends and peers. Although perceived unpopular cliques consisted of reciprocated friendships, Adler and Adler (1998) have suggested that they may be forced relationships based on availability since these children tend to be excluded by the larger peer group. Therefore, individuals in perceived unpopular cliques may not have strong positive feelings for one another because their affiliation may be motivated by a lack of an alternative source of friendship.
Many researchers have focused on studying those who are at extreme ends of social status continua (e.g., accepted versus rejected youths) or have used an average group primarily as a means of comparison. This study has demonstrated the importance of devoting attention to average status groups. Membership in perceived average cliques may be beneficial [End Page 429] for early adolescents, as it may foster generally positive friendship experiences. Indeed, the results showed that youths in perceived average cliques were reasonably liked by their peers, were well liked by their friends, and engaged in high levels of prosocial acts as well as low levels of aggressive behavior within their friendship cliques. This study provided support for Adler and Adler's (1998) qualitative finding that children in middle friendship circles may have the highest-quality friendships and engage in less aggression within their cliques because they are satisfied with their midlevel status and are less concerned with ascending the hierarchy.
Social Behavior and Peer/Friend Regard as a Function of Dominance Rank within the Clique
Unlike Adler and Adler's (1998) research with preadolescents, the current study demonstrated that status hierarchies within cliques are not exclusive to popular cliques in early adolescence, as social dominance was measurable within cliques at all levels of perceived popularity. Whether this difference is due to developmental differences between preadolescent and early adolescent samples is an important question for future research. Furthermore, the lack of interaction effects between clique perceived popularity and individual social dominance suggests that processes related to social dominance within cliques may be largely independent of the status of the clique. The findings demonstrate that the higher one's dominance rank within the clique, the more aggressive toward clique members one is likely to be regardless of the perceived popularity of the individual's clique.
Following the suggestion of Savin-Williams (1979; Savin-Williams & Freedman, 1977), I hypothesized that low dominants would be least liked by friends. Contrary to what was expected, low dominants were reasonably well liked by friends, possibly because they were the least aggressive members of each clique yet used levels of prosocial behavior comparable to that of other clique members. Compared to the rest of their clique, low dominants scored the lowest on peer social impact. Because these individuals are likely to be nonaggressive within their clique as well as within the larger peer group and may be more withdrawn (Lease et al., 2002; Savin-Williams, 1979; Savin-Williams & Freedman, 1977) yet get along with others, they may go somewhat unnoticed by many peers.
In contrast, high dominants were found to use more instrumental aggression (both relational and overt forms) toward the friends in their clique than any subordinate members used. Previous research has shown that instrumental aggression is often used by socially dominant youths to protect their position within the peer group (Hawley, 2003; Hawley et al., [End Page 430] 2002). The present results demonstrate that instrumental aggression may be a useful tool in maintaining a dominant position within friendship cliques as well. Additionally, the current findings indicate that high dominants scored higher than subordinate clique members on peer social impact regardless of the perceived popularity of their clique. High dominants may interact with their friends in a manner that is readily apparent to the entire peer group. Alternatively, these youths may utilize dominance strategies within the larger peer group as well as with their friends (Hawley et al., 2007), making them controversial social figures.
Interestingly, prosocial dominants, aggressive dominants, and bistrategic dominants were perceived by their friends to be equally socially dominant within cliques across all levels of perceived popularity. In other words, the type of strategy used by high dominants did not influence their cliques' perception of the degree of their social dominance. Hawley (2003) found that prosocial or bistrategic socially dominant early adolescents may find the most success at resource control (i.e., getting what they want) within the larger peer network. However, the findings in the current study indicate that all three strategies can be effective in maintaining the highest dominance rank within a clique.
The most intriguing findings regarding dominance strategies related to high dominants who used both prosocial and aggressive behaviors within their cliques. Such bistrategic dominants were more likely to be members of cliques higher in perceived popularity than were prosocial dominants, and both types of dominant adolescents were more liked by friends than were aggressive dominants. Interestingly, bistrategic dominants were the most liked by peers. The finding that high dominants who were bistrategic within their cliques were well liked and were members of higher status cliques provides additional support for Hawley and colleagues' (Hawley, 2003; Hawley et al., 2007) assertion that bistrategic adolescents may be the most socially successful individuals in a peer group. Using aggression within the clique may effectively allow social dominants to maintain power and control by inducing a degree of fear or intimidation in others (Adler & Adler, 1998; Merten, 1997), while prosocial behavior may allow them to offset negative repercussions. Although there may be costs to membership in cliques with bistrategic dominant leaders, their unpredictability may be alluring and exciting to some adolescents (Hawley et al., 2007).
Limitations and Future Directions
Although every effort was made to achieve a high participation rate, a 100% participation rate was not obtained. Thus, there may have been a few [End Page 431] members of each clique who were not accounted for. Consequently, it is possible that the dominance hierarchy within each clique may have been slightly different had every student in each school participated in the study. However, a dominance hierarchy was found to exist regardless of who held each position within the clique at the time of data collection.
The present study assessed the extent to which within-clique aggressive and prosocial behaviors were characteristic of early adolescents. What remains unknown is the frequency of such behaviors used among friends in cliques. For example, individuals may be regarded as aggressive because aggression is often a noticeable behavior or because they frequently act aggressively, or perhaps it is a combination of the two. It would be expected that the frequency of prosocial acts would be high whereas the frequency of aggressive acts would be fairly low, since it is unlikely that most youths would tolerate constant hostility within their cliques. Nevertheless, even occasional acts of aggression from clique members are liable to harm victims, especially since the aggression originates from an individual with whom the victim shares a social relationship. Therefore, studies of social-psychological adjustment as it relates to the frequency and perceived impact of within-clique aggression are warranted.
Also of interest for future research is a quantitative comparison of friendship quality across cliques of differing levels of perceived popularity and between clique members with differing levels of social dominance. Qualitative research has suggested that friendships within popular cliques may be of poorer quality (i.e., high conflict, victimization, unstable, restrictions on self-expression and choice of friends) than the friendships of lower status (i.e., average) youths (Adler & Adler, 1998; Merten, 1997, 2004). Furthermore, Hawley and colleagues (2007) found that socially dominant adolescents and their friends report having higher quality friendships than subordinate adolescents. Thus, it would be interesting to assess in future studies whether the friendship quality of dominant and subordinate early adolescents differ as a function of their clique's level of perceived popularity.
The findings from the current study have several implications. Methodologically, this research has shown that two distinct hierarchies can be distinguished among early adolescent peer groups: (1) a dominance hierarchy within cliques based on measures of individual social dominance and (2) a status hierarchy within the larger peer network based on measures of clique perceived popularity. Delineating these structures can provide a useful template from which researchers can understand how early adolescents organize themselves socially and function within such contexts. Additionally, this research has shown that within-clique aggression may be socially rewarded by affording high status to certain aggressive individuals, especially [End Page 432] when they have sufficient social prowess to offset their aggression with prosocial behaviors. Overall, the results from the present study offer a more comprehensive understanding of the complex associations between social status and the dynamics within early adolescent friendship cliques.
Leanna M. Closson, Department of Educational and Counselling Psychology and Special Education
1. The creation of dominance strategy subtypes distinguishing the form of aggression (i.e., relational vs. overt) used by high dominants relative to their clique was considered. For example, high dominants were classified as relationally aggressive dominant if their mean level of relational aggression was higher than the mean of their clique and if their scores for prosocial behavior and overt aggression were lower than the mean of their clique. Chi-square analyses revealed no gender differences in dominance strategy subtype when subdividing based on the form of aggression (χ2 [6, n = 74] =.57, p >.05). In addition, the creation of dominance strategy subtypes distinguishing the function of aggression (i.e., instrumental vs. reactive) used by high dominants relative to their clique was considered. Chi-square analyses revealed no gender differences in dominance strategy subtype when subdividing based on the function of aggression (χ2 [6, n = 74] = 6.75, p >.05). Due to unacceptably low sample sizes in each cell, a decision was made to aggregate across type of aggression for the dominance strategy analyses. Nevertheless, a pattern of findings from exploratory analyses (e.g., examination of means) alluded that dominant youths who use overt aggression (alone or in combination with relational aggression) toward the friends in their clique may be less liked by friends and peers compared to other subtypes of dominant youths, including those who are primarily relationally aggressive or bistrategic. However, bistrategic dominant youths may be more liked by peers and have higher levels of clique perceived popularity than other dominant youths regardless of the form of aggression they use. These findings represent trends in the data from exploratory analyses with very small sample sizes and should be interpreted with caution. Nonetheless, it may be reasonable for future studies with larger sample sizes to distinguish the forms of aggression when examining dominance strategies. Conversely, exploratory analyses did not demonstrate a pattern of findings to indicate a need to subdivide dominance strategies based on the function of aggression.