Вы находитесь на странице: 1из 13

International Journal of Behavioral Development 2008, 32 (6), 473485 http://www.sagepublications.

com

2008 The International Society for the Study of Behavioural Development DOI: 10.1177/0165025408093667

Who bullies whom? Social status asymmetries by victim gender


Philip C. Rodkin
University of Illinois at Urbana-Champaign, USA

Christian Berger
Universidad Alberto Hurtado, Chile

This study asks whether bullies have higher social status than their victims. Social status was measured by social preference, popularity, and physical competence as perceived by children and teachers. A survey instrument was introduced to enable identication of specic victims associated with specic bullies. The sample was 508 fourth and fth grade children from midwest U.S. elementary schools. Results indicated that peer- and teacher-perceived popularity were the optimal status measures for capturing heterogeneity in bullyvictim status imbalances. In addition, the gender of victims of male bullying was critical. Powerful, popularaggressive bullies and unpopular victims were found in samesex dyads, but unpopularaggressive boys were also identied as bullying popular girls. All bullies were disliked. Implications are drawn for peer sexual harassment and for innovations in sociometric technology. Keywords: bullying; gender; social status

Bullies have power over their victims by denition. In his pioneering work on bullying at school, Olweus (1993, p. 10) stated that the term bullying is not (or should not be) used when two students of approximately the same strength (physical or psychological) are ghting or quarreling. Of note here is the equal footing of psychological and physical: power differences between bullies and victims can take the form of psychological attributes like social status, social inuence, and interpersonal perception that are hard to quantify relative to physical attributes like size and strength (cf., Lippitt, Polansky, Redl, & Rosen, 1952; Sherif, 1956). Psychological power differentials continue to be a critical feature in how asymmetries between bullies and victims are conceptualized. Leff, Power, and Goldstein (2004, p. 270), in a discussion of bullying methodology, reiterate the Olweus (1993) denition and hypothesize that bullies should have higher social status than the victim. Evidence regarding the relative social status of bullies and victims is inconclusive (Boulton, 1999) because, in actuality, the social status of bullies and their victims has not been directly compared. Groups of children classied as bullies have been compared with groups of children classied as victims, but without dyadic understanding of who bullies whom social status asymmetries between bullies and their victims cannot be directly ascertained. In this study, we introduce a survey instrument that asks children to identify bullies together with the children whom each bully most often harasses. The new instrument is constructivist: children dene for themselves what is bullying in the context of peers who to them exemplify bullying behavior. The instrument does not require children to adhere to common denitional components of bullying such as the repetition of bullying, the intent of the bully, or the status

relationship between bullies and victims. Indeed, the development of the who bullies whom measure derives from the need to subject the status feature of bullying denitions to empirical scrutiny (Espelage & Swearer, 2003). Using the who bullies whom measure, we examine two potential sources of heterogeneity in social status asymmetries between bullies and victims. The rst source of heterogeneity concerns whether social status is operationalized as social preference or perceived popularity. There is good reason to suspect that perceived popularity will be more sensitive than social preference to status asymmetries between bullies and victims. The second source of heterogeneity concerns gender. Here we expect that boys who engage in cross-sex bullying may have physical but not psychological power (social status) over the girls whom they harass. We elaborate on each of these two possible sources of heterogeneity in turn.

Social status, social preference, perceived popularity


Olweus (2001, p. 7) implicated the dominant North American approach to childrens peer relations in explaining why the social status of bullies was underappreciated. The North American approach, according to Olweus (2001), tends to operationalize social status as social preference (i.e., being highly liked and hardly disliked) and to overemphasize connections between aggression and rejection. In essence, it made little sense to study the psychological power of bullies when aggressive children were viewed as generally rejected and marginalized. Recent studies of aggressive children, who may or may not be bullies, make clear that aggressors can attain high social status as long as social status is not dened exclusively as being

Correspondence should be sent to Philip C. Rodkin, 232A Col. Wolfe School, Mail Code 422, 403 E. Healey St., University of Illinois at Urbana-Champaign, Champaign, Illinois 61820; e-mail: rodkin@ illinois.edu or Christian Berger, Facultad de Psicologa, Universidad Alberto Hurtado, Almirante Barroso 26, Santiago 6500620, Chile; email: cberger@uahurtado.cl

The research reported in this article was made possible by Small Grant #20050079 from the Spencer Foundation. Christian Berger is a Fulbright / Fundacin Andes grantee. The views expressed are solely the responsibility of the authors. We thank the school and the children involved in this research and Gaebson (Sunny) Lee for her assistance in survey administration.

474

RODKIN AND BERGER / WHO BULLIES WHOM?

widely liked and hardly disliked (Farmer, Estell, Bishop, ONeal, & Cairns, 2003; LaFontana & Cillessen, 2002; Parkhurst & Hopmeyer, 1998; Rodkin, Farmer, Pearl, & Van Acker, 2000; Zeller, Vannatta, Schafer, & Noll, 2003). Farmer et al. (2003), for example, identied subtypes of aggressive children among 419 rural African American seventh and eighth graders and obtained measures of popularity, preference, and social network prominence from students and teachers. Aggressive boys who were popular and prominent were nonetheless overrepresented in the rejected sociometric category. Longitudinal evidence suggests that popularity and aggression may be positively related from middle childhood to adolescence (Cillessen & Mayeux, 2004; Rose, Swanson, & Waller, 2004). In their study of 905 children tracked from grades four through eight, Cillessen and Mayeux (2004) reported that aggression was negatively related to preference but positively related to peer perceptions of popularity. Positive correlations between popularity and aggression increased over grades four to eight in what was referred to as a shift from censure to reinforcement of aggressive behavior. Among bullies per se, the most direct examination of psychological power is by Vaillancourt, Hymel, and McDougall (2003). Interestingly, Vaillancourt et al.s purpose was not to test whether bullies were powerful, but to question whether bullies were uniformly unpopular. Victims social status was not ascertained. Subtypes of 555 Euro-Canadian bullies were identied in grades six through 10 and measured on status characteristics including peer nominations of who is powerful (Who can pressure others into doing things?), who is popular, who is liked and disliked, and power-related attributes like leadership, wealth, attractiveness, and athleticism. Bullies, also identied using peer nominations, varied in their social power. Half of all bullies were labeled high power and were popular and likeable, better looking, more likely to be named as leaders, and showed trends towards greater athleticism than bullies with less power. High power bullies were more physically and relationally aggressive than bullies with less power. All bullies were highly disliked. The social status of bullies and victims have been compared in a number of survey studies using a less extensive array of status measures than in Vaillancourt et al. (2003). Boulton (1999) reviews this work and nds wide variation in reported results. With regards to social preference, both bullies and victims are highly disliked. Victims are unlikely to be nominated as liked most, but bullies receive liked most nominations at proportional rates (Coie, Dodge, & Coppotelli, 1982; Graham, Bellmore, & Juvonen, 2003; Pellegrini, Bartini, & Brooks, 1999; Perry, Kusel, & Perry, 1988). Boulton (1999) did not nd consistent trends in comparisons of perceived popularity of bullies and victims, and he suggested that observational methods that examine bullies and their victims within the same situation may better reveal the powerful and central role that bullies can enjoy. For example, OConnell, Pepler, and Craig (1999) observed playground behavior to identify aggression among rst through sixth graders that involved bullies, their victims, and at least two nearby peers. They reported that bullying lasted longer when more peers were present, and that bullies were positively reinforced for their behavior 75% of the time, with peers intervening on the behalf of victims in only 25% of cases. In Boultons (1999) observational study of 89 British eight- and nine-year-olds, bullies and victims were low on social preference but male bullies afliated in larger peer groups than victims.

In sum, whether bullies enjoy social status over the children they harass may depend on how social status is operationalized and whether bullies and victims are examined conjointly within the same context. Our hypothesis in this study, using multiple measures of social status and a survey instrument that asks children to name bullies and victims together, is that bullies will be more popular but not more preferred than the children they harass.

Bullying, victimization, and gender


Our starting point in the consideration of bullying, victimization, and gender is the proposition that, albeit with some variation by bullying form (e.g., relational, physical), most bullies are boys but victims are both boys and girls (Espelage, Mebane, & Swearer, 2004; Scheithauer, Hayer, Petermann, & Jugert, 2006; Schwartz, Proctor, & Chien, 2001; Solberg & Olweus, 2003). Olweus (1993, p. 18) outlined the algebraic consequence of the preponderance of male bullies and the more similar prevalence rates of male and female victims by claiming that boys carried out a large part of the bullying to which girls were subjected (italics in original): 60% of fth through seventh grade girls whom Olweus (1993) reported as being harassed said that they were bullied by boys. Others nd lower prevalence rates of boygirl bullying (e.g., Craig, Pepler, Connolly, & Henderson, 2001; Pellegrini & Long, 2003; Russell & Owens, 1999). There may be concurrent or predictive linkages between bullying and peer sexual harassment (McMaster, Connolly, Pepler, & Craig, 2002; Pellegrini, 2002; Rodkin & Fischer, 2003). Although peer sexual harassment is often seen as a purely adolescent phenomenon, its origins may lie in middle childhood (Adler & Adler, 1998; Maccoby, 1998). The American Association of University Women (American Association of University Women Educational Foundation [AAUW], 2001, p. 25) reported that 38% of girls who experience sexual harassment say they rst experienced it in elementary school. Stein (1995) argues that peer sexual harassment in middle childhood is a form of bullying, opening her essay with an unforgettable account of a fth grade girl who is repeatedly attacked by a boy in ways that clearly resemble dating and domestic violence yet the girls requests for help are belittled by the teacher. Steins (1995) Bullyproofing protocol for fourth and fth graders and their teachers encourages participants to realize the linkages between elementary school bullying and sexual harassment. Malemale aggression is more prevalent than malefemale aggression (Pellegrini, 2007; Russell & Owens, 1999), but there is ample reason to expect that children will name a substantial proportion of boys bullying girls when given the opportunity. One advantage of the who bullies whom measure is that children can spontaneously nominate mixed-sex dyads. In the notoriously tense gender climate of middle childhood (Adler & Adler, 1998; Maccoby, 1998), boys and girls rarely choose one another as best friends or afliate in the same groups but negative relationships are easily apparent. Underwood, Schockner, and Hurley (2001) placed 8-, 10-, and 12year-olds in an experiment where they were teased by either a same- or opposite-sex confederate while losing at a computer game. Children who were teased by an opposite-sex peer showed more negative facial expressions, made more negative remarks, displayed more negative gestures, and were less likely to want to befriend their provocateur. Rodkin, Pearl, Farmer,

INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2008, 32 (6), 473485

475

and Van Acker (2003) assessed sociometric relationships of antipathy by identifying dyads of third and fourth grade children who nominated one another as liked least. Over three assessment periods, between 40 and 50% of antipathy dyads included one boy and one girl (see also Abecassis, Hartup, Haselager, Scholte, & Van Lieshout, 2002). These studies, which show childrens sensitivity to cross-sex teasing and the high prevalence of boygirl dyads of dislike, lead us to expect that we will nd cross-gender bullying once we ask who bullies whom. We also expect to nd that the denitional pattern of a powerful bully and powerless victim will not characterize cases where boys bully girls, at least with regards to psychological power. Evolutionary theory suggests that status-conferring properties of bullying will hold only in male, same-sex instances as dominance and resource allocations are established within peer groups of boys (Bjorklund & Pellegrini, 2002; Buss, 1999). Sroufe, Bennett, Englund, Urban, and Shulman (1993) studied 47 10- and 11-year-olds at summer camp and found that boys who were unusually aggressive towards girls violated gender norms of separation, were disliked by peers, and were perceived by camp counselors as socially incompetent. Conversely, females who are the victims of male bullying may not resemble low-status male victims but instead could be popular girls singled out for harassment by unpopular, socially incompetent boys, possibly as part of an inappropriate strategy for expressing romantic interest (Adler & Adler, 1998; Duncan, 1999; Maccoby, 1998; McMaster et al., 2002; Pellegrini, 2002). An alternative hypothesis would be that boys who mainly harass girls enjoy social status because they exemplify and extend negative middle childhood gender norms. The Olweus (1993) description of power asymmetries does not consider cross-sex bullyvictim dyads even though Olweus raised the problem of cross-sex bullying. Our analytic strategy follows in three parts. First, because the who bullies whom measure is a new instrument, we conduct descriptive analyses of the prevalence, stability, and behavioral characteristics associated with bullying and victimization. This helps us assess the strengths and limitations of the measure. Then we turn to the overarching study hypothesis that variation will qualify any general proposition on how social status is apportioned among bullies and victims. In the second section we examine how social status is operationalized. We expect that bullies should be seen as more popular than the children they harass, but not as more preferred. In the third section we analyze the gender composition of bullyvictim dyads. Our hypothesis is that social status asymmetries should favor male bullies whose victims are mainly boys, but not male bullies who predominately target girls.

in the larger school: several weeks before survey administration, parents were sent a letter describing the project with a form to sign if they declined their childs participation. Active consent procedures were used in the smaller school: here, parents or guardians written assent was required. Students written assent to participate was required in both elementary schools. The overall participation rate was 83.9% but this varied by school: participation was 91.1% in the larger, ethnically homogeneous school with passive consent and 60.2% in the smaller, ethnically diverse school with active consent. All results were recomputed using only children participating in the larger school but there were no differences in study ndings.

Measures
Survey measures were obtained from children and their teachers. Children were asked to identify bullies and the children they picked on, and to rate themselves and their peers on a variety of social and personality characteristics. Teachers also rated their students social and personality characteristics. Who bullies whom? The who bullies whom measure involves two questions repeated three times. First, children are asked: Are there some kids in your class who really like to bully other kids around? Please write the name of a kid that bullies other kids around. Children then write the rst name and last initial of the child who rst comes to mind as a bully. Second, children are asked: Which kids does this bully like to pick on the most? Six lines are provided for children to write the rst name and last initial of peers who t the description of being picked on by the particular bully they nominated. After naming a bully and children whom the bully most picks on, children can then nominate a second and third bully along with children whom those bullies are most likely to harass. Children were classied as bullies and/or victims if they were nominated by at least two peers (self-nominations included). Bullyvictim dyads were recognized if at least two peers nominated a specic bully as likely to harass the same victim. Peer nominations. Children were asked to nominate three peers in their classroom who best t descriptors for 14 items. Children were told that they could nominate same- or crosssex peers, themselves (all self-nominations were removed prior to analysis), and that peers could be nominated for more than one item. Scores for peer assessments were calculated from the quotient of the number of nominations received by a child for an item over the number of classmates nominating at least one peer for that item. A log transformation (with a small constant added to remove zero values) was applied to the peer scores to satisfy normality, and then peer scores were standardized by sex. A principal components analysis with varimax rotation was performed on the normalized peer nominations and revealed three components: social status, aggression, and prosocial behavior.We further disaggregated the social status component into perceived popularity and physical competence because we were interested in differentiating psychological and physical power. We disaggregated aggression into overt and relational subtypes. Finally, shy behavior was kept as a single-item factor. The resulting six peer nomination composites were as follows. 1 Perceived popularity consisting of three items ( = .83): popular (These kids are very popular. Theyre the most

Method Participants
The sample consisted of 508 fourth and fth grade children (275 boys, 233 girls; ages 1011) recruited from two elementary schools in the midwestern USA. The rst school had 17 classrooms (215 boys, 175 girls) and was over 98% European American; the second school had ve classrooms (60 boys, 58 girls) with an ethnic breakdown of 66.7% European American, 23.9% African American, and <10% other (Hispanic, Asian, and Native American). Passive consent procedures were used

476

RODKIN AND BERGER / WHO BULLIES WHOM?

popular kids in the whole class.), cool (These kids are really cool. I think theyre the coolest kids in the whole class.), want to be like (If I could be somebody else, Id want to be just like these kids.). Physical competence consisting of two items ( = .51): best looking (These kids are the best looking kids in our class.) and athletic (These kids are very athletic. They are best at sports and many outdoor games.). Relational aggression consisting of two items ( = .83): makes fun of others (These kids like to make fun of other kids and embarrass them in front of other people.) and says mean things (These kids say mean things to other kids, and they spread nasty rumors about other kids.). Overt aggression consisting of three items ( = .87): doesnt follow rules (These kids dont follow the rules, dont pay attention, and talk back to the teacher.), starts ghts (These kids start ghts. These kids push other kids around, or hit them.), and upsets everything (These kids have a way of upsetting everything when they get into a group dont share and try to get everyone to do things their way.). Prosocial consisting of three items ( = .87): cooperate (These kids are agreeable and they cooperate pitch in, share, and give everyone a turn.), good grades (These kids get good grades, usually know the right answer, and work hard in class.), nice (These kids are always willing to do something nice for somebody else.). Shy as a single item (These kids act very shy with other kids. Its hard to get to know these kids.).

with which it shared high correlations. This ve-item composite ( = .89) was relabeled prosocial. Six-month testretest stability for teacher reports ranged from .69 to .81 (all ps < .001, n = 390). The ICS-T has convergent validity with observational data and predictive validity over an eight-year period in predictions of adult adjustment, early school drop out, and teenage parenthood (Cairns, Leung, Gest, & Cairns, 1995). ICS-T scores were standardized by gender and classroom. Interpersonal Competence ScaleSelf. The Interpersonal Competence ScaleSelf (ICS-S) is a 21-item bidirectional questionnaire that students completed individually about themselves consisting of the same 18 items and scale anchors as the ICS-T plus three distracters. The afliation factor of the ICS-S was not supplemented with additional prosocial items as was done for the ICS-T. ICS-S factor consistency was lower than for the ICS-T ( values: perceived popularity, .62; physical competence, .54; aggression, .57; academic, .25; afliation, .44). Six-month testretest stability for ICS-S factors (including the single item of shy) ranged from .52 to .65 (all ps < .001, n = 390). ICS-S scores were standardized by gender.

Procedure
Children were surveyed in the fall (retest assessment was in the spring) during regular class hours through a group administration. Data collection took 45 min per classroom. Children were assured that their answers would be kept condential and they were told to cover their responses. Children were told not to talk and that they could stop participating at any time. During the survey, one administrator read the instructions and questions aloud while scanning the room to check for potential problems. Additional administrators provided mobile monitoring and assisted children as needed. During survey administration teachers remained in the classroom and completed the ICS-T for each participant. All surveys were identied and distributed in a manner that concealed the identity of the participants. Measures and procedures to protect the condentiality and rights of all participants were approved by the local university Institutional Review Board for the Protection of Human Participants and the research review boards of the two school districts involved in this research.

The correlation between perceived popularity and physical competence was high (r = .79), as was the correlation between relational and overt aggression (r = .85). Six-month testretest stability for peer nominations was examined in the 17 classrooms of the larger participating school (n = 390) and ranged from .45 to .79 (all ps < .001). Social preference. Children were allowed up to six choices for their nominations of children with whom they liked most (LM) and liked least (LL) to play with, and a social preference score was created by subtracting LL from LM. Sixmonth stability for liked most and liked least measures were .45 and .61 respectively, in line with values obtained in Jiang and Cillessens (2005) meta-analysis of sociometric status measures. Interpersonal Competence ScaleTeacher. The Interpersonal Competence ScaleTeacher (ICS-T) is an 18-item bidirectional questionnaire (including two distracter items) consisting of 7-point Likert scales that teachers completed for each participant in their class. The ICS-T scale anchors (after reversing negative items) were never, not, or no for 1, sometimes, so-so, or some for the midpoint of 4, and always, very, or lots for 7. The ICS-T yields composite scores on popularity ( = .91, three items: popular with boys, popular with girls, and lots of friends), olympian/physical competence ( = .76, three items: good at sports, good looking, and wins a lot), academic ( = .83, two items: good at math and good at spelling), and aggressive ( = .87, three items: always argues, gets in trouble, and always ghts). Shy was retained as a single-item factor and the original ICS-T component of afliation (two items: always smiles and always friendly) was augmented by three additional items (cheers up others, helpful, kind to others)

Results Who bullies whom? Prevalence and stability


This section provides information on bullying prevalence and stability as obtained through the who bullies whom measure. Presented as well are data on the prevalence of cross-sex bullying, the ability of children to name bullies together with the children they harass, and differences between bullies with and without specied victims. Prevalence. Table 1 presents the percentage of boys and girls who were named by at least two classmates as bullies and/or victims. Almost 23% of participants were directly involved in bullying: 13.2% were identied as victims, 7.3% as bullies, and 2.4% as both bullies and victims (bully/victims). Victims were more likely to be girls (18% vs. 9.1%) and bullies and bully/victims were more likely to be boys (bullies: 12% vs.

INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2008, 32 (6), 473485

477

1.7%; bully/victims: 2.9% vs. 1.7%). The percentage of bullies and bully/victims who were nominated in tandem with particular victims is given in the victim specied rows of Table 1; conversely, victim not specied rows give the percentage of bullies and bully/victims without a victim on whom at least two nominators agreed. In the case of victims, participants were considered as such if they were identied by at least two classmates as victim of a specic bully; in other words, all victims identied in this study were identied as such as part of a bullyvictim dyad. Children were likely to associate particular victims with particular male bullies: 70% of male bullies (23 of 33) were nominated with peer consensus on children whom they harassed. Nominators were less likely to identify particular victims for the small number of female bullies (25% or 1 of 4) or for bullies who were victims themselves (25% or 3 of 12). Because of the small number of female bullies all subsequent analyses are restricted to male bullies only. Stability. The who bullies whom measure was re-administered in spring to children within the larger of the two schools sampled in the fall (n = 390 of 508, 77%). Table 2 shows a strong association between the distribution of bullies, victims, bully/victims, and neither bullies nor victims (2(9, N = 390) = 206, p < .001). Over half of children classied in fall as a victim (51.9%) or bully (55.6%) retained this classication into spring; 86% of children who were not involved in fall bullying or victimization remained uninvolved into spring. The bully/victim category was unstable: fall bully/victims tended to migrate into a straight bully (42.9%) or victim (28.6%)

pattern. The marginal distribution of the full sample (i.e., Table 1 total column) was compared to the marginal distributions of fall (i.e., Table 2 total column) and spring (i.e., Table 2 total row) subsamples. There were no differences [fall: 2(3) p < 1; spring: 2(3) = 7.01, p > .05], indicating similar proportions of bullies, victims, and bully/victims throughout. Because of the interchange of bully/victims with bullies and victims, fall-to-spring stability coefcients (cf., KochenderferLadd & Wardrop, 2001) were calculated among bullies (i.e., aggregate of bully and bully/victim) and victims (i.e., aggregate of victim and bully/victim). Stability coefcients were moderate for victims (r(26) = .37, p = .05) and strong for bullies (r(15) = .61, p < .01). A substantial proportion of bullyvictim dyads appeared to be stable from fall to spring. There were 21 malemale dyads in both fall and spring assessments; of these 9 (42.9%) were durable across the year. There were 47 malefemale dyads in the fall and 48 in the spring; 24 (51%) of the fall dyads remained in spring. These percentages are not independent as bullies and victims could be involved in multiple dyads. Bullies were nominated in tandem with an average of 3.31 (SD = 2.13) victims. Across both assessment waves children named over twice as many malefemale as malemale bullyvictim dyads. Independent stability estimates were obtained by correlating the percentage of bully nominations received and the number of victims specied among the 17 bullies (including bully/victims) identied in the fall subsample (15 + 2, see Table 2). Stability was strong for both measures (percentage of bully nominations: r(15) = .66, p < .005; number of victims: r(15) = .73, p < .001). Bullies with and without specied victims. According to peers, bullies with specied victims were more relationally [M (SD) = 2.42 (0.76) vs. 1.19 (0.73), t(39) = 5.01, p < .001] and overtly [M (SD) = 2.28 (0.92) vs. 1.22 (1.03), t(39) = 3.38, p < .005] aggressive than bullies without specied victims. Likewise, teachers rated bullies with specied victims as more aggressive [M (SD) = 1.40 (0.77) vs. 0.67 (0.77), t(35) = 2.78, p < .01]. The two bully types did not differ on any other peer, teacher, or self reported measure.

Table 1
Percentage (number) of children identied as bullies, victims, or bully/victims
Bullyvictim category Victims Bullies Victims specied Victims not specied Bully/victims Victims specied Victims not specied Neither Total Boys 9.1 (25) 12.0 (33) 8.4 (23) 3.6 (10) 2.9 (8) 1.1 (3) 1.8 (5) 76.0 (209) 54.1 (275) Girls 18.0 (42) 1.7 (4) 0.4 (1) 1.3 (3) 1.7 (4) 0.0 (0) 1.7 (4) 78.5 (183) 45.9 (233) Total 13.2 (67) 7.3 (37) 4.7 (24) 2.6 (13) 2.4 (12) 0.6 (3) 1.8 (9) 77.1 (392) 100 (508)

Social status and socialpersonality characteristics of bullies and victims


This section features comparisons of the popularity, preference, and socialpersonality characteristics of bullies, victims, and bully/victims as obtained through the who bullies whom measure during the fall assessment. The rst set of comparisons contrasts children classied as bullies, victims, and

Note. Percentages within a column sum to 100 over the four bullyvictim categories, and across the total row.

Table 2
Stability from fall to spring of bullying and victimization
Wave 2 (spring) Wave 1 (fall) Victims Bullies Bully/victims Neither Total Victims 51.9 3.7 28.6 11.5 16.7 (27) (1) (2) (35) (65) Bullies 1.9 55.6 42.9 2.0 6.4 (1) (15) (3) (6) (25) Bully/victims 1.9 7.4 0.0 0.3 1.0 (1) (2) (0) (1) (4) Neither 44.2 33.3 28.6 86.0 75.9 (23) (9) (2) (262) (296) Total 13.3 (52) 6.9 (27) 1.8 (7) 77.9 (304) 100 (390)

Note. Number of participants are given in parentheses.

478

RODKIN AND BERGER / WHO BULLIES WHOM?

bully/victims. The second set contrasts bullies with the specic victims they harass. Between-group comparisons. Table 3 displays differences between bullies, victims, and bully/victims on standardized peer nominations as tested through a series of one-way analyses of variance (ANOVAs) with Tukey HSD post-hoc tests. With regards to the three status measures of social preference, perceived popularity, and physical competence, results indicate that victims were more highly preferred than bullies (Z: 0.15 vs. 1.11) but bullies and victims were similar and near average on perceived popularity and physical competence. Bully/victims received very few nominations on any status measure. Social preference was disaggregated into liked most and liked least components and compared across bullies, victims, and bully/victims. Results for liked least were

signicant (F(2,113) = 23.1, p < .001): bullies (Z = 1.13) and bully/victims (Z = 1.74) were highly disliked but victims received only average levels of dislike (Z = 0.02). Results for liked most were marginally signicant (F(2,113) = 2.95, p < .06): bullies (Z = 0.30) and victims (Z = 0.05) were nominated as liked most at average levels but bully/victims were rarely nominated (Z = 0.75). On relational and overt aggression and prosocial behavior, Table 3 indicates that victims were average (Z < 0.1) but bullies and bully/victims were aggressive (Z 2.0) and not prosocial (Z < 0.50). There were no differences on shyness. Table 4 shows a similar set of comparisons for teacherand self-rated behavioral characteristics. Between-group differences on teacher assessments were smaller than for peer nominations but followed a similar pattern. Turning to the upper panel of Table 4, bullies and victims were average on

Table 3
Peer-nominated behavioral characteristics of bullies, victims, and bully/victims
Bullies Peer nominations Social preference Perceived popularity Physical competence Relational aggression Overt aggression Prosocial Shy N Z 1.11b 0.07ab 0.04ab 2.05a 2.00a 0.58b 0.29 37 SD 1.29 1.22 1.14 1.18 1.32 0.71 0.66 Z 0.15a 0.34a 0.14a 0.03b 0.05b 0.06a 0.18 67 Victims SD 1.24 1.15 1.03 0.84 0.81 1.04 0.88 Bully/victims Z 1.74b 0.71b 0.69b 2.12a 2.34a 0.87b 0.17 12 SD 1.12 0.53 0.38 1.71 1.60 0.54 0.66 F(2,113) 19.4*** 4.52* 3.32* 51.5*** 49.8*** 9.11*** <1

Note. Variables represent log-transformed scores that have been standardized by gender. Means having the same subscript are not signicantly different at p < .05 using Tukeys HSD post-hoc comparison. *p < .05; **p < .01; ***p < .001.

Table 4
Teacher- and self-reported behavioral characteristics of bullies, victims, and bully/victims
Bullies Behavioral characteristics ICS-T Popularity Olympian Aggressive Prosocial Academic Shy ICS-S Popularity Olympian Aggressive Academic Afliation Shy Z SD Z Victims SD Bully/victims Z SD Fa

0.15a 0.01 1.12a 0.20a 0.60b 0.56 0.29 0.36 0.66 0.25 0.33 0.35

1.20 1.06 0.97 1.04 0.97 0.77 1.27 1.15 1.00 0.89 1.15 0.89

0.00a 0.08 +0.18b 0.06a 0.01a 0.15 0.03 0.04 0.16 0.20 0.03 0.03

1.19 1.01 1.01 1.03 0.88 1.01 1.01 0.85 0.87 0.92 1.00 1.05

1.12b 0.58 1.48a 1.12b 0.20ab -0.29 0.19 0.05 0.40 0.31 0.25 -0.09

0.79 0.88 0.97 0.88 1.11 0.97 1.16 1.45 1.16 1.15 1.32 1.01

4.32* 1.45 13.9*** 6.17** 4.14* 2.05 <1 1.47 2.72 2.94 1.14 <1

Note. Teacher ratings are standardized by classroom and gender; self ratings are standardized by gender. Means having the same subscript are not signicantly different at p < .05 using Tukeys HSD post-hoc comparison. a Univariate F values are df = (2,99) for teacher ratings (34 bullies, 57 victims, 11 bully/victims) and (2,94) for self ratings (27 bullies, 60 victims, 10 bully/victims). *p < .05; **p < .01; **p < .001.

INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2008, 32 (6), 473485

479

popularity and olympian/physical competence (|Z|s 0.15) but bully/victims were unpopular (Z = 1.12) and low on olympian (Z = 0.58); differences were signicant for popularity only. Teachers viewed bullies and bully/victims as very aggressive (Zs > 1.0) but victims were average on aggression (Z = 0.18). Bullies and victims had comparable, average scores on prosocial behavior (|Z|s 0.20) but bully/victims were not prosocial (Z = 1.12). Teachers rated victims as more academically competent than bullies (Zs = 0.01 vs. 0.60), with bully/victims falling between the two (M = 0.20). There were no differences on shyness. The lower panel of Table 4 presents self-rated characteristics. No signicant differences between bullies, victims, and bully/victims emerged from the ICS-S. Comparisons between bullies and their victims. Differences on peer- and teacher-report variables between bullies (including bully/victims) and the average scores of their victims were calculated and submitted to one-sample (i.e., paired) t-tests. On peer nominations, victims were more highly preferred than bullies [M (SD) = 0.94 (2.36), t(25) = 2.04, p = .05] due to lower levels of dislike only [M (SD) = 0.82 (1.73), t(25) = 2.43, p < .05; liked most: t < 1]. Bullies and victims were similar on perceived popularity and physical competence. In addition, bullies were more relationally aggressive [M (SD) = 2.23 (0.96), t(25) = 11.8, p < .001], overtly aggressive [M (SD) = 1.93 (1.31), t(25) = 7.50, p < .001], and less prosocial [M (SD) = 0.55 (1.25), t(25) = 2.25, p < .05] than their victims. According to teachers, bullies were more aggressive [M (SD) = 1.07 (1.20), t(22) = 4.30, p < .001], less academically competent [M (SD) = 0.54 (1.11), t(22) = 2.33, p < .05], and marginally less shy [M (SD) = 0.39 (0.94), t(22) = 2.01, p = .06] than their victims. There were no differences between bullies and their victims on teacher ratings of popularity, olympian, or prosocial behavior. Summary. Bullies were more likely than their victims to be nominated as liked least, but there were no differences between bullies and victims on perceived popularity or physical competence. Bully/victims had a problematic prole including high aggression and low social status. Bullies and bully/victims were highly aggressive. Results from where bullies were paired with their specic victims were comparable to conventional between-group comparisons.

Social status in malemale and malefemale bullyvictim dyads


This section examines psychological power asymmetries between bullies and victims as a function of victim gender. First comes a regression analysis that distinguishes between male bullies whose targets are primarily girls versus boys. The study hypothesis is that only boys who bully other boys will be popular. Next follows an exploratory analysis that compares the social status and behavioral characteristics of male and female victims of male bullying. Male bullies harassing male versus female victims. There were 26 male bullies with peer-specied victims (23 bullies and 3 bully/victims, see Table 1). The proportion of female victims for each male bully (M = 0.58, SD = 0.41) was used as the dependent variable in three (peer, teacher, self reports) hierarchical regression models. In Step 1 of each model were entered jointly: (1) the number of victims identied with the

bully (M = 3.31, SD = 2.13), and (2) the log-transformed proportion of bullying nominations received (untransformed M = 0.39, SD = 0.22). These variables control for bullying severity, as bullies are more of a problem when they have many victims and are identied by many peers. In Step 2 of each model were entered stepwise (p < .05 entry criterion) the social status and socialpersonality characteristics featured in Table 3 (peer nominations) and the top (teacher ratings) and bottom (self ratings) panels of Table 4. Peer relational and overt aggression were aggregated to reduce colinearity. Stepwise selection was used as an accommodation to the modest sample size involved in these analyses. Of the 26 bullies in the peer regression, three were omitted from the teacher regression because teachers did not complete reports for these children. Six bullies who did not complete self-reports were omitted from the self regression. Children omitted from teacher and self analyses were similar to included children on percentage of female victims, number of victims, proportion of bullying nominations, and all peer nomination variables. The results of the hierarchical models are presented in Table 5 from peer (upper panel), teacher (middle panel), and self (lower panel) perspectives. Bullies who harassed girls were rarely nominated as popular by their peers ( = 0.56, t = 3.35, p < .01) and were rated as unpopular by their teachers ( = 0.62, t = 2.68, p < .05). Bullies who harassed girls were unlikely to be nominated by peers as shy ( = 0.42, t = 2.67, p < .01). Bullying severity (Step 1) did not discriminate boys who bully girls versus boys. There were no differences on social preference, physical competence/olympian, or aggression. Bullies self reports did not discriminate the predominant gender among the children they harass. Results were unchanged when liked most and liked least variables were substituted for peer social preference. To further illustrate Table 5 effects, bullies were categorized into bully mostly boys ( 50%, n = 11) and bully mostly girls (> 50%, n = 15) and compared using two-sample t-tests on peer perceived and teacher-assessed popularity. These tests were signicant (p < .01) and showed that boys who bullied boys were popular according to peers (M = 0.83, SD = 1.35) and moderately popular according to teachers (M = 0.25, SD = 1.29). Boys who bullied girls were unpopular according to peers (M = 0.74, SD = 0.49) and teachers (M = 1.00, SD = 0.89). Disaggregation of the ICS-T Popularity factor revealed that boys who bullied girls were rated as being unpopular among both girls (r = .50, p < .05) and boys (r = .42, p < .05). Male versus female victims of male bullying. Male and female victims of male bullying were compared by grouping the social and personality characteristics described in Tables 3 and 4 into three conceptually distinct sets. The measures within each set were used as dependent variables in three MANOVAs (or Hottelings T 2) with gender as a between-subjects factor. Set 1 included seven variables that index social status (peer perceived popularity, peer physical competence, peer social preference, teacher popularity, teacher olympian, self popularity, self olympian). Set 2 included three variables that index aggression (peer, teacher, and self; relational and overt aggregated). Set 3 included the eight remaining socialpersonality characteristics (peer prosocial, peer shy, teacher prosocial, teacher academic, teacher shy, self academic, self afliative, self shy). MANOVAs indicated victim gender differences among Set

480
Table 5

RODKIN AND BERGER / WHO BULLIES WHOM?

Social status and behavioral characteristics of boys who bully girls: Hierarchical regression models using peer, teacher, and self reports
Peer nominations (n = 26) Step 1a (Constant) Bully nominations Number of victims Perceived popularity Shy Social preference Physical competence Aggressive Prosocial B 0.76 0.103 0.016 0.19 0.23 SE B 0.17 0.083 0.041 0.057 0.083

t 4.27*** 1.23 <1 3.35** 2.67* 1.43 <1 <1 <1

R2

0.25 0.082 0.56 0.42 0.34 0.25 0.20 0.09

.080 .380 .538

Step 2 Excluded

Teacher reports (n = 23) Step 1a (Constant) Bully nominations Number of victims Popularity Prosocial Olympian Aggressive Academic Shy 0.73 0.18 0.067 0.21 0.25 0.13 0.063 0.079 2.98** 1.36 1.08 2.68* 1.01 1.07 <1 <1 <1

0.44 0.314 0.62 0.31 0.36 0.25 0.17 0.15

.053 .312

Step 2 Excluded

Self reports (n = 20) Step 1a (Constant) Bully nominations Number of victims Shy Popularity Afliation Olympian Academics Aggressive 0.35 0.040 0.091 0.26 0.13 0.055 1.31 <1 1.65 1.90 1.79 1.03 <1 <1 <1

0.087 0.48 0.39 0.38 0.24 0.13 0.082 0.004

.186

Excluded

a Entries for Step 1 variables reect values obtained after the inclusion of social-personality characteristics that passed the p < .05 selection criterion. *p < .05; **p < .01; ***p < .001.

1 status measures (F(1,7,43) = 3.35, = .647, p = .006), but not in Sets 2 or 3 (F < 1). Five of the seven social status measures were signicant in follow-up univariate tests [p .001: Peer Popularity (F = 12.4), Self Popularity (F = 11.4); p .01: Teacher Popularity (F = 8.06), Peer Olympian (F = 7.75); p .05: Peer Social Preference (F = 5.81); ns: Teacher Olympian (F = 2.82, p = .10), Self Olympian (F < 1)]. Figure 1 illustrates the results of the Set 1 analysis: female victims of male bullying had high social status; male victims of male bullying had low social status. Effects were strongest for measures of popularity and weaker for social preference and physical competence/olympian. Summary. Psychological power asymmetries between male bullies and their victims in elementary classrooms differed by whether the victim was a boy or a girl. Boys who mainly bullied other boys were popular according to teachers and peers, but boys who mainly bullied girls were unpopular. In addition, female victims had high social status while male victims did not.

Discussion
The purpose of this research was to clarify the relevance of social status to the proposition that bullies may have a psychological power advantage over the children they victimize (Olweus, 1993). Social status was conceptualized as an aspect of psychological power (Leff et al., 2004; Vaillancourt et al., 2003) and measured by social preference, popularity, and physical competence as perceived by children and teachers. Bullies were distinguished by whether peers identied them with mainly male or female victims. Results showed that patterns of social status asymmetry between bullies and victims varied according to the index of social status used and victim gender. Bullies had more social status than their victims when status was measured by popularity rather than preference, and when boys bullied other boys more than girls. Findings were obtained through a new survey instrument, the who bullies whom measure, that showed stability over a six-month period, expected behavioral correlates, and prevalence rates of bullying and victimization that accord with previous work.

INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2008, 32 (6), 473485

481

0.6 0.5 0.4 0.3

***

Popularity
** ** * ***

Olympian Social Preference

Social status (Z)

0.2 0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Peer

Teacher

Staff

Peer

Teacher

Staff

Girls Gender of victims


Figure 1.

Boys

Social status of male and female victims of male bullying. *p < .001; **p < .001, ***p < .001.

Olweus (2001) concern that preference-based measures of social status would not reveal bullies psychological power advantage over victims was borne out in this study as in others (Boulton, 1999; Vaillancourt et al., 2003). Regardless of whether their victims tended to be girls or other boys, bullies were low in social preference (i.e., Z < 1.0) because aggression engenders high levels of dislike. By contrast, correlations between bullying and aggressive behavior with likeability (i.e., liked most nominations) were weakly negative to zero. The development of two-dimensional sociometric status assessment procedures by Coie et al. (1982) was spurred by this recognition that nominations of dislike were strongly related to aggressive behavior but nominations of likeability were not. In one sense, the study hypothesis that bullies would be more popular than the children they victimize was not supported. Teacher, peer, and self perspectives on popularity and physical competence were equal and average (i.e., |Z| < 0.5) for bullies and victims alike. These null ndings bring to mind Boulton (1999), who did not nd consistent trends in the literature in bullyvictim differentials on peer perceptions of popularity. The differential patterns of preference from popularity and physical competence in bullyvictim comparisons show that social status is multidimensional, including features such as perceived popularity, dominance, being cool, and physical prowess and attractiveness that are distinct from social preference and possibly from one another (Cillessen & Mayeux, 2004; Parkhurst & Hopmeyer, 1998; Rodkin et al., 2000). For example, Cillessen and Mayeux (2004) found that preference and popularity could better be discriminated in concurrent and longitudinal analysis than the commonly recognized subtypes of relational and overt aggression. Aggressive victims of bullying (i.e., bully/victims) were an exception;

they had a distinct maladaptive prole including low social status regardless of how status was measured (Schwartz et al., 2001). The advantages of popularity as an index of social status were evident once account was taken of victim gender. As hypothesized, stepwise regression analyses indicated that peerand teacher-perceived popularity were uniquely associated with whether bullies predominately picked on other boys or girls. Boys who bullied other boys were popular. Boys who bullied girls were unpopular. Bullies physical competence, including their athleticism and attractiveness, did not distinguish the predominant gender of victims. Similarly, bullies were highly disliked and extremely aggressive regardless of whether they harassed boys or girls. Of all the social and personality characteristics of bullies included in the stepwise analyses, only bullies popularity was associated with victim gender across multiple perspectives. Of victims social and personality characteristics included in MANOVAs, popularity was the optimal discriminant of victim gender. Female victims were more popular than male victims on peer, teacher, and self reports. To a lesser extent, peers also viewed female victims as more preferred and physically competent. Bullying is a gendered phenomenon: most preadolescent girls who are victims of bullying point to boys as perpetrators (Olweus, 1993). It would have been surprising if social status dynamics between bullies and victims were not sensitive to victim gender, given deep connections between power and gender in child and adult cultures (Maccoby, 1998; Rodkin & Fischer, 2003), Indeed, study ndings supported the hypothesis that social status asymmetries would vary by whether males bullied boys or girls. The classic image of a high status male bullying a defenseless, marginalized boy was seen in this data, but so was a complementary image where psychological

482

RODKIN AND BERGER / WHO BULLIES WHOM?

power ran in the opposite direction: an unpopular, rejected aggressive boy bullying a popular girl. Malefemale bullying does not contradict Olweus (1993) power imbalance criterion (boys are presumably more physically powerful than the girls whom they harass), but what exactly this criterion entails is now an open question for future investigation. This is also the case for other denitional components of bullying, for example what satises criteria that bullying have extended duration and repeated frequency (Kochenderfer-Ladd & Ladd, 2001). The popularity advantages enjoyed by boys when bullying other boys can be interpreted from perspectives from social development and evolutionary psychology. Social development perspectives are rooted in Lewin, Lippitt, and Whites (1939) work on the emergence of powerful (but disliked) leaders victimizing helpless scapegoats among 10-year-old boys in camp settings experimentally manufactured to have autocratic versus democratic social climates. Also in this tradition are Sherifs (1956) Robbers Cave experiments where settings were manipulated so as to variously promote aggressive versus cooperative behavior, and Lippitt et al.s (1952) camp study of psychological power dynamics among emotionally disturbed and normal preadolescents. Social development perspectives explain that aggression can emerge rapidly as a valued behavior from intra- and intergroup tensions, overly restrictive group roles, and inequitable cultural contexts (Bukowski & Sippola, 2001; Cairns, 1979; Olweus, 1978; Rodkin & Hodges, 2003; Salmivalli, 2001; Wright, Giammarino, & Parad, 1986). Underlying many of these formulations is a belief that the popularity of bullies over victims is an all-too-common result of maladaptive social conditions that deny inherent human rights and dignity (Bukowski & Sippola, 2001; Olweus, 2001). Evolutionary perspectives underscore that male aggression is an adaptive solution to a variety of social problems, such as getting others resources by force or threat of force and appearing competitive in conict situations relative to other males (i.e., intrasexual rivals). A strength of evolutionary theory is specicity in predicting that status-conferring properties of male bullying will be limited to same-sex instances (Bjorklund & Pellegrini, 2002; Buss, 1999; Pellegrini, 2002). Strict adherence to evolutionary-oriented views does not easily coexist with the sentiment, fundamental to most ecological and social development perspectives, that the social status of bullies is a pathological variant of social life as opposed to the essence of what social life has always been (Pellegrini, 2007; Smith, 2007). This study does not discriminate between or synthesize social development and evolutionary perspectives; it does however highlight the importance of advancing plausible explanations to account for the popularity of same-sex aggression and bullying in elementary school environments. The large proportion of children who reported that boys bullied girls is disturbing, but has precedence in current literature (AAUW, 2001; Olweus, 1993; Rodkin & Fischer, 2003; Stein, 1995). Middle childhood is a period of uneasy relations between boys and girls (Adler & Adler, 1998; Maccoby, 1998) where antipathy between genders can run high (Abecassis et al., 2002; Rodkin et al., 2003). Stein (1995) argues that peer sexual harassment among children and adolescents is frequent, dangerous, and too often dismissed as mere romantic interest rather than bullying. McMaster et al. (2002) suggest that crosssex harassment may be distinct from same-sex harassment: only cross-sex harassment shows a developmental increase over the secondary school years, and different children are likely to harass cross- and same-sex peers. The popular girls identied in this study as the victims of male bullying are at risk of being

overlooked, as they appear well-adjusted with little need for assistance. Fortunately, study ndings support Sroufe et al. (1993) that boys who bully girls break rather than exemplify acceptable gender norms. The behavioral prole of boys who bully girls unpopular, aggressive, low in academic competence and prosocial behavior, anything but shy admits the possibility that these bullies are incompetently expressing immature romantic attraction towards what in evolutionary terminology is a valued resource (i.e., a popular girl). This study lacked the measures and power to reliably determine girls reactions to male harassment alleged by their peers. Since bullying was not predened for participants, ndings could reect little more than an elevation of normative push and poke courtship (Maccoby, 1998). Early adolescent girls may indeed have some attraction to aggressive boys (Bukowski, Sippola, & Newcomb, 2000; Pellegrini, 2002), although girls are only attracted to aggressive boys who are popular (Rodkin, Farmer, Pearl, & Van Acker, 2006). All in all, the possibility that the children of this study are reporting very unhealthy levels of malefemale bullying should be taken seriously. Despite recognition that bullying occurs disproportionately between specic dyads (Coie et al., 1999; Olweus, 1978), the relationships that bullies and victims have to one another are overlooked. Children are routinely classied as bullies and victims, but rarely is it known which bullies harass which victims. In other words: who bullies whom? The who bullies whom measure was constructed in response to this problem and its psychometric properties were promising for an initial study. Obtained prevalence rates of 7.3% bullies, 13.2% victims, and 2.4% bully/victims are within the range of similar investigations with American (Schwartz et al., 2001) and international (Olweus, 1993; Solberg & Olweus, 2003) populations, perhaps with bullies underidentied. In comparison, Pellegrini et al. (1999) determined bullies and victims using high (Z > 0.80) scores on self-reports on the Olweus Senior Questionnaire form and they were distributed as: 14% bullies, 19% victims, 5% aggressive victims. Nansel et al. (2001) surveyed a representative sample of 15,686 students in grades 6 through 10 in public and private schools throughout the USA on the Olweus (1993) measure of bullying. Self-reported prevalence rates were 13% bullies, 10.6% victims, and 6.3% bully/victims. Some comparable international prevalence rates during middle childhood are: for Italy, 10.8% bullies and 17.5% victims at rates of once a week or more during primary school (Fonzi et al., 1999, Table 9.2); for Belgium, 5.6% bullies and 9.1% victims at least once a week during primary school (Vettenburg, 1999); for Portugal, 19.6% bullies and 21.9% victims three or more times during primary school (de Almeida, 1999); and for Japan, 25.5% bullies and 21.9% victims currently are or were in the last year during elementary school (Morita, Soeda, Soeda, & Taki, 1999). Prevalence rates for the present study t comfortably within this admittedly wide range of estimates. In the present study, fourth and fth grade children were capable of responding to the complex question of naming bullies and victims conjointly. Bully and victim categories showed strong stability from fall to spring of the school year. Behavioral characteristics of bullies, victims, and bully/victims were typical of those obtained using traditional bullying measures. This study is alone in suggesting that cross-sex harassment is more prevalent than same-sex harassment (cf., McMaster et al., 2002; Pellegrini, 2007; Russell & Owens, 1999); whether these estimates reect limitations of the instrument should be investigated.

INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2008, 32 (6), 473485

483

Study limitations
There are many aspects of this research that future work could strengthen. First and foremost, renement of the who bullies whom measure should remedy its current insensitivity to the problem of female bullying and determine whether key results change when bullying is predened for children. This may involve adding instructions to children on what bullying means or specically directing children to consider both female and male bullies. We understand that there are girls who bully others, and that a central task of any measure that purports to assess bullying and victimization is to identify these aggressive, bullying girls. This limitation is an important challenge for future research. Espelage and Swearer (2003) discuss the complexities of dening and assessing bullying and peer victimization with reference to the importance of providing research participants with a statement of what is bullying. They conclude that there are strengths to having a consensual, standardized denition of bullying. However, denitions can become predenitions for child and adolescent research participants that can prime a student against speaking honestly (Espelage & Swearer, 2003, p. 369). When simply asked to report on bullies and the kids they pick on, boys and girls spontaneously nominate male bullies who are physically and relationally aggressive. Interrelated dyadic, group, and intergroup phenomena related to bullying and victimization t prevailing theories that emphasize socialecological contexts (Bukowski & Sippola, 2001; Coie et al., 1999; Espelage & Swearer, 2004; Hawker & Boulton, 2001; Juvonen & Graham, 2001; Salmivalli, 2001). Nonetheless, a second weakness of this research is that the who bullies whom measure, on its own, is not well suited for capturing the group nature of much bullying. Groups and aggressive behavior create complex opportunities for one another. Xie, Swift, Cairns, and Cairns (2002) examined how seventh graders from the Carolina Longitudinal Study spoke about their interpersonal conicts. Most conicts involved group and not just dyadic processes. Some forms of aggression, termed social aggression by Xie et al. (2002), is non-confrontational, more characteristic of girls, and uses the peer ecology as an active vehicle of attack. In cases of social aggression, which includes behaviors like gossip, rumor, and social ostracism, the identity of the perpetrator is not obvious and may become what Garandeau and Cillessen (2006) have termed invisible aggression. Unlike other forms of aggression that are reciprocated in kind, social aggression tends to be followed with direct relational aggression (that could also involve the social network). We should ask in the future: how can the who bullies whom measure be employed to capture the kind of grouporiented, social and invisible aggression bullying noted by Garandeau and Cillessen (2006) and Xie et al. (2002)? Bullies rely upon interpersonal similarity and difference and the internal organization of their groups to achieve goals that are important to them. Bullies tend to be friends with other bullies (Pellegrini et al., 1999). In addition, research showing that aggressive children interact with nonaggressive children has been growing. Aggressive children engage their peer ecology (using both aggressive and prosocial behaviors) at higher rates than nonaggressive children, and interact with aggressive and nonaggressive children alike (Pepler, Craig, & Roberts, 1998). During middle childhood, the relationship between bullies and victims involves much of the elementary classroom (OConnell et al., 1999; Pierce & Cohen, 1995). Many children who are not themselves aggressive validate

bullies with applause, or play supporting roles in bully-led peer groups (OConnell et al., 1999; Salmivalli, Huttunen, & Lagerspetz, 1997). Bullies who use proactive or instrumental aggression are often popular within their groups (Pellegrini et al., 1999) and their groups tend to be larger than those of nonbullies (Boulton, 1999). Salmivalli et al. (1997) reported that bullying was a group activity in which group members had different, distinct roles (e.g., leading the attack, assisting, reinforcing) and where bullies relied on their network of supporters, subordinates, and scapegoats to establish and exercise inuence. Hawkins, Pepler, and Craig (2001) suggested that just as peers can enable bullies, they can also be successful at intervening in a bullying episode. These studies show that although bullies preferentially afliate with one another, they are not generally segregated from their nonaggressive peers and as such bullying can be part of a heterogeneous classroom social network (Rodkin, 2004). Identication of who bullies whom provides only one aspect of the entire participant role structure and group context of bullying. A third limitation of this research is that the peer ecologies of bullying take on multiple forms, more than could be handled in the present analysis. Bullies who are also victims, or who predominately aggress in reaction to provocation, face rejection and stigmatization. Their segregation from mainstream peer ecology makes them easy to detect. Other bullies use peers as allies, and as such may be difcult to isolate or earmark for adjustment difculties. As one reviewer of this article suggested, there may be some children who bully from strength they are in a powerful position and abuse it and those who bully for other reasons perhaps, because they are bullied themselves at home and need to bully others at school to obtain self-esteem, a classic frustrationaggression issue (Rodkin & Wilson, 2007). There are many different causes and motivations for bullying, and in the context of the who bullies whom measure this means that we should not stop at identifying bullyvictim pairs but also understand how these pairs form and their phenomenologies for the victims and perpetrators of harassment. A fourth limitation is our weak ndings regarding shyness. Surprisingly, we did not replicate the typical nding that victims are more shy than their victims. Why might this be? One reason is that our victim identication procedure determined that many girls were victims of harassment, and since female victims were likely to be popular in this study (Figure 1) they were also unlikely to be shy. Thus, the zero-order correlation between victimization and shyness may have been suppressed by the high popularity of some female victims. Another reason for our lack of ndings regarding shyness is that we used relatively poor measures of shyness, internalizing behavior, and inhibition. The Cairns et al. (1995) ICS scale of shyness is not state-of-the-art, and we used only a single item peer nomination. Nonetheless, our results did include one signicant effect regarding shyness: boys who bullied girls were less popular and less shy than boys who bullied other boys. This nding adds to the portrait of boys who bully girls as unpopular, aggressive, and possibly as dysregulated and externalizing children. Future research on shy children in the study of victimization would be well-served by incorporating measures designed to tap into the multidimensional construct of inhibition. Harrist, Zaia, Bates, Dodge, and Pettit (1997), following over 500 kindergarteners over four years, were able to disaggregate social withdrawal into subtypes of unsociable, passiveanxious, activeisolate, and sad/depressed. Gazelle et al. (2005) focused

484

RODKIN AND BERGER / WHO BULLIES WHOM? American Association of University Women Educational Foundation (AAUW). (2001). Hostile hallways: Bullying, teasing, and sexual harassment in school. Washington, DC: American Association of University Women. Bjorklund, D.F., & Pellegrini, A.D. (2002). The origins of human nature: Evolutionary developmental psychology. Washington, DC: American Psychological Association. Block, J. (1971). Lives through time. Berkeley, CA: Bancroft. Boulton, M.J. (1999). Concurrent and longitudinal relations between childrens playground behavior and social preference, victimization, and bullying. Child Development, 70, 944954. Bukowski, W.M., & Cillessen, A.H.N. (Eds.) (1998). Sociometry then and now: Building on six decades of measuring childrens experiences in the peer group. San Francisco: Jossey-Bass. Bukowski, W.M., & Sippola, L.K. (2001). Groups, individuals, and victimization: A view of the peer system. In J. Juvonen & S. Graham (Eds.), Peer harassment in school: The plight of the vulnerable and victimized (pp. 355377). New York: Guilford Press. Bukowski, W.M., Sippola, L.K., & Newcomb, A.F. (2000). Variations in patterns of attraction to same- and other-sex peers during early adolescence. Developmental Psychology, 36, 147154. Buss, D.M. (1999). Evolutionary psychology: The new science of the mind. Boston: Allyn and Bacon. Cairns, R.B. (1979). Social development: The origins and plasticity of interchanges. San Francisco: W.H. Freeman. Cairns, R.B., Leung, M., Gest, S., & Cairns, B. (1995). A brief method for assessing social development: Structure, reliability, stability, and developmental validity of the interpersonal competence scale. Behavioral Research and Therapy, 33, 725736. Cillessen, A.H.N., & Bukowski, W.M. (Eds.) (2000). Recent advances in the measurement of acceptance and rejection in the peer system. San Francisco: JosseyBass. Cillessen, A.H.N., & Mayeux, L. (2004). From censure to reinforcement: Developmental changes in the association between aggression and social status. Child Development, 75, 147163. Coie, J., Cillessen, A., Dodge, K., Hubbard, J., Schwartz, D., Lemerise, E., & Bateman, H. (1999). It takes two to ght: A test of relational factors and a method for assessing aggressive dyads. Developmental Psychology, 35, 11791188. Coie, J., Dodge, K., & Coppotelli, H. (1982). Dimensions and types of social status: A cross-age perspective. Developmental Psychology, 18, 557570. Craig, W.M., Pepler, D., Connolly, J., & Henderson, K. (2001). Developmental context of peer harassment in early adolescence: The role of puberty and the peer group. In J. Juvonen & S. Graham (Eds.), Peer harassment in school: The plight of the vulnerable and victimized (pp. 242262). New York: Guilford Press. De Almeida, A.M.T. (1999). Portugal. In P.K. Smith, Y. Morita, J. Junger-Tas, D. Olweus, R. Catalano, & P. Slee (Eds.), The nature of school bullying: A cross-national perspective (pp. 174186). London: Routledge. Duncan, N. (1999). Sexual bullying: Gender conict and pupil culture in secondary schools. New York: Routledge. Espelage, D., Mebane, S., & Swearer, S. (2004). Gender differences in bullying: Moving beyond mean level differences. In D. Espelage & S. Swearer (Eds.), Bullying in American schools: A social-ecological perspective on prevention and intervention (pp. 1535). Mahwah, NJ: Erlbaum. Espelage, D.L., & Swearer, S.M. (2003). Research on school bullying and victimization: What have we learned and where do we go from here? School Psychology Review, 32, 365383. Espelage, D.L., & Swearer, S.M. (Eds.) (2004). Bullying in American schools: A socialecological perspective on prevention and intervention. Mahwah, NJ: Erlbaum. Farmer, T.W., Estell, D.B., Bishop, J.L., ONeal, K., & Cairns, B.D. (2003). Rejected bullies or popular leaders? The social relations of aggressive subtypes of rural African American early adolescents. Developmental Psychology, 39, 9921004. Fonzi, A., Genta, M.L., Menesini, E., Bacchini, D., Bonino, S., & Costabile, A. (1999). Italy. In P.K. Smith, Y. Morita, J. Junger-Tas, D. Olweus, R. Catalano, & P. Slee (Eds.), The nature of school bullying: A cross-national perspective (pp. 140156). London: Routledge. Garandeau, C.F., & Cillessen, A.H.N. (2006). From indirect to invisible aggression: A conceptual view on bullying and peer group manipulation. Aggression and Violent Behavior, 11, 641654. Gazelle, H., Putallaz, M., Li, Y., Grimes, C.L., Kupersmidt, J.B., & Coie, J.D. (2005). Anxious solitude across contexts: Girls interaction with familiar and unfamiliar peers. Child Development, 76, 227246. Graham, S., Bellmore, A., & Juvonen, J. (2003). Peer victimization in middle school: When self- and peer views diverge. Journal of Applied School Psychology, 19, 117134. Gronlund, N.E. (1959). Sociometry in the classroom. New York: Harper & Brothers.

on the anxious solitude component of shyness for over 200 fourth grade girls using observations, teacher ratings, and peer nominations. Gazelle et al. (2005) found that anxious solitary girls, who are often the victims of mistreatment among their peers, displayed fewer aversive, avoidant behaviors in unfamiliar peer contexts where the group nature of bullying is less of a factor than in familiar peer groups. A weakness of the present research is that it was not designed for a sophisticated examination of shyness, inhibition, and withdrawal and their place in the bullyvictim dynamic. It is important to understand that inhibition and avoidance are fundamental behavior systems with deep biological, psychological, and social roots that are relevant to most every aspect of personality and social development but are yet (compared to aggression) extremely difcult to reliably detect (Block, 1971; Kagan, 1994). A fth set of weakness relates to technical aspects of study methodology. The within-dyad analyses we presented had low power because, even in a sample of over 500, there were fewer than 30 bullies. As a result, the within-dyad analyses could do little more than replicate (or not) the more conventional analytic comparisons that we presented in Tables 2 and 3. We should always be grateful to have fewer, rather than more, aggressive bullies but there are serious analytic challenges for researchers who attempt to study a meaningful but rare type of child such as a bully. Another limitation of the method is that results would be stronger if nominations of unpopularity were gathered so that differentials from popularity could be calculated as is done for social preference (LaFontana & Cillessen, 2002).

Final implications
Bullies have power over their victims by denition, but what that power is and how it is expressed is a matter of empirical investigation. Boys who bully other boys t the classic Olweus (1978) pattern of a popular bully and an unpopular victim. Boys who bully girls showed the opposite pattern of an unpopular bully and a popular victim. In the cross-gender case, bullies may have physical power over their victims but not the psychological power that high social status affords. This study detected heterogeneity in bullyvictim relationships by including multiple measures of social status and a measure that allowed bullies to be named conjointly with the children they harass. Substantial amounts of cross-sex bullying was revealed using these measures that would otherwise have gone undetected. Educators should be attentive to the characteristics and safety of popular girls, and to take reports of harassment seriously. Researchers should continue to advance sociometric technologies that will allow the structure and dynamics of childrens social environments to be represented more completely (Bukowski & Cillessen, 1998; Cillessen & Bukowski, 2000). These tasks have a synergy. Sociometric data keyed into important social problems such as bullying are of great interest and usefulness to teachers (Gronlund, 1959) and are a critical element in the responsibility of society to manage school violence effectively (Mulvey & Cauffman, 2001). What adults overlook children may report quite easily.

References
Abecassis, M., Hartup, W.W., Haselager, G.J.T., Scholte, R.H.J., & Van Lieshout, C.F.M. (2002). Mutual antipathies and their signicance in middle childhood and adolescence. Child Development, 73, 15431556. Adler, P.A., & Adler, P. (1998). Peer power. Preadolescent culture and identity. New Brunswick, NJ: Rutgers University Press.

INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2008, 32 (6), 473485 Harrist, A.W., Zaia, A.F., Bates, J.E., Dodge, K.A., & Pettit, G.S. (1997). Subtypes of social withdrawal in early childhood: Sociometric status and socialcognitive differences across four years. Child Development, 68, 278294. Hawker, D.S.J., & Boulton, M.J. (2001). Subtypes of peer harassment and their correlates: A social dominance perspective. In J. Juvonen & S. Graham (Eds.), Peer harassment in school: The plight of the vulnerable and victimized (pp. 378397). New York: Guilford Press. Hawkins, D.L., Pepler, D.J., & Craig, W.M. (2001). Naturalistic observations of peer interventions in bullying. Social Development, 10, 512527. Juvonen, J., & Graham, S. (Eds.) (2001). Peer harassment in school: The plight of the vulnerable and victimized. New York: Guilford Press. Kagan, J. (1994). Galens prophecy. Cambridge, MA: Harvard University Press. Kochenderfer-Ladd, B., & Ladd, G.W. (2001). Variations in peer victimization: Relations to childrens maladjustment. In J. Juvonen & S. Graham (Eds.), Peer harassment in school:The plight of the vulnerable and victimized (pp. 2548). New York: Guilford Press. Kochenderfer-Ladd, B., & Wardrop, J. (2001). Chronicity and instability of childrens peer victimization experiences as predictors of loneliness and social satisfaction trajectories. Child Development, 72, 134151. LaFontana, K.M., & Cillessen, A.H.N. (2002). Childrens perceptions of popular and unpopular peers: A multi-method assessment. Developmental Psychology, 38, 635647. Leff, S.S., Power, T.J., & Goldstein, A.B. (2004). Outcome measures to assess the effectiveness of bullying-prevention programs in the schools. In D.L. Espelage & S.M. Swearer (Eds.), Bullying in American schools: A socialecological perspective on prevention and intervention (pp. 269294). Mahwah, NJ: Erlbaum. Lewin, K., Lippitt, R., & White, R. (1939). Patterns of aggressive behavior in experimentally created social climates. Journal of Social Psychology, 10, 271299. Lippitt, R., Polansky, N., Redl, F., & Rosen, S. (1952). The dynamics of power: A eld study of social inuence in groups of children. In G.E. Swenson, T.M. Newcomb, & E.L. Hartley (Eds.), Readings in social psychology (rev. ed., pp. 623636). New York: Holt. Maccoby, E.E. (1998). The two sexes: Growing up apart, coming together. Cambridge, MA: Harvard University Press. McMaster, L.E., Connolly, J., Pepler, D., & Craig, W.M. (2002). Peer-to-peer sexual harassment in early adolescence: A developmental perspective. Development and Psychopathology, 14, 91105. Morita, Y., Soeda, H., Soeda, K., & Taki, M. (1999). Japan. In P.K. Smith, Y. Morita, J. Junger-Tas, D. Olweus, R. Catalano, & P. Slee (Eds.), The nature of school bullying: A cross-national perspective (pp. 309323). London: Routledge. Mulvey, E.P., & Cauffman, E. (2001). The inherent limits of predicting school violence. American Psychologist, 56, 797802. Nansel, T.R., Overpeck, M., Pilla, R.S., Ruan, W.J., Simons-Morton, B., & Scheidt, P. (2001). Bullying behaviors among US youth: Prevalence and association with psychosocial adjustment. Journal of the American Medical Association, 285, 20942100. OConnell, P., Pepler, D., & Craig, W.M. (1999). Peer involvement in bullying: Insights and challenges for intervention. Journal of Adolescence, 22, 437452. Olweus, D. (1978). Aggression in the schools: Bullies and whipping boys. New York: Hemisphere. Olweus, D. (1993). Bullying at school. Oxford: Blackwell. Olweus, D. (2001). Peer harassment: A critical analysis and some important issues. In J. Juvonen & S. Graham (Eds.), Peer harassment in school:The plight of the vulnerable and victimized (pp. 120). New York: Guilford Press. Parkhurst, J.T., & Hopmeyer, A. (1998). Sociometric popularity and peerperceived popularity: Two distinct dimensions of peer status. Journal of Early Adolescence, 18, 125144. Pellegrini, A.D. (2002). Bullying and victimization in middle school: A dominance relations perspective. Educational Psychologist, 37, 151163. Pellegrini, A.D. (2007). Is aggression adaptive? Yes: Some kinds are and in some ways. In P.H. Hawley, T.D. Little, & P.C. Rodkin (Eds.), Aggression and adaptation:The bright side to bad behavior (pp. 85105). Mahwah, NJ: Erlbaum. Pellegrini, A.D., Bartini, M., & Brooks, F. (1999). School bullies, victims, and aggressive victims: Factors relating to group afliation and victimization in early adolescence. Journal of Educational Psychology, 91, 216224. Pellegrini, A.D., & Long, J.D. (2003). A sexual selection theory longitudinal analysis of sexual segregation and integration in early adolescence. Journal of Experimental Child Psychology, 85, 257278. Pepler, D.J., Craig, W.M., & Roberts, W.L. (1998). Observations of aggressive and nonaggressive children on the school playground. Merrill-Palmer Quarterly, 44, 5576. Perry, D.G., Kusel, S.J., & Perry, L.C. (1988). Victims of peer aggression. Developmental Psychology, 24, 807814.

485

Pierce, K.A., & Cohen, R. (1995). Aggressors and their victims: Toward a contextual framework for understanding childrens aggressorvictim relationships. Developmental Review, 15, 292310. Rodkin, P.C. (2004). Peer ecologies of aggression and bullying. In D.L. Espelage & S.M. Swearer (Eds.), Bullying in American schools: A socialecological perspective on prevention and intervention (pp. 87106). Mahwah, NJ: Erlbaum. Rodkin, P.C., Farmer, T.W., Pearl, R., & Van Acker, R. (2000). Heterogeneity of popular boys: Antisocial and prosocial congurations. Developmental Psychology, 36, 1424. Rodkin, P.C., Farmer, T.W., Pearl, R., & Van Acker, R. (2006). Theyre cool: Social status and peer group supports for aggressive boys and girls. Social Development, 15, 175204. Rodkin, P.C., & Fischer, K. (2003). Sexual harassment and the cultures of childhood: Developmental, domestic violence, and legal perspectives. Journal of Applied School Psychology, 19, 177196. Rodkin, P.C., & Hodges, E.V.E. (2003). Bullies and victims in the peer ecology: Four questions for psychologists and school professionals. School Psychology Review, 32, 384400. Rodkin, P.C., Pearl, R., Farmer, T.W., & Van Acker, R. (2003). Enemies in the gendered societies of middle childhood: Prevalence, stability, associations with social status, and aggression. In E.V.E. Hodges & N. Card (Eds.), Enemies and the darker side of peer relationships (pp. 7388). San Francisco: Jossey Bass. Rodkin, P.C., & Wilson, T. (2007). Aggression and adaptation: Psychological record, educational promise. In P.H. Hawley, T.D. Little, & P.C. Rodkin (Eds.), Aggression and adaptation:The bright side to bad behavior (pp. 235267). Mahwah, NJ: Erlbaum. Rose, A.J., Swenson, L.P., & Waller, E.M. (2004). Overt and relational aggression and perceived popularity: Developmental differences in concurrent and prospective relations. Developmental Psychology, 40, 378387. Russell, A., & Owens, L. (1999). Peer estimates of school-aged boys and girls aggression to same- and cross-sex targets. Social Development, 8, 364379. Salmivalli, C. (2001). Group view on victimization: Empirical ndings and their implications. In J. Juvonen & S. Graham (Eds.), Peer harassment in school:The plight of the vulnerable and the victimized (pp. 398419). New York: Guilford. Salmivalli, C., Huttunen, A., & Lagerspetz, K.M.J. (1997). Peer networks and bullying in schools. Scandinavian Journal of Psychology, 38, 305312. Scheithauer, H., Hayer, T., Petermann, F., & Jugert, G. (2006). Physical, verbal, and relational forms of bullying among German students: Age trends, gender differences, and correlates. Aggressive Behavior, 32, 261275. Schwartz, D., Proctor, L.J., & Chien, D.H. (2001). The aggressive victim of bullying: Emotional and behavioral dysregulation as a pathway to victimization by peers. In J. Juvonen & S. Graham (Eds.), Peer harassment in school: The plight of the vulnerable and victimized (pp. 145174). New York: Guilford Press. Sherif, M. (1956). Experiments in group conict. Scientic American, 195, 5458. Smith, P.K. (2007). Why has aggression been thought of as maladaptive? In P.H. Hawley, T.D. Little, & P.C. Rodkin (Eds.), Aggression and adaptation:The bright side to bad behavior (pp. 6583). Mahwah, NJ: Erlbaum. Solberg, M.E., & Olweus, D. (2003). Prevalence estimation of school bullying with the Olweus bully/victim questionnaire. Aggressive Behavior, 29, 239268. Sroufe, L.A., Bennett, C., Englund, M., Urban, J., & Shalman, S. (1993). The signicance of gender boundaries in preadolescence: Contemporary correlates and antecedents of boundary violation and maintenance. Child Development, 64, 455466. Stein, N. (1995). Sexual harassment in schools: The public performance of gendered violence. Harvard Educational Review, 65, 145162. Underwood, M.K., Schockner, A.E., & Hurley, J.C. (2001). Childrens responses to same- and other-gender peers: An experimental investigation with 8-, 10-, and 12-year-olds. Developmental Psychology, 37, 362372. Vaillancourt, T., Hymel, S., & McDougall, P. (2003). Bullying is power: Implications for school-based intervention strategies. Journal of Applied School Psychology, 19, 157176. Vettenburg, N. (1999). Belgium. In P.K. Smith, Y. Morita, J. Junger-Tas, D. Olweus, R. Catalano, & P. Slee (Eds.), The nature of school bullying: A crossnational perspective (pp. 187204). London: Routledge. Wright, J.C., Giammarino, M., & Parad, H.W. (1986). Social status in small groups: Individual-group similarity and the social mist. Journal of Personality and Social Psychology, 50, 523536. Xie, H., Swift, D.J., Cairns, B.D., & Cairns, R.B. (2002). Aggressive behaviors in social interaction and developmental adaptation: A narrative analysis of interpersonal conicts during early adolescence. Social Development, 11, 205224. Zeller, M., Vannatta, K., Schafer, J., & Noll, R.B. (2003). Behavioral reputation: A cross-age perspective. Developmental Psychology, 39, 129139.

Вам также может понравиться