Social network predictors of bullying and victimization.

Mouttapa, Michele ; Valente, Tom ; Gallaher, Peggy 等

The school context provides an opportunity for youth to socialize with selected peers, independently from adults (Youniss & Smollar, 1989). Friends make unique contributions to each other's learning, emotional support, and socialization beyond that of their parents (Hartup & Sancilio, 1986). Validation from friends provides psychosocial support that leads to healthy development and adjustment (Ladd, Kochenderfer, & Coleman, 1996; Harris, 1995). However, adolescents also face pressures to live up to the norms of their friendship group (Brown, Dolcini, & Leventhal, 1997), which may include involvement in bullying behaviors. For this reason, the friendship network, the pattern of friendships among individuals within a group, is an important aspect of adolescent school bullying.

Friendship networks are associated with several health risk behaviors, including smoking (Alexander, Piazza, Mekos, and Valente, 2001; Ennett & Bauman, 1994), risky sexual behaviors (Ennett, Bailey, & Federman, 1999), drug use (Bauman & Ennett, 1996), and syringe sharing among drug users (Valente & Vlahov, 2001). Friendship network characteristics are also associated with bullying (Huttunen, Salmivalli, & Lagerspetz, 1996) and victimization (Graham & Juvonen, 1998).

School Bullying--Prevalence and Correlates

Bullying in elementary schools and high schools is well documented and is recognized as a growing problem in the United States, Australia, several European nations, and some Asian countries, including Japan (Smith & Brain, 2000). The social context in which bullying occurs in Western and Eastern cultural settings may have similarities (Schwartz, Farver, Chang, & Lee-Shin, 2002). Within the United States, The Kaiser Family Foundation (Acre [CNN report], 2001) found that 8- to 15-year-olds considered bullying a "big problem," ranking higher than racism, AIDS, and peer pressure to use drugs and alcohol. The Centers for Disease Control and Prevention (2001) reported that one in eight high school students nationwide was in at least one physical fight on school property during the past year. Both bullying and victimization are associated with intrapersonal problems such as anxiety and depression (Kumpulainen & Rasanen, 2000; Salmon & West, 2000; Kumpulainen, Rasanen, & Puura, 2002), eating disorders (Kaltiala-Heino, Rimpela, Rantanen, & Rimpela, 2000), low self-esteem (O'Moore & Kirkham, 2001), and less satisfaction with school (Kochenderfer & Ladd, 1996; Karatzias, Power, & Swanson, 2002).

Who Are Bullies, Victims, and Aggressive Victims?

Since bullies, victims and aggressive victims (those who are both bullies and victims) may have unique patterns of friendships and social status, it is important to clearly define the behaviors and common characteristics of these three groups. Bullies are those students who physically and/or emotionally harm another student repeatedly over time (Olweus, 1991). An imbalance of power exists, such that the victim has difficulty defending him/herself from aggressors (Olweus, 1991). In this case, the aggressor also has the distinction of being a "bully" because there is no retaliation. Bullies represent approximately 7-15% of the school-aged population (Pelligrini, 1998), and have been described as having a strong need to dominate others (Olweus, 1991) and the social skills and understanding of others' emotions to do so (Sutton, Smith, & Sweetenham, 1999). Collins and Bell (1996) found that bullies have higher peer-nominated scores on sociability and leadership relative to other students.

Victims are those students who are frequent targets of aggressive, hurtful actions, and provide little defense against their aggressors. Victims represent approximately 2-10% of the school-aged population (Pelligrini, 1998), and have been characterized by their cautious, sensitive, and quiet mannerisms (Olweus, 1991) and low self-esteem (Collins & Bell, 1996).

Aggressive victims, or "bully-victims" (e.g., Andreou, 2000), are those who engage in aggressive behaviors and are also victims of aggression. It is believed that aggressive victims represent 2-10% of the student population (Pelligrini, 1998), and are characterized by their reactivity, poor emotional regulation, academic difficulties, and peer rejection (Schwartz, 2000), as well as learning difficulties (Kaukiainen et al., 2002). Generally, previous studies have limited aggressive victim status to those students who score extremely high on raw or standardized measures of both aggression and victimization (see Schwartz, Proctor, & Chien, 2001).

Dominance Theory and Social Cognitive Theory

Dominance theory (Hawley, 1999) and social cognitive theory (Bandura, 2002) provide explanations of how the social network may influence bullying behaviors. Dominance theory posits that students use aggression against weaker students to gain access to resources, including high sociometric status among peers, whereas social cognitive theory posits that adolescents model their friends' behaviors, including aggressive behaviors. The two theories are not mutually exclusive. However, dominance theory suggests that aggression is associated with high sociometric status, whereas social cognitive theory suggests that aggression is associated with peers' aggressive behaviors.

Gender Differences in Bullying Behaviors

The types of bullying that males and females engage in vary. Compared to females, males are more often involved in physical forms of bullying (e.g., kicking, pushing), whereas females are more often involved in indirect forms of bullying (e.g., rumor spreading, social ostracism) (Crick, Casas, & Ku, 1999; Baldry & Farrington, 1999; Rivers & Smith, 1994). There is also evidence that indirect forms of aggression are more often tolerated among males, and are associated with social acceptance among their peers (Salmivalli, Kaukiainen, & Lagerspetz, 2000).

The strategies that victims use to cope with bullying also vary by gender. Male victims are less likely to tell anyone that they were bullied (Cowie, 2000) and more often retaliate with aggression relative to female victims (Salmivalli, Karhunen, & Lagerspetz, 1996). Female victims, on the other hand, more often respond to bullying with helplessness (Salmivalli, Karhunen, & Lagerspetz, 1996).

Ethnic Differences in Bullying and Victimization

The ethnic composition of the classroom has been related to classroom levels of aggression (Rowe, Almeida, & Jacobson, 1999), aggression among ethnic majorities (Graham & Juvonen, 2002), and victimization among ethnic minorities (Hanish & Guerra, 2000). There is also some evidence that attitudes toward fighting vary by ethnicity (Arbona, Jackson, McCoy, & Blakely, 1999). Rodkin et al. (2000) found that "tough boys," aggressive students with high centrality (popularity) scores, were disproportionately African American. Relatively few studies have examined bullying and social networks in a multiculrural setting.

Social Network Analysis

Social network analysis is a set of methods and techniques used to analyze social relationships (Scott, 2000; Wasserman & Faust, 1994; Valente, 1995). Network analysis requires relational data, or information about who is connected to whom within a group (e.g., friendship connections within a classroom of students; (Scott, 2000). Using network analysis, researchers can determine whether sociometric status, or one's social position within a group, is associated with individual attributes (e.g., leadership qualities, extraversion). Network analysis can also be used to determine whether the specific attitudes, beliefs, and behaviors of one's social ties (e.g., friends, coworkers) influence one's own attitudes, beliefs, and behaviors.

Sociometric status, aggression, and victimization. Network analysis has been employed to determine whether sociometric status is related to aggression and victimization. Centrality, an index of popularity, is linked to both prosocial and antisocial behaviors (Pakaslathi, 2001; Gest, Graham-Bermann, & Hartup, 2001). In their study of 4th- to 6th-grade males, Rodkin, Farmer, Pearl, and Van Acker (2000) found that "model boys" (those who were perceived as nonaggressive, athletic, leaders, cooperative, studious, and outgoing) and "tough boys" (those who were perceived as popular, aggressive, and physically competent) both occupied central positions in their network of classmates. Furthermore, social norms can vary across classrooms such that aggression, rather than prosocial behavior, is related to high centrality in some classrooms (Xie, Cairns, & Cairns, 1999). In these studies, the causal relationship between agression and sociometric status was not determined.

Some studies have specifically examined the sociometric characteristics of bullies and victims. Compared to other students, bullies have larger friendship groups (Huttunen, Salmivalli, & Lagerspetz, 1996), higher sociometric rankings (Bjoerkqvist, Oesterman, Lagerspetz, Landau, Caprara, Vittoro, & Fraczek, 2001), and earlier dating experiences after controlling for pubertal development (Connolly, Pepler, Craig, & Taradash, 2000). Victimization has been associated with loneliness and peer rejection (Bjoerkqvist et al., 2001; Graham & Juvonen, 1998). However, there is longitudinal evidence that reciprocal friendships (e.g., situations where students nominate a friend and receive a friendship nomination from that friend) protect students against victimization (Boulten, Trueman, Chau, Whitehand, & Amatya, 1999). Reciprocal friendships may also buffer the psychological consequences of victimization, including withdrawal and depression (Hodges, Malone, & Perry, 1997; Hodges, Boivin, Vitaro, & Bukowski, 1999).

Friends' aggressive behaviors. Some studies have employed network analytic techniques to determine whether friends' aggressive behaviors are associated with one's own aggressive behaviors. Friends are highly similar in their aggressive behaviors, as well as other behaviors, traits, and indicators of social status (Haselager, Hartup, Van Lieshout, & Riksen-Walraven, 1998; Kupersmidt, DeRosier, & Patterson, 1995). This finding has been documented in studies of preadolescent and adolescent males (e.g., Tremblay, Masse, Vitaro, & Dobkin, 1995; Poulin, Cillessen, Hubbard, & Coie, 1997). Among females, there is longitudinal evidence that friends' bullying behaviors more strongly predict bullying even compared to their own baseline bullying behaviors (Salmivalli Lappalainen, & Lagerspetz, 1998). However, not all adolescents are equally influenced by their friends' behaviors. For example, exposure to disruptive friends leads to increases in delinquency only among those students who are moderately disruptive themselves (Vitaro, Tremblay, Kerr, Pagani, & Bukowski, 1997).

The Present Study

The application of network analysis to bullying in schools has made an important contribution to the literature by providing evidence that the friendship network is relevant. In their review, Naylor and Cowie (1999) comment that peer-led support systems, including befriending, conflict resolution, and counseling programs, have, in some cases, successfully reduced bullying and created a prosocial classroom climate.

To date, social network studies have generally examined either bullying (e.g., Rodkin et al., 2000; Xie, Cairns, & Cairns, 1999) or victimization (e.g., Boulton et al., 1999; Hodges et al., 1997), but not aggressive victimization, or the simultaneous occurrence of both behaviors. Because many students engage in aggressive behaviors, and, to a similar extent, become victimized (see Twemlow, Fonagi, & Sacco, 2001), it is important to examine the friendship patterns of aggressive victims. Network analysis of bullies, victims, and aggressive victims would provide new insights into the social etiology and persistence of bullying (e.g., situations where one student continuously victimizes another student) and aggression (e.g.,situations where two students victimize each other to a similar extent.

In the present study, we assessed physical and verbal forms of aggression and victimization in a sample of Southern California 6th graders, of whom the majority were Hispanic or Asian American. From the data, we identified self-reported bullies, victims, and aggressive victims. We examined the following: (1) whether bullies, victims, and aggressive victims differed from their classmates on sociometric variables and their friends' involvement in physical and verbal aggression, and (2) whether ethnic differences existed among bullies, victims, and aggressive victims. All analyses were performed on the entire sample and among females and males separately.

Consistent with dominance theory (Hawley, 1999) and previous findings (Pelligrini, 1998; Pelligrini & Smith, 1998), we expected that bullies would occupy more central network positions and victims would occupy less central network positions relative to their classmates. Consistent with social cognitive theory (Bandura, 2002) and previous findings (Poulin et al., 1997; Salmivalli et al., 1998), we expected that the friends of bullies and aggressive victims would score higher on self-reported aggression relative to the friends of other students. Similar to the findings of Graham and Juvonen (2002), who also examined aggression and victimization in a Latino-majority Southern California sample, we predicted that bullies would be disproportionately Latino, where as victims would be disproportionately Asian and other ethnic minorities.

METHOD

Sample

The data described in this article are from the baseline survey of a longitudinal school-based experimental trial of smoking prevention strategies in an urban population of primarily Latino and Asian adolescents in Southern California. The respondents were 6th-grade students from 16 schools participating in this study.

Student recruitment. A total of 47 Southern California school districts were identified for possible participation in the study with 36 being solicited to participate (11 districts were too far from the research center). Of the 36 contacted, 10 districts declined to participate. The 26 districts that agreed to participate had 150 schools. Of these 150 schools, 104 were approached for participation. To be eligible for the study, a school met the following requirements: (1) administrator approval, (2) a majority of students who were Hispanic/Latino or multiethnic (no single predominant ethnicity), with at least 30% Asian American, (3) geographic proximity to the researchers, and (4) 80% consent rates from parents. Of the 104 schools approached, 38 did not receive administrative approval, 28 did not meet ethnicity or driving distance criteria, and 5 did not meet parental consent criteria. Of the 33 schools remaining, 9 were randomly selected for pilot testing of survey instruments and curriculum materials, and 24 were randomly selected for inclusion in the study. Sixteen of the 24 schools were then randomly assigned to the program group to receive a smoking prevention curriculum and 8 were assigned to the control group.

All 6th-grade students in the participating schools were invited to participate in the study. Students who agreed to participate and provided written parental consent were included. Of the 4,427 students invited, 3,326 (75.13%) provided parental consent. Of those students, 217 did not complete the survey because they were absent from school on the day of data collection or chose not to participate, leaving 3,109 who completed the baseline survey.

Students attending one of the 16 program schools also completed a shorter survey, which included items that asked students to write down the names of their five closest friends. The purpose of collecting friendship data was to investigate whether friendship patterns are associated with smoking and using other substances. A total of 1,368 students provided complete data on all of the variables of interest as well as friendship data. These students comprised the analytic sample in this study.

Procedure

Students completed a 160-item paper-and-pencil survey in their classrooms during a single class period (45-50 minutes). Trained data collectors, who were not previously acquainted with the students, distributed the surveys. The surveys were identified only by a code number, not with the students' names or any other identifying information. Because all the students were attending schools in which their classes were conducted only in English (as required by California law), a basic level of English-language proficiency was assumed and the surveys were provided only in English. However, students were encouraged to ask the data collectors to clarify the meanings of any unfamiliar words.

Measures

Categorization of bullies, victims, and aggressive victims. Four items were adopted from Olweus (1991) to assess self-reported physical and verbal forms of aggression and victimization during the past three months. The aggression items were the following: "Did you push or hurt another kid?" "Did you threaten another kid or say something mean to him or her?" The victimization items were the following: "Did another kid hit you, push you, or hurt you in any way?" "Did another kid threaten you or say something mean to you?" The response options for all items were: 3 = a lot, 2 = sometimes, 1 = once in a while, and 0 = never. A total aggression score and a total victimization score were calculated by summing responses on their two respective items. Hence, bullying scores and victimization scores ranged from 0 to 6. Three dichotomous variables, bully (yes/no), victim (yes/no), and aggressive victim (yes/no) were created. Students were classified as "bullies" if they scored 4 or higher on aggression and less than 4 on victimization, "victims" if they scored less than 4 on aggression and 4 or higher on victimization, and "aggressive victims" if they scored 4 or higher on both aggression and victimization. We used a cutoff of 4 so that students who were moderately to frequently involved in bullying, victimization, and aggressive victimization were identified.

Friendship network variables. Friendship network variables were assessed with the item: "Name your five best friends in your class." Students selected friends from their classroom roster, which was included in the survey materials. Five blanks were provided to fill in their friends' first and last names. No specific rank order was assigned to the blanks. Each student in the classroom was assigned a numeric code. At a later time, data collectors corrected the spelling of friends' names where necessary, and wrote the numeric code of each friend on the survey for data entry. The friendship network data were used to assess the following sociometric variables: nominations sent, nominations received, reciprocity, friends' bullying, and friends' victimization.

Nominations sent was a count of the number of classmates a student nominated as friends, and ranged from 0 to 5. Nominations received was a count of the number of classmates who nominated the student as a friend, and ranged from 0 to 15. Reciprocity was the proportion of nominations sent that was reciprocated with a nomination received. Friends' aggression was the mean aggression score of the classmates that the student nominated, and ranged from 0 to 6. Friends' victimization was the mean victimization score of the classmates that the student nominated, and ranged from 0 to 6.

Ethnicity. Self-reported ethnicity was assessed with eight dichotomous questions: "Are you: (1) White? (2) Chinese/Chinese-American? (3) Pacific Islander? (4) Filipino? (5) Korean/Korean-American? (6) Vietnamese/Vietnamese-American? (7) Latino/Hisapnic? (8) Black/African-American?" Response options for each were: 1 = yes and 0 = no. Since Latinos and Asians comprised the majority of our sample (53.8% and 22.8%), respectively) students were classified as Latino, Asian (those who answered "yes" to being Chinese, Pacific Islanders, Filipinos, Koreans, or Vietnamese), or other (those who did not identify themselves as being either Asian or Latino).

Analyses

Gender differences. Frequencies on the outcome variables and means on the independent variables were calculated by gender. Gender differences were assessed with chi-square tests and F-tests.

Regression of bullies, victims, and aggressive victims. The dichotomous outcome variables in this study were bullying, victimization, and aggressive victimization. Logistic regression models were estimated to determine whether sociometric status variables (friendship nominations sent, received, and reciprocated), friends' influences (friends' involvement in bullying and victimization), and ethnicity were associated with each of the dependent variables separately: classification as a bully, victim, and aggressive victim. Each dependent variable was examined among the entire sample, among males separately, and among females separately. Therefore, a total of nine logistic regression models were estimated. The Bonferroni adjustment was used to correct for the experiment-wide error rate. Therefore, we report those results that were significant at the .006 level of probability (.05 divided by 9 logistic regression models).

RESULTS

Demographic Characteristics of the Sample

A total of 1,368 students (52.9% female) provided complete data, and comprised the analytic sample. Their mean age was 11.3 years (SD = 0.53). Latinos were the ethnic majority in this sample (53.8%), followed by Asians (22.8%). The remaining 23.4% included non-Hispanic Whites, African Americans, and other ethnic groups (classified as "other").

Students with complete data, who were included in the analyses, differed from students with missing data on all of the variables of interest. Students with complete data were more often bullies (p < .001) and victims (p < .0001) and less often aggressive victims (p < .0001) relative to those with missing data. The mean scores of friends' bullying and victimization were lower among those with complete data (p < .01). Those with complete data sent and received more friendship nominations (p < .0001), had a higher proportion of reciprocated friendships (p < .05), were more often Asian (p < .0001), and less often Latino (p < .01).

Gender Differences on Ethnicity and Network Variables.

Table 1 presents gender differences on network characteristics and outcome variables. Males had more friends who were bullies (3.6) and victims (4.0) relative to females (3.1 and 3.8, respectively). Females had a higher proportion of reciprocated friendships (58.7%) relative to males (54.3%). Consistent with previous findings (e.g., Crick, Casas, & Ku, 1999), males were more often bullies (9.3%) and aggressive victims (36.3%) relative to females (5.8% and 26.1%), respectively).

Logistic Regression Analyses

Bullies. Table 2 presents logistic regression results for bullies. As predicted, the friends of bullies reported more aggression, both in the entire sample (AOR [adjusted odds ratio] = 1.32, p < .0001) and in the subsample of females (AOR = 1.46, p < .006). Female bullies received fewer friendship nominations (AOR = 0.77, p < .0001), but had a higher proportion of reciprocated friendships (AOR = 7.45, p < .006). This finding suggests that female bullies have smaller, more cohesive friendship groups relative to other females. Contrary to our predictions, male bullies did not score higher on any of the sociometric measures.

Victims. Table 3 presents logistic regression results for victims. As predicted, victims received fewer friendship nominations (AOR = 0.91, p < .0001), and were disproportionately Asian (AOR = 1.57, p < .0001). Furthermore, the friends of victims reported less aggression (AOR = 0.83, p < .006). Contrary to previous findings (e.g., Boulton et al., 1999), victims did not have a lower proportion of reciprocated friendships relative to other students.

Aggressive victims. Table 4 presents logistic regression results for aggressive victims. As predicted, the friends of aggressive victims were more aggressive (AOR = 1.15, p < .006). When the analyses were stratified by gender, however, this association held true for females (AOR = 1.23, p < .006), but not males (AOR = 1.13, ns). Finally, the friends of aggressive victims were less victimized (AOR = 0.88, p < .006).

DISCUSSION

Several studies have examined the friendship network characteristics of bullies and victims. However, the present study compared bullies, victims, and aggressive victims with other adolescents on multiple network characteristics. Our findings suggest that social cognitive theory, rather than social dominance theory, best explains the friendship patterns associated with bullying, victimization, and aggressive victimization among adolescents. Consistent with social cognitive theory (Bandura, 2002), we found that bullies and aggressive victims tended to nominate friends who are also aggressive. These findings suggest that the presence of aggressive friends is associated with participation in aggression, whereas the presence of nonaggressive friends is associated with less participation in aggression. Similarly, in a longitudinal study, Warman and Cohen (2000) found that preadolescents who discontinued their aggressive behaviors were friends with students who were less aggressive compared to those students who continued their aggressive behaviors.

Our findings also indicate that the presence of aggressive friends is associated with lower rates of victimization, whereas the presence of nonaggressive friends is associated with higher rates of victimization. Aggressive students (aggressive victims, and to a lesser extent, bullies) were friends with students who were less often victimized. Victims, on the other hand, were friends with students who were less often aggressive. Previous research indicates that reduced victimization is associated with "having a friend help" among kindergarten males (Kochenderfer & Ladd, 1997), and having friends who are physically capable of fulfilling a protective function among early adolescents (Hodges, Malone, & Perry, 1997). Hence, a possible explanation for our findings is that aggressive students defend their friends when attacked by aggressors, whereas nonaggressive friends may leave students vulnerable to aggressors. It is also possible that students aggregate in friendship groups, according to preferences for aggressive behaviors.

Dominance theory (Hawley, 1999) posits that aggression facilitates access to a central position in the peer network. On the contrary, male bullies did not differ from other males on measures of sociometric status. Such findings are similar to those of Rodkin et al. (2000), who found that males who occupy central positions in the peer network were heterogeneous in regard to aggressive behaviors. A clearly different pattern was observed among female bullies. Female bullies received fewer friendship nominations, but their friendships were more often reciprocated. This suggests that female bullies occupy less central network positions, but have stronger ties to their friends. This might have occurred because physical and verbal bullying is less common among adolescent females (e.g., Crick, Casas, & Ku, 1999), as well as in this study. Hence, the relatively few female bullies may have selected each other as friends based on their similar preferences for direct forms of bullying. Haselager et al. (1998) suggests that aggressive students select each other as friends based on their similar preferences for various behaviors, including aggressive behaviors. This may be especially the case for female bullies who use physical and verbal forms of aggression. Since females tend to use indirect forms of bullying (e.g., rumor spreading, social ostracism), further research is needed to determine whether indirect bullying is associated with high sociometric status among females. Furthermore, future studies should examine whether a central position in the friendship network is associated with dominance over classmates among males and females.

Consistent with dominance theory, victims occupied less central positions in the friendship network relative to other students, as evidenced by the fewer friendship nominations they received. Contrary to previous research (e.g., Boulton et al., 1999). victims did not have fewer reciprocated friendships. Such findings suggest that the number of connections in the friendship network, rather than the reciprocation of friendships, may protect students against victimization.

This study was conducted on a primarily Latino sample of adolescents. However, some of the schools in the study had an Asian majority. Hence, we conducted preliminary analyses to determine whether ethnic majorities (e.g., Asians in the predominantly Asian schools, and Latinos in the predominantly Latino schools), were more often bullies, and ethnic minorities were more often victims. No such differences were found. The general finding was that Asians were the most frequently victimized ethnic group. Furthermore, when grouped together, the remaining students (Whites, African Americans, and other ethnic groups) were not victimized more frequently than Latinos, the majority ethnic group. In predominantly White schools in Europe, Asian students are more often victims of racist name-calling (Boulton, 1995; Moran, Smith, Thompson, & Whitney, 1993). Perhaps Asian students in our sample were victimized more often because of their ethnicity, or were more likely to report victimization relative to other ethnic groups. More in-depth research is needed to determine why Asian Americans report higher rates of victimization.

Limitations

Because the present study included only those students who attended schools that met all of the inclusion criteria, the results may not generalize to adolescents attending schools with different ethnic compositions, low administrative approval, and low parental consent. The results may or may not generalize to regions outside of Southern California. Furthermore, since students with complete data differed from students with missing data, the results may not generalize as well to Latino students, aggressive victims, and those who have friends that are more frequently aggressive and victimized.

The results are based on adolescents' self-reports, which might be biased. Self-reports tend to yield lower rates of bullying and victimization relative to peer reports (Pakaslathi & Keltikangas-Jarvinen, 2000). Schwartz, Proctor, and Chien (2001) suggest that self-reported aggressive victims and peer-reported aggressive victims may not be equivalent groups. To address these issues, future studies should utilize both peer and self-reported measures of bullying and victimization. Ideally, it would be useful to obtain data on who aggresses against whom. Such information will help determine whether a student is a bully in some social situations, a victim in others, and both in yet other situations, as suggested by Pellegrini (1998).

Self-reports of bullying and victimization in this study were based on perceived frequencies within the past three months, rather than on actual frequencies. To obtain more accurate frequencies, perhaps daily or weekly occurrences of bullying and victimization should be assessed in future studies through short surveys or diaries.

The results are based on cross-sectional analyses. Therefore causal relationships between friendship network characteristics and bullying, victimization, and aggressive victimization cannot be inferred. Longitudinal studies would help determine whether aggressive students select each other as friends, and whether new associations with aggressive students result in the adoption of aggressive behaviors.

Implications

The present study adds to the existing literature by examining the association of school bullying with several characteristics of the friendship network in a sample of primarily Latino and Asian adolescents, who are frequently underrepresented in studies of bullying. We found that friends' involvement in aggression was a strong predictor of aggression. This is consistent with social cognitive theory (Bandura, 2002), which posits that behaviors are learned and reinforced within the peer group. Consistent with dominance theory (Hawley, 1999), victims had fewer social connections relative to other students. Furthermore, we found that the friends of aggressive students were victimized less often than other students. In sum, the findings suggest that violence prevention efforts targeting highly aggressive students may also effectively reduce the aggressive behaviors of their friends. Furthermore, assertiveness training in handling aggressive situations could be beneficial for both aggressive and nonaggressive students alike. Such training may help students defend themselves and their friends effectively from aggressors using nonviolent strategies.

This research was supported by the University of Southern California Transdisciplinary Tobacco Use Research Center (TTURC), funded by the National Institutes of Health (Grant 1 P50 CA84735-01), the California Tobacco-Related Disease Research Program (TRDRP; Grant 7PT-7004), and the National Cancer Institute (Grant T32 CA 09492). The authors thank Gaylene Gunning, Steven Cen, and the TTURC/IRP project staff for assistance with data collection and data management. The authors would also like to thank Donna Spruijt-Metz, Stove Sussman, Allen Tien, and Mary Ann Pentz for their thoughtful comments.

Table 1
Gender Differences on Network Characteristics and Outcome Variables

 Males (n = 645) Females (n = 723)
 M SD M SD F
Network Characteristics
 Nominations Sent 4.63 0.78 4.70 0.72 2.89
 Nominations Received 4.53 2.61 4.67 2.52 1.01
 Reciprocated 0.54 0.30 0.59 0.29 7.66 **
 Friends' Bullying 3.59 1.09 3.08 0.90 86.94 ***
 Friends' Victimization 4.02 1.02 3.77 1.03 20.50 ***
 n % n % [[chi.sup.2]
Outcome Variables
Bully 60 9.3 42 5.8 6.03 **
Victim 197 27.3 167 25.9 0.32
Aggressive Victim 234 36.3 189 26.1 16.40 ***

*p < .05, **p < .01, ***p < .001.

Table 2
Logistic Regression of Bullying on Friends' Behavior,
Sociometric Status, and Ethnicity, by Gender

 Total (n = 1368)
 AOR 95% CI
Friends'
Behavior
 Aggressive 1.32 *** 1.16-1.52
 Victimized 0.81 * 0.67-0.97
Sociometric
Status (a)
 Sent 1.13 * 1.01-1.27
 Received 0.98 0.90-1.07
 Reciprocated 1.88 0.40-8.78
Ethnicity (b)
 Latino 1.32 0.67-2.60
 Asian 1.46 0.68-3.18

 Males (n = 645)
 AOR 95% CI
Friends'
Behavior
 Aggressive 1.21 * 1.02-1.44
 Victimized 0.87 0.63-1.20
Sociometric
Status (a)
 Sent 0.92 0.71-1.18
 Received 1.11 0.95-1.29
 Reciprocated 0.83 0.09-7.60
Ethnicity (b)
 Latino 1.82 * 1.11-2.99
 Asian 1.67 * 1.06-2.61

 Females (n - 723)
 AOR 95% CI
Friends'
Behavior
 Aggressive 1.46 ** 1.11-1.92
 Victimized 0.75 0.51-1.09
Sociometric
Status (a)
 Sent 1.96 0.87-4.43
 Received 0.77 *** 0.68-0.87
 Reciprocated 7.45 ** 2.05-27.02
Ethnicity (b)
 Latino 0.82 0.21-3.16
 Asian 0.99 0.24-4.07

Note. Underlined values represent odds ratios that remained
significant after Bonferroni adjustment.

(a) Sociometric status indicators derived from friendship nominations
sent, received, and reciprocated.

(b) Reference group = other.

*p < .05, **p < .006, ***p < .0001.

Table 3
Logistic Regression of Victimization on Friends' Behavior,
Sociometric Status, and Ethnicity, by Gender

 Total (n = 1368)
 AOR 95% CI
Friends'
Behavior
 Aggressive 0.83 ** 0.73-0.94
 Victimized 1.10 0.95-1.27
Sociometric
Status (a)
 Sent 0.88 0.73-1.07
 Received 0.91 *** 0.86-0.95
 Reciprocated 1.07 0.76-1.50
Ethnicity (b)
 Latino 1.41 0.92-2.15
 Asian 1.57 *** 1.23-2.01

 Males (n = 645)
 AOR 95% CI
Friends'
Behavior
 Aggressive 0.80 * 0.65-0.99
 Victimized 1.14 0.90-1.45
Sociometric
Status (a)
 Sent 0.89 0.72-1.10
 Received 0.89 0.78-1.01
 Reciprocated 1.34 0.51-3.50
Ethnicity (b)
 Latino 1.49 * 1.05-2.14
 Asian 1.67 ** 1.18-2.36

 Females (n = 723)
 AOR 95% CI
Friends'
Behavior
 Aggressive 0.86 0.70-1.06
 Victimized 1.04 0.85-1.26
Sociometric
Status (a)
 Sent 0.85 0.68-1.06
 Received 0.92 ** 0.87-0.98
 Reciprocated 0.88 0.46-1.69
Ethnicity (b)
 Latino 1.34 0.75-2.41
 Asian 1.46 * 1.01-2.10

Note. Underlined values represent odds ratios that
remained significant after Bonferroni adjustment.
(a) Sociometric status indicators derived from friendship
nominations sent, received, and reciprocated.

(b) Reference group = other.

*p < .05, **p < .006, ***p < .0001.

Table 4
Logistic Regression of Aggressive Victimization on Friends'
Behavior, Sociometric Status, and Ethnicity, by Gender

 Total (n = 1368)
 AOR 95% CI
Friends'
Behavior
 Aggressive 1.15 ** 1.05-1.26
 Victimized 0.88 ** 0.81-0.96
Sociometric
Status (a)
 Sent 1.10 0.86-1.40
 Received 0.93 0.84.1.04
 Reciprocated 1.13 0.65.1.97
Ethnicity (b)
 Latino 0.84 0.57-1.23
 Asian 1.12 0.77-1.62

 Males (n = 645)
 AOR 95% CI
Friends'
Behavior
 Aggressive 1.13 0.96-1.33
 Victimized 0.79 * 0.65.0.97
Sociometric
Status (a)
 Sent 1.08 0.83-1.40
 Received 0.95 0.83-1.09
 Reciprocated 0.93 0.48-1.78
Ethnicity (b)
 Latino 0.58 0.39-0.86
 Asian 0.95 0.67-1.35

 Females (n = 723)
 AOR 95% CI
Friends'
Behavior
 Aggressive 1.23 ** 1.05-1.43
 Victimized 0.98 0.92-1.04
Sociometric
Status (a)
 Sent 1.15 0.82-1.62
 Received 0.91 * 0.82-1.00
 Reciprocated 1.55 0.76-3.17
Ethnicity (b)
 Latino 1.32 0.79-2.20
 Asian 1.44 0.85-2.45

Note. Underlined values represent odds ratios that remained significant
after Bonferroni adjustment.

(a) Sociometric status indicators derived from friendship nominations
sent, received, and reciprocated.

(b) Reference group = other.

*p < .05, **p < .006, ***p < .0001.

REFERENCES

Acre, R. (2001). Study: Kids rate bullying and teasing as "big problem." Retrieved August 28, 2001, from www9.cnn.com/2001/US/03/08 violencesurvey/index.html.

Alexander, C., Piazza, M., Mekos, D., & Valente, T. (2001). Peers, schools, and adolescent cigarette smoking. Journal of Adolescent Health, 29, 22-30.

Andreou, E. (2000). Bully/victim problems and their association with psychological constructs in 8- to 12-year-old Greek schoolchildren. Aggressive Behavior, 26, 49-56.

Arbona, C., Jackson, R. H., McCoy, A., & Blakely C. (1999). Ethic identity as a predictor of attitudes of adolescents toward fighting. Journal of Early Adolescence, 19, 323-340.

Baldry, A., & Farrington, D. (1999). Types of bullying among Italian school children. Journal of Adolescence, 22, 423-426.

Bandura, A. (2002). Social cognitive theory in cultural context. Applied Psychology: An International Review, 51, 269-290.

Bauman, K. E., & Ennett, S. T. (1996). On the importance of peer influence for adolescent drug use: Commonly neglected considerations. Addiction, 91, 185-198.

Bjoerkqvist, K., Oesterman, K., Lagerspetz, K. M. J., Landau, S. F., Caprara, G. V., & Fraczek, A. (2001). Aggression, victimization, and sociometric status: Findings from Finland, Israel, Italy, and Poland. In J. M. Ramirez & D. S. Richardson (Eds.), Cross-cultural approaches to research on aggression and reconciliation (pp. 111-119). New York: Nova Science Publishers.

Boulton, M. J. (1995). Parents of bully/victim problems in mixed race groups of children. Social Development, 4, 277-293.

Boulton, M. J., Trueman, M., Chau, C., Whitehand, C., & Amatya, K. (1999). Concurrent and longitudinal links between friendship peer victimization: Implications for befriending interventions. Journal of Adolescence, 22, 461-466.

Brown, B. B., Dolcini, M. M., & Leventhal, A. (1997). Transformations in peer relationships at adolescence: Implications for health-related behavior. In J. Schulenberg, J. L. Maggs, & K. Hurrelmann (Eds.), Health risks and developmental transitions during adolescence. New York: Cambridge University Press.

Centers for Disease Control and Prevention. (2001). YRBSS: Results. Youth 2001 on line. Retrived January 16, 2003, from www.cdc.gov/nccdphp/dash/yrbs/2001/youtho1online.htm

Collins, K., & Bell, R. (1996). Peer perceptions of aggression and bullying behavior in primary schools in Northern Ireland. Annals of the New York Academy of Sciences, 794, 336-338.

Connolly, J., Pepler, D., Craig, W., & Taradash, A. (2000). Dating experiences of bullies in early adolescence. Child Maltreatment, 5, 299-310.

Cowie, H. (2000). Bystanding or standing by: Gender issues in coping with bullying in English schools. Aggressive Behavior, 26, 85-97.

Crick, N., Casas, J., & Ku, H. (1999). Relational and physical forms of peer victimization in preschool. Developmental Psychology, 35, 376-385.

Ennett, S. T., Bailey, S. L., & Feldman, E. B. (1999). Social network characteristics associated with risky behaviors among runaway and homeless youth. Journal of Health and Social Behavior, 40, 63-78.

Ennett, S. T., & Bauman, K. E. (1994). The contribution of influence and selection to adolescent peer group homogeneity: The case of adolescent cigarette smoking. Journal of Personality and Social Psychology, 67, 653-663.

Gest, S. D., Graham-Bermann, S. A., & Hartup, W. W. (2001). Peer experience: Common and unique features of number of friendshipos, social network centrality, and sociometric status. Social Development, 10, 23-40.

Graham, S., & Juvonen, J. (1998). Self-blame and peer victimization in middle school: An attributional analysis. Developmental Psychology, 34, 587-599.

Graham, S., & Juvonen, J. (2002). Ethnicity, peer harassment, and adjustment in middle school: An exploratory study. Journal of Early Adolescence, 22, 173-199.

Hanish, L. D., & Guerra, N. G. (2000). The roles of ethnicity and school context in predicting children's victimization by peers. American Journal of Community Psychology, 28, 201-223.

Harris, J. R. (1995). Where is the child's environment? A group socialization theory of development. Psychological Review, 102, 458-489.

Hartup, W. W., & Sancillo, M. F. (1986). Children's friendships. In E. Schopler & G. B. Mesibov (Eds.), Social behavior in autism (pp. 61-80). New York: Plenum.

Haselager, G. J., Hartup, W. W., Van Lieshout, C. F., & Riksen-Walraven, J. M. (1998). Similarities between friends and nonfriends in middle childhood. Child Development, 69, 1198-1208.

Hawley, P. H. (1999). The ontogenesis of social dominance: A strategy-based evolutionary perspective. Develomental Review, 19, 97-132.

Hodges, E. V., Boivin, M., Vitaro, F., & Bukowski, W. M. (1999). The power of friendship: Protection against an escalating cycle of peer victimization. Developmental Psychology, 35, 94-101.

Hodges, E. V., Malone, M. J., & Perry, D. G. (1997). Individual risk and social risk as interacting determinants of victimization in the peer group. Developmental Psychology, 33, 1032-1039.

Huttunen, A., Salmivalli, C., & Lagerspetz, K. (1996). Friendship networks and bullying in schools. Annals of the New York Academy of Sciences, 794, 355-359.

Kaltiala-Heino, R., Rimpela, M., & Rimpela, R. (2000). Bullying at school--an indicator of adolescents at risk for mental disorders. Journal of Adolescence, 23, 661-674.

Karatzias, A., Power, K. G., & Swanson, V. (2002). Bullying and victimization in secondary schools: Same or separate entities? Aggressive Behavior, 28, 45-61.

Kaukiainen, A., Salmivalli, C., Lagerspetz, K., Tamminen, M., Vauras, M., Maeki, H., & Poskiparta, E. (2002). Learning difficulties, social intelligence and self-concept: Connections to bully-victim problems. Scandinavian Journal of Psychology, 43, 269-278.

Kochenderfer, B. J., & Ladd, G. W. (1996). Peer victimization: Cause or consequence of school maladjustment? Child Development, 67, 1305-1317.

Kochenderfer, B. J., & Ladd, G. W. (1997). Victimized children's responses to peers' aggression: Behaviors associated with reduced versus continued victimization. Development & Psychopathology, 9, 59-73.

Kumpulainen, K., & Rasanen, E. (2000). Children involved in bullying at elementary school age: Their psychiatric symptoms and deviance in adolescence. Child Abuse and Neglect, 24, 1567-1577.

Kumpulainen, K., Rasanen, E., & Puura, K. (2002). Psychiatric disorders and the use of mental health services among children involved in bullying. Aggressive Behavior, 27, 102-110.

Kupersmidt, J. B., DeRosier, M. E., & Patterson, C. P. (1995). Similarity as the basis for children's friendships: The roles of sociometric status, aggressive and withdrawn behavior, academic achievement and demographic characteristics. Journal of Social and Personal Relationships, 12, 439-452.

Ladd, G. W., Kochenderfer, B. J., & Coleman, C. C. (1996). Friendship quality as a predictor of young children's early school adjustment. Child Development, 67, 1103-1108.

Moran, S., Smith, P. K., Thompson, D., & Whitney, L. (1993). Ethnic differences in experiences of bullying: Asian and White children. British Journal of Educational Psychology, 63, 431-440.

Naylor, P., & Cowie, H. (1999). The effectiveness of peer support systems in challenging school bullying: The perspectives and experiences of teachers and pupils. Journal of Adolescence, 22, 467-479.

Olweus, W. (1991). Bully/victim problems among schoolchildren: Basic facts and effects of a school based intervention program. In D. Pepler & K. Rubin (Eds.), The development and treatment of childhood aggression (pp. 411-448). Hillsdale, NJ: Erlbaum.

O'Moore, M., & Kirkham, C. (2001). Self-esteem and its relationship to bullying behavior. Aggressive Behavior, 27, 269-283.

Pakaslahti,. (2001). Peer-attributed prosocial behavior among aggressive/preferred, aggressive/non-preferred, non-aggressive/preferred and non-preferred adolescents. Personality & Individual Differences, 30, 903-916.

Pakaslahti, L., & Keltikangas-Jarvinen, L. (2000). Comparison of peer, teacher, and self-assessments on adolescent direct and indirect aggression. Educational Psychology, 20, 177-190.

Pelligrini, A. D. (1998). Bullies and victims in school. A review and call for research. Journal of Applied Developmental Psychology, 19, 165-176.

Pellegrini, A. D., & Smith, P. K. (1998). Physical activity play: The nature and function of a neglected aspect of play. Child Development, 69, 577-598.

Poulin, F., Cillessen, A. H. N., Hubbard, J. A., Coie, J. D., et al. (1997). Children's friends and behavioral similarity in two social contexts. Social Development, 6, 224-236.

Rivers, I., & Smith, P. K. (1994). Types of bullying behavior and their correlates. Aggressive Behavior, 20, 359-368.

Rodkin, P. C., Farmer, T. W., Pearl, R., & Van Acker, R. (2000). Heterogeneity of popular boys: Antisocial and prosocial configurations. Developmental Psychology, 36, 14-24.

Rowe, D. C., Almeida, D. M., & Jacobson, K. C. (1999). School context and genetic influences on aggression in adolescence. Psychological Science, 10, 277-280.

Salmivalli, C., Karhunen, J., & Lagerspetz, K. M. J. (1996). How do victims respond to bullying? Aggressive Behavior, 22, 99-109.

Salmivalli, C., Kaukiainen, A., & Lagerspetz, K (2000). Aggression and sociometric status among peers: Do gender and type of aggression matter? Scandinavian Journal of Psychology, 41, 17-24.

Salmivalli, C., Lappalainen, M., & Lagerspetz, K. (1998). Stability and change of behavior in connection with bullying in schools: A two-year follow-up. Aggressive Behavior, 24, 205-218.

Salmon, G., & West, A. (2000). Physical and mental health issues related to bullying in schools. Current Opinion in Psychiatry, 13, 375-380.

Schwartz, D. (2000). Subtypes of victims and aggressors in children's peer groups. Journal of Abnormal Child Psychology, 28, 181-192.

Schartz, D., Farver, J. M., Chang, L., & Lee-Shin, Y. (2002). Victimization in South Korean children's peer groups. Journal of Abnormal Child Psychology, 30, 113-128.

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. 147-174). New York: Guilford Press.

Scott, J. (2000). Social network analysis: A handbook (2nd ed.). London: Sage Publications.

Smith, P., & Brain, P. (2000). Bullying in schools: Lessons from two decades of research. Aggressive Behavior, 26, 1-9.

Sutton, J., Smith, P., & Swettenham, J. (1999). Social cognition and bullying: Social inadequacy or skilled manipulation? British Journal of Developmental Psychology, 17, 435-450.

Tremblay, R. E., Masse, L. C., Vitaro, F., & Dobkin, P. L. (1995). The impact of friends' deviant behavior on early onset of delinquency: Longitudinal data from 6 to 13 years of age. Development and Psychopathology, 7, 649-667.

Twemlow, S. W., Fonagy, P., & Sacco, F. C. (2001). An innovative psychodynamically influenced approach to reduce school violence. American Journal of Child and Adolescent Psychiatry, 40, 377-376.

Twemlow, S. W., Sacco, F. C., & Williams, P. (1996). A clinical and interactionist perspective on the bully-victim-bystander relationship. Bulletin of the Menninger Clinic, 60, 296-313.

Valente, T. W. (1995). Network models of the diffusion of innovations. Cresskill, NJ: Hampton Press.

Valente, T. W., & Vlahov, D. (2001). Selective risk taking among needle exchange participants: Implications for supplemental interventions. American Journal of Public Health, 91, 406-411.

Vitaro, F., Tremblay, R. E., Kerr, M., & Bukowski, W. M. (1997). Disruptiveness, friends' characteristics, and delinquency in early adolescence: A test of two competing models of development. Child Development, 68, 676-689.

Warman, D. M., & Cohen, R. (2000). Stability of aggressive behaviors and children's peer relationshps. Aggressive Behavior, 26, 277-290. Wasserman, S., & Faust, K. (1994). Social networks analysis: Methods and applications. Cambridge, UK: Cambridge University Press.

Xie, H., Cairns, R. B., & Cairns, B. D. (1999). Social networks and configurations in inner-city schools: Aggression, popularity, and implications for students with EBD. Journal of Emotional and Behavior Disorders, 7, 147-155.

Youniss, J., & Smollar, J. (1989). Adolescents' interpersonal relationships in social context. In T. J. Berndt & G. W. Ladd (Eds.), Peer relationships in child development (pp. 300-316. Oxford, England: John Wiley & Sons.

Tom Valente, Peggy Gallaher, Louise Ann Rohrbach, and Jennifer B. Unger, University of Southern California, Department of Preventive Medicine. Requests for reprints should be sent to Michele Mouttapa, PhD., Morgan State University Center for Health Disparities Solutions, 1700 E. Cold Spring Lane, Montebello Complex, Room 103-D, Baltimore, MD 21251.