Close
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Social network and group self-identification predictors of school violence
(USC Thesis Other)
Social network and group self-identification predictors of school violence
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
SOCIAL NETWORK AND GROUP SELF-IDENTICATION PREDICTORS
OF SCHOOL VIOLENCE
Copyright 2004
by
Michele Mouttapa
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
PREVENTIVE MEDICINE
August 2004
Michele Mouttapa
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 3145250
INFORMATION TO USERS
The quality of this reproduction is dependent upon the quality of the copy
submitted. Broken or indistinct print, colored or poor quality illustrations and
photographs, print bleed-through, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.
®
UMI
UMI Microform 3145250
Copyright 2004 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Dedication
To Papa, Mama, Peter, and Andre Mouttapa, with much love. To Brian, with
a wonderful future ahead of us. And to God, who has taken me on this journey.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Acknowledgements
There are a countless number of people to thank for their support over the
years, and here are only a few of them. I would like to thank my advisor, Dr.
Jennifer Unger, for her copious amounts of support, advice, and feedback at all
times, any time, with great encouragement. I would also like to thank Dr. Tom
Valente and Dr. Luanne Rohrbach for providing me with the opportunity to work
outside of academia and providing excellent guidance all along. Dr. Steve Sussman
has provided excellent advice at a moment’s notice with enthusiasm and humor, and
I greatly appreciate that. I also greatly appreciate Dr. Ron Avi Astor’s commitment
to being the outside member of my dissertation committee even while overseas.
I would also like to thank Drs. Sohaila Shakib and Jie Weiss for their
friendship and experiences in writing papers. I learned so much through the process.
A special thanks to all of the staff and students at IPR whose friendly faces and acts
of kindness have made this journey worthwhile.
Last, but not least, I would like to like my parents for their countless prayers
and eager support, to Laura for countless phone calls and great laughs, and Brian for
his never-ending love.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
iv
Table of Contents
Dedication ii
Acknowledgements iii
List of Figures and Tables vi
Abstract vii
CHAPTER I: INTRODUCTION 1
What is violence? 1
School violence: Prevalence 3
The importance of studying adolescent school violence 5
General characteristics of perpetrators, victims, and aggressive victims 6
The importance of friends during adolescence 10
Dominance theory and social cognitive theory 11
Social network analysis 12
Social network characteristics of perpetrators and victims 14
Insights gained through social network analysis 21
Peer group self-identification 23
Peer groups and violence: Theoretical models 27
Peer groups and violence: Previous findings 29
Limitations of previous studies 32
This dissertation 35
CHAPTER II: METHODS 41
Sample 41
Procedure 43
Measures 43
CHAPTER III: STUDY 1 51
Predictions 51
Sample 53
Analyses 53
Results 54
Conclusions 59
CHAPTER IV: STUDY 2 65
Predictions 66
Sample 67
Analyses 68
Results 69
Conclusions 78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
V
CHAPTER V: STUDY 3 86
Predictions 87
Sample 88
Analyses 89
Results 90
Conclusions 96
CHAPTER VI: DISCUSSION 103
General implications 104
Recommendations for everyday school practices 106
Limitations 107
Future directions 112
REFERENCES 115
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
vi
List of Figures and Tables
Figure 1: Conceptual Model 36
Table 1: Gender Differences on Network Characteristics and 55
Outcome Variables
Table 2: Logistic Regression of Perpetrator Status 56
Table 3: Logistic Regression of Victim Status 57
Table 4: Logistic Regression of Aggressive Victim Status 59
Table 5: Factor Loadings of Peer Group Self-Identification Items 71
Table 6: Differences Between 7th Grade High-Risk and Low-Risk Students 73
Table 7: Sixth Grade Characteristics by 7th Grade Group Identification 74
Table 8: Logistic Regression of High-Risk Peer Group Self-Identification 75
Table 9: Peer Group Self-Identification and Friendship Nominations 77
Table 10: Proportion of High-Risk Friends by Peer Group 78
Table 11: Differences among Physically and Verbally Violent Subgroups 92
Table 12: Logistic Regression of Physical and Verbal Violence 93
Table 13: Logistic Regression of 8th Grade Rumor Spreading 95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
vii
Abstract
School violence is prevalent, and its existence has been recognized in the
United States as well as in other countries. School violence is most prevalent during
adolescence, especially during the transition into middle school. The three studies of
this dissertation examined the associations of social network characteristics and
high-risk peer group self-identification and violence among middle school
adolescents who attended 24 schools throughout Southern California. The students
were surveyed during the 6th , 7th , and 8th grades. A total of 2,292 students completed
the survey during all three assessments.
In a cross-sectional design, Study 1 examined whether 6th grade violent
perpetrators, victims, and aggressive victims (those who are perpetrators and
victims) differed from other students on the following friendship network
characteristics: nominations sent, nominations received, friendship reciprocity, level
of friends’ violent perpetration, and level of friends’ victimization. Friends’ level of
perpetration was positively associated with one’s own perpetrator (p < 0.0001) and
aggressive victim (p < 0.01) status, and negatively associated with victim status (p <
0.01).
In a longitudinal design, Study 2 examined whether 6th grade perpetrator,
victim, and aggressive victim status, as well as social network characteristics, were
associated with 7th grade high-risk peer group self-identification. Aggressive victim
status was associated with high-risk peer group self-identification (p < 0.001).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
viii
Furthermore, high-risk students had a higher proportion of high-risk friends relative
to low-risk friends (p < 0.0001). Self-reported “ jocks” and “popular students”
received the largest number of friendship nominations.
In a longitudinal design, Study 3 examined whether 6th grade social network
characteristics and 7th grade peer group self-identification were associated with 8th
grade perpetrator, victim, and aggressive victim status. Friends’ perpetration level
was positively associated with one’s own perpetrator status (p < 0.05). The number
of friendship nominations received was negatively associated with victim status (p <
0.05). Last, aggressive victim status was positively associated with the number of
nominations received (p < 0.05) and high-risk peer group self-identification (p <
0.05).
The findings suggest that perpetrators, victims, and aggressive victims have
unique correlates, antecedent variables, and outcome variables associated with their
status. Implications are discussed.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1
Chapter I: Introduction
School violence is a multifaceted construct that includes both criminal
behaviors and non-criminal, aggressive behaviors that occur in school (Furlong &
Morrison, 2000). Criminal behaviors include homicide, weapon possession in
school, and physical fighting, to name a few. Aggressive behaviors consist of
actions such as verbal threats, teasing, milder forms of physical violence (e.g., less
harmful hitting or shoving), and antisocial behavior in general (See Furlong &
Morrison, 2000; Zeira, Astor, & Benbenishty, 2003). The commonality among all of
these behaviors is that they often cause physical and/or psychological harm to fellow
students and disrupt the learning environment at school (Furlong & Morrison, 2000).
What is violence?
The term “violence” is used frequently in everyday life and in many different
contexts. Therefore it is important in research to define violence as clearly as
possible and to articulate which types of violence are being studied. The Panel on
the Understanding and Causes of Violent Behavior has defined violence as “behavior
by persons that intentionally threatens, attempts, or actually inflicts physical harm”
(Reiss & Roth, 1993). The behaviors included in this definition are mostly included
in definitions of aggression.
Similarly Aronson (1980) defined aggression as “a behavior aimed at causing
harm or pain” (Aronson, 1980). As such, aggression is separate from competitive
behavior that is not intended to do harm. For example the act of studying diligently
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2
with the intention of receiving the highest grades in the class is not an act of
aggression. However stealing another student’s notes with the intention of having
him or her performing poorly on a test is an act of aggression. Secondly, the
negative impact that a behavior has on a recipient does not determine whether that
behavior is aggression. For example an ineffective attempt to beat up a classmate is
aggression since there is harmful intent, despite that no obvious harm was done. On
the other hand, an accidental shove may be very painful, and may even leave bruises,
but is not aggression since there is no harmful intent.
As one can see, definitions of violence and aggression have considerable
overlap in the literature. Since Furlong & Morrison (2000) suggest that aggression is
a subset of school violence, I will follow their example and use the term “violence”
throughout this dissertation to refer to all behaviors that are intended to cause harm.
Bullying: a subset of school violence. “Bullying” is another hot topic in
recent literature. Theoretically, school bullying is a special case of violence in which
one or more students physically and/or emotionally harm another student repeatedly
over time during school (Olweus, 1991). Some students are both aggressors and
victims (Twemlow, Fonagi, & Sacco, 2001). These students engage in violent
behaviors, and, to a similar extent, are victims of violence. Bullying, however,
implies that an imbalance of power exists, such that the victim has difficulty
defending him or herself from aggressors (Olweus, 1991). In this case, the aggressor
also has the distinction of being a “bully.”
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3
The researcher needs to know whether a power differential exists among
pairs of students who engage in violence with each other to determine whether
bullying has occurred. For example, one can examine whether Student A more
frequently victimizes Student B, or the opposite, or both are equally victimized by
each other. To date, however, relatively few studies of “bullying” have specifically
examined bullying since the majority of studies have not assessed the relational
characteristics of individual pairs of students. Hence, these studies fall into the more
general topic of school violence.
School violence: Prevalence
School violence is prevalent and is a salient concern among today’s youth
and social scientists. Findings from the 2003 Youth Risk Behavior Survey (CDC,
2003) suggest that approximately 1 in 3 high school students in the U.S. were
involved in at least one physical fight within the past year, and slightly over 1 in 8 on
school property. Six percent of students carried a weapon on school property within
the past month, and 9% reported being threatened or injured with a weapon on
school property within the past year. Nearly 1 in 3 reported that someone had stolen
or deliberately damaged their property at school within the past year. Such items
included cars, clothing, and books. Finally, over 5% reported missing one or more
days of school during the past month because they felt unsafe at school or on their
way to and from school.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4
Other studies have also examined the prevalence of aggressive victimization,
the extent to which students are both perpetrators and victims. In his review,
Pelligrini (1998) estimated that 7 to 15% of the school-aged population are bullies
(those who are perpetrators but not victims), 2 to 10% are passive victims (those who
are victims but not perpetrators), and 2 to 10% are aggressive victims (those who are
both). The prevalence of perpetration, victimization, and aggressive victimization in
a sample may vary depending on the source, as self-reports tend to yield lower rates
of perpetration and victimization relative to peer reports (Pakaslathi & Keltikangas-
Jarvinen, 2000).
Research has paralleled youth’s concerns about school violence. The number
of citations with the key words “school violence” appearing in the PSYCHINFO
electronic database has grown exponentially since the 1990s (Furlong & Morrison,
2000). In 2003 alone, 211 journal articles can be found under the search term
“school violence” in PSYCHINFO.
In this chapter I will review several contemporary studies (from 1995 until
the present) in which authors have attempted to assess and examine relatively
common, mild to moderately violent behaviors at school defined as “bullying” and
“aggression.” Hence I did not include those studies that examined extreme acts of
violence such as use of firearms and homicide in this review.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5
The importance of studying adolescent school violence
School violence in the form of physically and verbally aggressive behaviors
begins as early as preschool and is most prevalent during the middle school years
(Twemlow, Fonagy, & Sacoo, 2001). Such behaviors are precursors to serious
violence during adulthood (Pakiz, Reinherz, & Giaconia, 1997). Contrary to
common beliefs, previous research suggests that violence does not have a cathartic
effect. In other words violence does not reduce tensions that would otherwise lead to
future violence. Rather, violence breeds more violence, greater dislike for victims,
and more justification for violent actions (Aronson, 1980).
Although violent tendencies may be instinctive among humans, it is believed
that the situational context largely determines whether violence actually occurs
(Aronson, 1980). Hence, examining risk and protective factors of school violence
during preadolescence and adolescence is of great importance. Furthermore,
violence among middle school students may be more effectively prevented and/or
reduced by school-level interventions relative to high school students, who may have
more contact with older, more violent peers (Furlong & Morrison, 2000).
Several studies have examined associations between violent perpetration,
victimization, and individual characteristics, for example, race/ethnicity, age,
depression, hostility and anxiety (Espelage and Swearer, 2003). However, it has
been argued that violence is the result of interactions between individual
characteristics and the social context (Espelage, 2002). In this chapter I will discuss
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6
both individual and relational characteristics that have been associated with violent
perpetration and victimization.
General characteristics of perpetrators, victims, and aggressive victims
Perpetrators, victims, and aggressive victims of school violence have unique
patterns of correlates, antecedent variables and consequences associated with their
behaviors. This section briefly summarizes findings in the literature.
Gender differences. There is plenty of evidence suggesting that violent
perpetration differs between girls and boys. Compared to girls, boys are more often
involved in physical forms of aggression (e.g., kicking, pushing), whereas females
are more often involved in relational forms of aggression (e.g., rumor spreading,
social isolation; Crick, Casas, & Ku, 1999; Baldry & Farrington, 1999; Rivers &
Smith, 1994). Relational forms of aggression are associated with social acceptance
among males (Salmivalli, Kaukiainen, & Lagerspetz, 2000), whereas females more
often defend victims from their aggressors (Salmivalli, Lappalainen, & Lagerspetz,
1998).
Ethnic differences. 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; Boulton, 1995). These findings held true even
when Whites were not an ethnic minority (e.g., primarily Latino schools, primarily
African-American schools). Rodkin et al. (2000) found that “tough boys,”
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7
aggressive students with high centrality scores, were disproportionately African
American. These findings may be confounded with other variables such as
socioeconomic status, exposure to violence in the community and in the media, and
the like.
Characteristics of perpetrators. Perpetrators of violence are often referred to
as “bullies” in the literature. Bullies 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). Although bullies may be
disliked among several peers (Warman & Cohen, 2000; Prinstein and Cillessen,
2003), they are also well liked among other aggressive students because of their
“savoir faire”, sophisticated interpersonal skills used in dominating others (See
Rodkin et al., 2000). Bullies have higher peer-nominated scores on sociability and
leadership (Collins & Bell, 1996), have earlier dating experiences (Connolly, Pepler,
Craig, & Taradash, 2000), and are more physically abusive towards their
boyfriends/girlfriends (Connolly et al., 2000) relative to other students. High school
students report that they and other students engage in bullying in order to create
excitement, gain attention, and gain social acceptance (Owens, Shute, & Slee, 2000).
Unlike aggressive victims, who will be described later, bullies do have the ability to
regulate their emotions (Schwartz, 2000).
Characteristics of victims. Those who report being a victim have higher rates
of intrapersonal problems including self-blame, loneliness, anxiety, low self-worth
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8
(Graham & Juvonen, 1998), and, to the greatest extent, depression (Hawker &
Boulton, 1999). There is evidence that such characteristics are both the cause (Bond,
Carlin, Thomas, Rubin, & Patton, 2001; Kochenderfer & Ladd, 1996; Hodges &
Perry, 1999) and consequence (Hodges & Perry, 1999) of victimization.
Victims miss more days of school and report more physical health complaints
(Salmon & West, 2000). Victims tend to remain victims even when they change
schools (Salmivalli, Lappalainen, & Lagerspetz, 1998) or interact with groups of
previously unacquainted peers (Dodge, 1983). Adolescents believe that girl victims
provoke bullies with their helplessness, whereas boy victims provoke bullies with
their unsuccessful attempts to fight back (Salmivalli & Lagerspetz, 1996). It has
been demonstrated that a dose-response relationship exists between the duration of
victimization and the amount of stress experienced by the victim (Sharp, Thompson,
& Arora, 2000). Generally victims are disliked among their peers (Kochenderfer-
Ladd, 2003; Prinstein and Cillessen, 2003).
Characteristics of aggressive victims. Some studies have examined
aggressive victims, a subset of perpetrators and victims. Aggressive victims, or
“bully-victims” (e.g., Andreou, 2000) are those who engage in aggressive behaviors
and are also victims of aggression. Aggressive victims have been described as
reactive aggressors (reacting to being bullied), whereas bullies have been described
as proactive aggressors (initiating bullying; Roland & Idsoe, 2000). Aggressive
victims are also characterized by their poor emotional regulation (Schwartz, 2000),
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9
high reactivity when provoked (Schwartz, 2000), and hostility (Evans, Heriot, &
Friedman, 2002). Such characteristics may be the result of a hostile home
environment, and can lead to rash displays of aggression and subsequent peer
retaliation (Schwartz, Dodge, Pettit, & Bates, 1997). Aggressive victims tend to
have particularly high levels of anxiety and depression (Kumpulainen & Rasanen,
2000; Kumpulainen, Rasanen, & Puura, 2002). Aggressive victims are also
generally disliked by their peers (Andreou, 2000).
Longitudinal findings. Longitudinal evidence suggests that higher levels of
impulsivity, anger, and depression leads to increased aggression over time (Espelage,
Bosworth, & Simon, 2001). The experience of being victimized may lead one
become increasingly anxious and depressed even into adulthood (Roth, Coles, &
Heimberg, 2002). Interpersonal problems such as low self-esteem may cause one to
be the target of bullying, and the experience of being victimized may lead to lower
self-esteem (Hodges & Perry, 1999).
Violence and the school climate. A positive school climate (Ma, 2002),
namely, the presence of disciplinary actions, strong parental involvement, and high
academic standards in schools, is associated with lower rates of aggression after
adjusting for confounders such as gender, socioeconomic status, and martial status of
parents. Furthermore, longitudinal evidence suggests that a positive school climate
is associated with lower rates of aggression seven years later in young adulthood
(Kasen, Cohen, & Brook, 1998). On the other hand, violence that goes unchecked
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
by teachers and school staff may disrupt the school climate. Dupper and Meyer-
Adams (2002) suggest that low-level violence may anger and alienate students,
thereby contributing to a hostile school environment and lower academic
performance (Dupper & Meyer-Adams, 2002). Anger and hostility among students
may lead to higher rates of violence, which may then lead to the decreased ability for
teachers to establish order in the classroom and teach their classes effectively. Over
time, students may learn to engage in violent behaviors to achieve goals, as will be
discussed later in this chapter.
The importance of friends during adolescence.
Peer relations are a salient feature in the lives of adolescents, as they have
increased opportunities to socialize with each other outside of adult supervision
compared to earlier years. Although the number of social contacts increases during
adolescence (Pellegrini & Bartini, 2001), close friends become increasingly similar
in their values, preferences, and behaviors (Cairns, Neckerman, and Cairns, 1989).
This may be the result of peer influence or the selection of like-minded peers.
Hence, an adolescent’s friendship group may indicate whether or not he or she
engages in violence, or is victimized by other peers. Both dominance theory
(Hawley, 1999) and social cognitive theory (Bandura, 2002) can illustrate how the
peer context is associated with school violence.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11
Dominance theory and social cognitive theory
Dominance theory (Hawley, 1999) explains why violence occurs among
classmates, whereas social cognitive theory (Bandura, 2002) explains how peers may
learn violent behaviors. Hence, the two theories are not mutually exclusive.
Dominance theory. Dominance theory posits that a status hierarchy exists
among classmates, and one moves up this hierarchy through the use of violent
tactics. Pellegrini and Bartini (2001) suggest that as boys move into adolescence,
physical aggression is increasingly used to establish dominance among other boys.
Among girls, Owens, Shute, and Slee (2000) found that gaining attention and social
acceptance among classmates was a primary reason why 15-year old Australian girls
engaged in rumor spreading, a form of relational violence. Salmivalli and
Lagerspetz (1996) suggest that helplessness among girls provokes other students to
victimize them.
Social cognitive theory. Social cognitive theory (Bandura, 2002) posits that
violent behaviors are first learned through vicarious reinforcement. For example,
adolescents may observe a classmate gain access to material resources (e.g., lunch
money) and social resources (e.g., attention, praise) by bullying another student. If
these observers believe that they have the skills to do so, they may model the peer’s
violence in hopes of acquiring such rewards. Since violence among students often
occurs in the presence of other classmates (Hawkins, Pepler, & Craig, 2001), a social
cognitive approach towards violence is highly feasible.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
Social network analysis, the methods and techniques for analyzing social
relationships, for example, friendships (Scott, 2000; Wasserman & Faust, 1994), can
be utilized to test social dominance and social cognitive models of school violence.
Social network analysis
Social network analysis is a set of techniques used to analyze social
relationships (Scott, 2000; Wasserman & Faust, 1994). By asking each student to
list the names of his or her friends, the network analyst can map out the social
network, the constellation of all social ties within a group, for example, a classroom
of students (Scott, 2000). Several characteristics of one’s friendships and his or her
location in the social network can be assessed. Such network characteristics may
include reciprocity, centrality, and the behaviors of one’s nominated friends, all
which have been examined in previous studies of school violence.
Reciprocity is the degree that the students a person names as friends also
name him or her as a friend (e.g., Valente, Gallaher, & Mouttapa, in press; Pellegrini,
Bartini, & Brooks, 1999). Reciprocity can be defined as the number of reciprocal
friends one has, or the proportion of reciprocal friendships one has from the total
number of friends he or she nominated. Centrality is the degree to which a person
occupies a central position in the network (Scott, 2000). There are several ways to
measure centrality. One measure is indegree, a count of the number of times a
person is nominated among their classmates in response to the question, “Who are
your best friends in the classroom?” (See Valente, Gallaher, & Mouttapa, in press).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13
Those who are directly and indirectly connected to many other classmates would be
considered “central” in the friendship network, whereas those who are connected to
relatively few classmates would be considered “peripheral.”
By linking friendship nominations to the nominees’ individual data, the
network analyst can also determine the extent to which an individual’s friends report
that they themselves are perpetrators or victims of violence. Hence, the researcher
can assess variables such as the mean level of violent perpetration (e.g., Espelage,
Holt, and Henkel, 2003) and victimization (Browning, Cohen, & Warman, 2003)
among one’s nominated friends.
Testing theoretical models of violence. What does friendship reciprocity and
centrality have to do with school violence? Dominance theory (Hawley, 1999)
suggests that some students engage in violence in order to acquire social status
among classmates. Hence, violence may be positively associated with high network
centrality. Dominance theory also suggests that a social hierarchy exists among
classmates such that victims are those students who are less assertive, socially
isolated, and less capable of defending themselves from potential perpetrators.
Hence, victimization may be positively associated with low network centrality and
low friendship reciprocity.
Social cognitive theory (Bandura, 2002) suggests that individuals may model
the violence of friends, and friends may assist in and reinforce each other’s violent
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
14
behaviors. Hence, violence may be positively associated with the level of violence
among one’s friends.
Social network characteristics of perpetrators and victims
Numerous studies have identified friendship network characteristics that are
associated with violent perpetration and victimization. The following network
characteristics will be discussed in this chapter: friends’ violent behaviors, friends’
victimization, friendship reciprocity, and peer network centrality.
Friends’ violent behaviors: Cross-sectional findings. Social cognitive theory
(Bandura, 2002) suggests that groups of friends engage in similar amounts of
violence because they model and reinforce each other’s violent behaviors. Students
may also initially select each other as friends based on similarities of violent
behavior. Cross-sectional analyses indicate that friends are most similar to each
other in their frequencies of fighting (Haselager, Hartup, van Lieshout, & Riksen-
Walraven, 1998; Kupersmidt, DeRosier, & Patterson, 1995), bullying (Haselager et
al., 1998; Pellegrini, Bartini, & Brooks, 1999), aggressive behavior (Tremblay,
Masse, Vitaro, & Dobkin, 1995; Poulin & Boivin, 2000), and disruptive behavior
(Haselager et al., 1998). Espelage and Holt (2001) found that 75% of bullies
nominated at least one friend who was a bully, whereas non-bullies were much less
likely to nominate bullies as friends.
Friends tend to carry out similar roles in violent incidents (e.g., assisting the
bully, defending the victim, being an outsider, etc.; Salmivalli, Lappalainen, &
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
15
Lagerspetz, 1998). Friends may also play complimentary roles in violent incidents.
Salmivalli, Huttunen, and Lagerspetz (1997) identified three general clusters of
friendship groups: (1) bullies, their assistants, and their reinforcers, (2) victims and
defenders of victims, and (3) outsiders, those who are not involved in bullying.
Although friends tend to engage in similar levels of violence, violent students
do not exclusively associate with each other. Farmer, Leung, Rodkin, Cadwallader,
Pearl, & Van Acker (2002) noted that in their sample, two thirds of aggressive 4th -6th
grade boys and one third of aggressive girls were in non-aggressive to moderately
aggressive friendship groups.
Friends’ violent behaviors: Longitudinal findings. Longitudinal studies
suggest that friends’ violent perpetration predicts subsequent individual perpetration.
Espelage, Holt, and Henkel (2003) identified peer cliques among graders,
clusters of students who report having more contact with each other than with other
students. They found that the bullying behaviors of one’s peer clique largely
determined his or her own bullying behaviors one year later. However, Tremblay et
al. (1995), in their study of 11-13 year-old boys, found no association between the
aggressive behaviors of reciprocal friends, classmates who nominated each other as
friends, and individual aggression one year later. Such findings may indicate that the
influence of the entire peer clique is stronger than that of individual friends, or that
females are more strongly influenced by their friends’ violence relative to males.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
16
The stability of friendships with violent classmates also plays an important
role in the continuance of violence. Specifically, lasting friendships with aggressive
students has been associated with stable or increased amounts of aggression over
time. Furthermore, discontinued relationships with aggressive students, and/or
lasting friendships with non-aggressive students, have been associated with
decreased aggression (Bemdt, Hawkins, & Jiao, 1999; Warman & Cohen, 2000).
Highly aggressive students often lose reciprocal friends, whereas moderately
aggressive students often gain reciprocal friends over time. Hence, moderately
aggressive students may have a strong influence on the aggressive behaviors of their
friends.
Friends’ violent behaviors: Influence or selection? There is evidence that
similarities in aggressive behaviors precede friendship formation rather than the
reverse. Among 4th -6th grade boys, Poulin and Boivin (2000) found that similarities
in aggressive behaviors predicted later friendship formation, whereas dissimilarities
in aggressive behaviors among friends predicted later friendship breakups. They did
not find increasing similarities in aggression among students who remained friends.
These findings support a friendship selection model, that violent students select each
other as friends because of their similar preferences for violence, rather than an
influence model, friends becoming increasingly similar over time. It is not known
whether Poulin and Boivin’s (2000) findings generalize to female adolescents as well
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
17
as males. Bauman & Ennett’s (1996) found similar temporal relationships between
adolescent friendships and drug use in their review of social network studies.
Friends’ victimization. Dominance theory (Hawley, 1999) suggests that
victimized students have low social status; thus they may aggregate in friendship
groups. Social cognitive theory (Bandura, 2002) suggests that victimized students do
not counterattack their aggressors because they do not believe that they have the
skills to do so. Neither do their friends. In both cases, victimization in the friendship
group should be associated with one’s own victimization.
Victims tend to have other victims as friends. These groups of victims
generally lack the strength and assertiveness to protect each other from potential
aggressors (Hodges, Malone, & Perry, 1997). Moreover, the presence of friends who
are not victimized may be protective against victimization. Browning, Cohen, and
Warman (2003) found that adolescents who discontinued their victim status were
friends with adolescents with lower victimization scores at baseline compared to
those children who maintained their victim status in the second year.
Friendship reciprocity: Cross-sectional associations. Even though friendships
between adolescents are often unstable over time, it is common for an adolescent to
have one or more reciprocal friends at any given time. Dominance theory (Hawley,
1999) suggests that the presence of reciprocal friends, versus the presence of no
friends or non-reciprocal friends, may serve as a protective function against potential
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
18
aggressors (See Lewis & Feiring, 1989). Hence, friendship reciprocity should be
associated with lower rates of victimization.
Cross-sectional studies suggest that victimization is negatively associated
with the number of reciprocal friends (Pellegrini, Bartini, & Brooks, 1999; Boulton,
Trueman, Chau, Whitehand, & Amatya, 1999; Hodges, Boivin, Vitaro, & Bukowski,
1999). Hodges, Malone, and Perry (1997) suggest that the presence of reciprocal
friends who are physically strong and assertive may protect students who are high on
internalizing behaviors (e.g., social withdrawal) and/or externalizing behaviors (e.g.,
disruptiveness) from potential aggressors.
Friendship reciprocity: Longitudinal associations. Longitudinal studies
suggest that the continuous presence of reciprocal friends leads to decreases in
victimization (Boulton et al., 1999; Hodges & Perry, 1999; Hodges et al., 1999).
There is also evidence that victimization also leads to decreases in friendship
reciprocity (Hodges & Perry, 1999). However, associations between friendship
reciprocity and victimization are attenuated when externalizing and internalizing
behaviors are controlled for (Hodges & Perry, 1999).
Friendship reciprocity: Gender differences. The association between
friendship reciprocity, violence, and victimization varies by gender. Rys and Bear
(1997) found that friendship reciprocity was associated with lower rates of physical
th
and verbal aggression among 6 grade boys but not among girls. Other findings
(Huttunen, Salmivalli, & Lagerspetz, 1996) suggest that boy defenders of victims
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
19
have a greater number of reciprocal friendships relative to all other boys, and girl
victims have substantially fewer reciprocal friendships relative to all other girls.
Peer network centrality. For students, centrality among classmates may be
associated with a greater control over scarce resources (e.g., attention, valued items)
and can be a manifestation of social dominance (See Gest, 2001). Dominance theory
(Hawley, 1999) suggests that violence is used to acquire desirable social positions
among classmates. Hence, dominance theory suggests that violence is associated
with high network centrality.
Peer network centrality: Cross-sectional associations. Violence has been
associated with high network centrality. Salmivalli, Huttunen, and Lagerspetz
(1997) had Finnish 6th graders identify their classmates’ social groups and also
identify each of their classmates for their general role in bullying (bullies, bullies’
assistants, bullies’ reinforcers, victims, defenders of victims, or outsiders).
Assistants and reinforcers had the largest friendship groups, whereas victims and
their defenders had the smallest friendship groups. Espelage and Holt (2001) found
that bullying was associated with popularity among 6th grade males.
Peer network centrality: Violence used to dominate. Students may employ
violent tactics to maintain desired social positions. Pellegrini and Bartini (2001)
found that teacher-reported dominance (e.g., leadership qualities, assertiveness) was
associated with aggression among boys at the beginning of the 6th grade. By the end
of the 6th grade, however, aggression decreased as positions of social dominance had
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20
been established. The findings of Roland & Idsoe (2000) suggest that boys engage
in bullying to establish dominance over their classmates, whereas girls engage in
bullying to establish affiliations with other classmates.
Peer network centrality: Violence among popular students. Prinstein and
Cillessen (2003) had high school students nominate the “most popular” and “least
popular” students in their class. Popularity nominations were positively associated
with aggression used to obtain a reward, for example, attention (proactive
aggression), rather than for revenge (reactive aggression). Furthermore, popularity
was positively associated with reputational victimization, classmates attempting to
destroy one’s social reputation, and negatively associated with other forms of
victimization (e.g., physical). Such findings suggest that popular students may also
be targets for some forms of victimization, possibly by other popular students.
Peer network centrality: Not all popular students are violent. Studies indicate
that both violent and non-violent students often occupy central network positions.
Farmer et al. (2003) identified two popular groups of adolescent boys: “tough boys”
and “model boys.” They found that tough boys were considered more aggressive
than model boys. Similarly Prinstein and Cillessen (2003) found that both high and
low levels of popularity were associated with aggression among high school
students. Bagwell, Coie, Terry, and Lochman (2000) suggest that peer-identified
aggressive 4th graders were most popular among groups of other aggressive students,
but not groups of non-aggressive students.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
21
Peer network centrality: Longitudinal associations. Prinstein and Cillessen
(2003) examined longitudinal associations between popularity and aggression among
high school students. Popularity predicted increases in aggression, but aggression
did not predict increases in popularity. Such findings are contrary to that of social
dominance theory (Hawley, 1999), which suggests that violence earns one’s
popularity. It is possible that use of violence does not boost social status once
positions of popularity are established prior to high school.
Tnsights gained through social network analysis
Research on violence has primarily focused on the individual characteristics
and developmental patterns of violent perpetrators. However, less has been done to
identify characteristics of the social context within school settings that trigger
violence (Furlong & Morrison, 2000). Examining friendships among classmates
through social network analysis has allowed researchers to examine the social
context in which school violence occurs.
First, social network analysis has provided unique insights into the social
relationships of violent perpetrators, often called “bullies.” Evidence suggests that
violent students tend to select each other as friends (Poulin &Boivin, 2000).
However, their friendship groups are not limited only to violent students; they often
have non-violent friends (Farmer et al., 2002). Those violent students who do have
lasting friendships with non-violent students tend to have lower rates of violence
over time (Bemdt, Hawkins, & Jiao, 1999; Warman & Cohen, 2000). Such findings
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
22
suggest that the non-violent friends of violent students may well serve as role models
and leaders in violence prevention programs. In school-based studies, peer-
nominated leaders have often been recruited to assist in program delivery, and have
been successful in reducing other problem behaviors such as substance use (Valente
et al., in press).
Second, social network analysis has provided insights into why violent
students may engage in violence. It has been suggested that middle school students
engage in violence to establish dominance among some students and affiliations
among others Roland & Idsoe (2000). By high school, violence is a strategy used
among some popular students to obtain a desired reward, for example, attention
(Prinstein and Cillessen, 2003). In their review Pellegrini and Bartini (2001) suggest
that at the beginning of middle school, violence may be a salient strategy to establish
status. Once dominance is established by the high school years, however, the
frequent use of violence is costly because one risks alienating themselves from
groups of friends that are becoming increasingly stable. Nevertheless the occasional
use of violence among dominant students in certain contexts (e.g., bullying a
physically weaker student, spreading rumors another popular student to damage their
reputation) may help demonstrate and maintain their dominance.
However, many popular students are not violent and may never have used
violence to attain popularity (Farmer et al., 2003). Such findings suggest that the
school’s encouragement of prosocial skills and the discouragement of violence
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
23
should occur prior to middle school. Prinz, Blechman, and Dumas (1994) found that
a school-based social skills training program for 1st through 3rd graders was
successful in increasing social skills and decreasing aggression six months later.
Third, social network analysis has provided unique insights into the social
relationships of victims. Victims tend to associate with each other (Hodges, Malone,
& Perry, 1997). However, those victims who do have non-victimized friends have
lower rates of victimization over time (Browning et al., 2003). Victims generally
belong to smaller social circles (Salmivalli et al., 1997) and have fewer reciprocal
friends (Pellegrini et al., 1999; Boulton et al., 1999; Hodges et al., 1999). The
findings suggest that developing a supportive network of classmates around victims
may help reduce victimization. Menesini, Codecasa, Benelli, and Cowie (2003) did
exactly this with Italian middle school students. They found that a befriending
intervention prevented the increase of bullying behaviors and positive attitudes
towards bullying by the end of the intervention relative to a control group.
In sum, the findings of social network studies on school violence have
provided valuable recommendations for specific strategies to reduce violence.
Peer group self-identification
The previous section suggests that violent perpetration and victimization
among adolescents is strongly associated with characteristics of their social network,
for example, their friends’ behaviors, friendship reciprocity, and centrality.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
24
Similarly, the “peer crowd” or “peer group” that adolescents belong to might also
have associations with violence.
What is a peer group? A peer group refers to a large group of students who
share similar characteristics and interests, but may or may not be friends (Brown,
1989). Furthermore peer groups may only be a self-image that students adopt as
opposed to actual friendship groups (Sussman, Unger, & Dent, 2004). Peer groups
are generally described with labels that depict stereotypical group characteristics,
including personality, activities, dress, music, values, language, and lifestyle (See
Michell, 1997; La Greca, Prinstein, & Fetter, 2001). The names of labels may differ
and new labels may also develop. For example, those students who wear baggy
clothing and listen to rap music have been called “whiggers” (Urberg et al., 2000)
and “rappers” (Sussman et al., 2000).
The stereotypes of different peer groups may vary depending on which group
the observer belongs to. For example, “preps”(students who are popular and heavily
involved in school activities) may be viewed by other groups as cliquish (for
example, “gangsters” and “nerds”), but they may see themselves as being open and
friendly (Brown, 1989). It is believed that peer groups may emerge to help students
cope with the problems associated with adolescent life including physical changes,
psychological changes, and transitioning to middle schools and high schools (Brown,
1996).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
25
Even though belonging to the same peer group does not automatically imply
friendship, peer groups may channel friendships among similar adolescents. The
group stereotype creates expectations not only of outward appearances but also of
attitudes, beliefs, and behaviors of individuals belonging to that group. Thus,
knowing which peer group a classmate belongs to may help an adolescent quickly
decide whether or not to pursue the classmate as a friend (Urberg et al., 2000).
Merten (1996) contends that adolescents seldom identify with one particular peer
category (as he calls it), but identify who they are by rejecting particular categories
(e.g., not being a “brain”). Nevertheless, mutual rejections for particular peer groups
may channel friendships as well.
Assessing peer groups. One way to assess adolescent peer groups is for
respondents to indicate the name of the group at school they feel they belong to the
most (See Sussman, Unger, & Dent, 2004). As such this measure has been called
“peer group self-identification.” The measure has also been referred to as “peer
crowd affiliation” (e.g., La Greca, Prinstein, & Fetter). Generally the protocol
requires students to either write the name of the group or to select a specific group
from a list of groups derived from previous work or pilot studies. Groups are then
collapsed into general categories. This strategy assesses one’s perceptions of which
group he or she belongs to, rather than assessing patterns of individual friendships as
is done in social network analysis.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
26
Other studies have utilized peer reports of group identification (e.g., Michell,
1990; Urberg, Degimencioglu, Tolson, & Halliday-Scher, 2000). This method
requires students to indicate the peer groups that their classmates belong to. Both
self-reports and peer reports of group identification have their strengths and
weaknesses. For example, social desirability can influence the accuracy of self-
reports such that the adolescent indicates membership in the groups they want to
belong to rather than actually belonging to (Shakib et al., in preparation). On the
other hand, the accuracy of peer reports relies on the perspectives of a few
informants, who may or may not have complete knowledge of all their classmates.
High-risk groups. Studies that have examined peer group self-identification
have consistently identified a general group of students referred to as “high-risk”
youth, including peer groups such as “gang members”, “stoners”, “burnouts”,
“druggies”, “taggers”, “rappers”, and “heavy metalers/rockers”(See Sussman, Unger,
& Dent, 2004). These youth are generally low on academic achievement, are not
involved in school activities, and engage in risky activities such as substance use.
High-risk youth are also higher on psychological symptoms. They are more
depressed, impulsive, and rebellious relative to their classmates (Barber, Eccles, &
Stone, 2001). Last, students who identify with high-risk groups are more likely to be
involved in violence one year later. These findings will be discussed in greater detail
shortly.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
27
Peer groups and violence: Theoretical models
Dominance theory and social cognitive theory suggest that motivation for
reinforcement (e.g., increased social status) and behaviors of specific peers (e.g.,
friends’ violent perpetration) influence involvement in violence. Network analysis
can assess relationships one has with specific students, and therefore is an ideal tool
to test these theories. The peer group, however, consists of students who may share
similar interests, ideologies, and behaviors, but may or may not have social contact
with each other (Brown, 1989). Identification with high-risk peer groups may be
associated with adolescent violence, regardless of who their specific friends are. The
social-interactional perspective, labeling theory, and primary socialization theory
suggest that peer group self-identification is associated with involvement in violence.
Keep in mind that these theories are not mutually exclusive from dominance
and social cognitive theories. It is possible that motivations, behaviors of others, and
self-identification are all associated with violence.
The social-interactional perspective. The social-interactional perspective
(Eron & Slaby, 1994) suggests that violent perpetration and victimization precede
involvement with high-risk peer groups. First, maladaptive parent-child
relationships and ineffective parenting practices lead to violent behaviors in the
social-interactional perspective. Next, violence may interfere with academic
performance and positive relationships with peers. Finally, by late childhood and
early adolescence, academic failure and peer rejection among violent students puts
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
28
them at increased risk for depressed mood and identification with high-risk peer
groups. Such groups may be actual groups and/or self-images.
Labeling theory. Both labeling theory and theory of secondary deviation
posit that identification with high-risk groups leads to later violent perpetration and
victimization. These theories are not mutually exclusive from the social-
interactional perspective. The former may explain how peer groups emerge, whereas
the latter may explain the effects of identifying with peer groups.
Labeling theory (See Bynum & Thompson, 1989) posits that behaviors are
largely determined by one’s social role, or a set of expectations within a social group
or society that influences behavior. For example, if a student becomes known as a
“tough kid,” it is expected that he or she bullies others, engages in deviant behaviors,
and so on. Over time it becomes difficult to escape one’s label, as labels affect the
way in which people react to the appearance and behaviors of the labeled individual
and the way in which friendships are formed. Hence those who identify with high-
risk groups may be more prone to engage in violence because others expect them to
engage in violence. Others may provoke the labeled individual through direct
confrontations, gossip, or simply by being fearful in his or her presence.
Theory of secondary deviation. The theory of secondary deviation (See
Bynum & Thompson, 1989) takes labeling theory one step further. Both theories
posit that others’ perceptions of an individual influence him or her to engage in
violence. However, theory of secondary deviation suggests that peers’ perceptions
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
29
of an individual influence his or her involvement in violence only if those
perceptions are integrated into the self-concept (See Bynum & Thompson, 1989).
For example, if one has been labeled a “bully” among his or her peers, it does not
influence him or her to bully unless he or she believes in that label. Hence those
who identify with high-risk groups may be more prone to engage in violence because
they view themselves as violent. Hence peer group self-identification, but not
necessarily peer reports of group identification, would be associated with violent
perpetration. The dataset analyzed for this dissertation only contains assessments of
peer group self-identification. Hence, theory of secondary deviation was tested.
Peer groups and violence: Previous findings
This dissertation examines associations between peer group self-
identification and violence. However in the discussion of previous findings I do also
include two studies that examined associations between peer reports of group
identification and violence. These studies are informative especially as literature in
peer group identification and violence is still growing.
Peer reports. In the United Kingdom, Michell (1990) conducted loosely
structured focus groups of 11 to 13-year-olds to identify peer groups and behaviors
associated with each of the peer groups. She identified three main categories of
groups: “top” (popular) students, “middle” (average) students, and “bottom”
students, which included “troublemakers” and “loners”. The students reported that
“troublemakers” were the most violent group, were described as “always looking for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
30
fights,” and were generally avoided by their classmates. The “troublemakers” in this
study resemble a subset of students who are highly aggressive and unpopular among
their classmates (Rodkin et al., 2000).
Urberg et al. (2000) had selected peers label each of their classmates as
belonging to one of the following social crowds identified in previous interviews:
“burnouts” (unpopular students who are uninvolved in school activities), “preps”
(popular students), “jocks” (athletic students), “whiggers” (students who wear baggy
overalls and listen to rap music), “nerds” (unpopular students), “average students”,
“alternatives” (students who wear black clothing and listen to heavy metal music),
and “brains” (smart students). They found that “whiggers” had the highest level of
self-reported delinquency, including violence. Their delinquency levels were
significantly higher than that of “brains”, “nerds”, and “preps”.
Self-reports. Sussman’s articles on continuation high school students
(Sussman et al., 1999; Sussman et al., 2000; Sussman, Unger, & Dent, 2004) suggest
that self-reported identification with certain peer crowds is associated with violence.
The longitudinal study reported in the articles required participants to indicate which
one of 17 peer groups they belonged to. The peer groups were previously identified
in focus groups, and the high-risk groups included “gang members”, “stoners”,
“burnouts”, “druggies”, “taggers”, “rappers”, and “heavy metalers/rockers.” These
groups were classified as “high-risk” based on previous literature. Identification
with high-risk groups was associated with higher rates of violent perpetration relative
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
31
to low-risk groups one year later (Sussman et al., 1999; Sussman et al., 2000) and
five years later (Sussman, Unger, & Dent, 2004). Self-reported identification with a
high-risk group was also associated with higher rates of victimization one year later
relative to low-risk groups (Sussman et al., 2000). Sussman et al. (2000) also found
that “ jocks” and “hotshots” (popular, athletic students who place a high importance
on academics) were most likely to avoid dangerous situations.
Finally, La Greca, Prinstein, and Fetter (2001) examined whether self-
identification with “jocks”, “brains”, “burnouts”, “populars”, “non-conformists”, and
“average” students was associated with higher levels of aggression and antisocial
behavior among respondents’ three best friends. No differences were found, which
indicated that friends’ aggressive and antisocial behaviors were similar across the
peer groups. Unfortunately this study did not examine whether self-identified peer
crowds differed in their own levels of aggressive and antisocial behavior.
In sum the findings suggest that peer-identified and self-identified
membership to specific groups are associated with violent perpetration and
victimization. Specifically, “troublemakers”, “whiggers”, “stoners”, “burnouts”,
“druggies”, “taggers”, “rappers”, and “heavy metalers/rockers” may be at higher risk
for violent perpetration and victimization, whereas “brains”, “nerds”, “preps”,
“jocks”, and “hotshots” may be at lower risk. No known studies have examined
whether targeting specific peer groups in violence prevention programs is an
effective strategy to reduce violent perpetration and victimization.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
32
Timitations of previous studies
The findings of the studies discussed in this chapter have provided evidence
that social network characteristics and peer group identification are associated with
violent perpetration and victimization. However several limitations do exist in such
studies. This section addresses a few of those limitations.
Few studies have examined non-Caucasian ethnic majorities. Since social
network studies have most often been conducted with a primarily Caucasian sample
of students, their findings may not generalize well to other ethnic groups or to
ethnically heterogeneous schools. For example, does the association between
friendship reciprocity and low rates of victimization hold true only in primarily
Caucasian schools, or also schools with a non-Caucasian ethnic majority?
Additional research needs to be conducted with multiethnic samples and samples
with non-Caucasian ethnic majorities.
Aggressive victimization needs to be further examined. Aggressive victims
have generally not been distinguished from violent perpetrators and victims in
previous studies. It is common to examine perpetration and victimization as separate
constructs. However, aggressive victims do differ from those who are only
perpetrators or only victims. For example, perpetrators are believed to have the
ability to regulate their emotions whereas aggressive victims do not (Schwartz,
2000). Aggressive victims generally react to victimization with violence, whereas
victims generally do not (Andreou, 2000). Social network studies suggest that
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
33
aggressive victims are less frequently accepted (Pellegrini, Bartini, & Brooks, 1999;
Schwartz, 2000) and more frequently rejected (Schwartz, 2000) by their classmates
relative to other students including bullies. Finally Schwartz (2000) contended that
aggressive victims are rejected and victimized by their classmates because of their
excessive displays of anger and distress when provoked by other classmates
(Schwartz, 2000).
No known studies have examined whether aggressive victims tend to identify
with high-risk groups more frequently than their classmates. However other
evidence warrants that perpetrators, victims, and aggressive victims differ on many
characteristics. Each of the three groups may have a unique set of social network
and peer group characteristics that are antecedents or consequences of their
involvement in violence.
Multiple network characteristics have not been assessed. Relatively few
studies have examined the association between multiple network characteristics,
violent perpetration, and victimization. For example, some studies have only
assessed friendship reciprocity (e.g., Boulton et al., 1999; Hodges et al., 1999), and
others have examined only peers’ bullying behaviors (e.g., Haselager et al., 1998;
Salmivalli, Lappalainen, & Lagerspetz, 1998). Assessing multiple network
characteristics in a study would make it possible to determine their relative strength
of associations with violent perpetration and victimization. For example, does
violence among one’s friends a play a lesser or greater role in violent perpetration
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
34
relative to one’s network centrality? Is peer reciprocity or friends’ level of
victimization more strongly associated with victimization?
Network characteristics and peer group self-identification have not been
exam ined together. Since no known studies have assessed social network
characteristics and peer group self-identification together, the relative association of
each of them with violent perpetration and victimization is not known. For example,
is identification with a high-risk group more strongly associated with perpetration
and victimization than is one’s position in the social network or the behaviors of
one’s specific friends? Furthermore, are network characteristics and group
identification separate constructs, or are they highly correlated with each other?
Assessing peer group self-identification may further explain some of the
findings of social network research. For example, not all aggressive students are
popular, and not all popular students are aggressive (Rodkin et al., 2000; Farmer et
al., 2003). It is possible that the attitudes and beliefs of specific peer groups largely
determine their involvement in violent perpetration regardless of their popularity
among other classmates. It is also possible that certain peer groups may be more
popular than others, and popularity provides leverage for violent perpetration,
especially during adolescence.
Secondly, social network research indicates that aggressive students do not
exclusively associate with each other and often have several non-aggressive friends
(Farmer et al., 2002). It is possible that self-identification with high-risk groups is
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
35
associated with violent perpetration, but self-identification may be a highly
individual construct that bears little resemblance to actual friendship choices.
Assessing both social network characteristics and peer group identification will help
determine whether violent perpetration and victimization are more strongly
influenced by either one of the two or by both.
Assessing the social network and group identification in one study can help
determine whether friendship networks actually correspond to self-reported peer
groups. In other words, do groups of friends consistently refer to themselves with
the same peer group name, for example, “skaters”, “artistic kids”, and “rockers?”
Sussman, Unger, and Dent (2004) suggest that peer group self-identification may
only reflect the image that one wants to project to peers, or one’s preferences for
things (e.g., music, clothes, etc.), as opposed to friendship choices. They
recommend that future research is needed to tackle this hypothesis.
This dissertation
In this dissertation I examine the frequency of self-reported violence among
middle school students, specifically physical (e.g., hitting, pushing), verbal (e.g.,
threats, saying mean things), and relational (e.g., spreading rumors) forms of
violence. Zeira, Astor, and Benbenishty (2003) caution that researchers should not
limit the focus of school violence to a small set of behaviors. I do acknowledge that
this dissertation only covers a relatively small set of all possible violent behaviors.
Other violent behaviors that may be prevalent among young adolescents, including
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
36
weapon carrying and vandalism, were not assessed. Furthermore, bullying was not
assessed in this dissertation since relationships between individual perpetrator-victim
pairs were not assessed. Despite these shortcomings, the present dissertation studies
still provide preliminary evidence on how the social context is associated with school
violence.
This dissertation will address the limitations of previous studies by (1)
studying a primarily Latino and Asian sample of adolescents, (2) assessing violent
perpetration, victimization, as well as aggressive victimization, (3) assessing
multiple characteristics of the social network, and (4) also assessing peer crowd self-
identification. The data analyzed for this dissertation were from three waves of a
school-based experimental trial of smoking prevention strategies in Los Angeles
County, from Grades 6 through 8. The conceptual model of this dissertation is
presented in Figure 1 below.
Figure 1
Conceptual Model
Perpetration
Friends’ Behaviors Perpetrator
S ta tu s
Victimization
High-risk
Aggressive
V ic tim S ta tu s
Peer Group Self-
id e n tific a tio n
Aggressive
V ic tim S ta tu s
Nominations Sent
Social Status
Nominations
Reciprocated
Victim Status
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
37
This model integrates aspects of social learning theory (Bandura, 2002), the
social-interactional perspective (Eron & Slaby, 1994), dominance theory (Hawley,
1999), and theory of secondary deviation (Bynum & Thompson, 1989) to determine
associations with perpetrator, victim, and aggressive victim status (those who are
perpetrators and victims). Variables believed to be associated with violence are
friends’ behaviors (involvement in perpetration and victimization), social status
(nominations received and reciprocated), and peer group self-identification with
groups that were categorized as “high-risk.” This model assumes that the
associations depicted in the model are not mediated by ethnicity and gender. Hence,
they were not included in the model but were included as covariates in the analyses.
A more detailed description of these measures will be presented in Chapter 2.
Each of the theories best explains one particular aspect of school violence.
First, consistent with social cognitive theory (Bandura, 2002) and previous findings
(e.g., Poulin Cilessen, Hubbard, et al., 1997; Salmivalli, Lappalainen, & Lagerspetz,
1998), friends’ involvement in violence is believed to be associated with one’s own
perpetration. This association is depicted by the first row of associations on the top
portion of the model. Specifically, perpetrators are more likely to have friends with
higher rates of perpetration and lower rates of victimization. This may be because
violent students select friends with similar violent tendencies and avoid those with
dissimilar tendencies (e.g. Poulin and Boivin, 2000), or because violent friends
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
38
influence each other’s violent behaviors through reinforcement (e.g., Dishion,
McCord, & Poulin, 1999), or both. Perpetrator status is not believed to be as
strongly associated with social status since both violent and non-violent students
often occupy central network positions (Farmer et al., 2003; Prinstein and Cillessen,
2003; Bagwell, Coie, Terry, & Lochman, 2000).
Second, consistent with dominance theory (Hawley, 1999), and previous
findings (e.g., Pellegrini et al., 1999; Boulton et al., 1999; Hodges et al., 1999;
Salmivalli, Lappalainen, & Lagerspetz, 1998), it is believed that social status is
associated with victim status. This association is depicted in the third row of
associations on the bottom portion of the model. Specifically victim status is
believed to be associated with a lower number of nominations received and a lower
proportion of nominations reciprocated. Previous research also suggests that
victimized students are likely to have more victimized and fewer violent friends
(Hodges, Malone, & Perry, 1997). However, as violence generally decreases
through adolescence (Petit, 1997), who gets victimized may largely depend on who
lies on the periphery of the social network. Hence, it is believed that low social
status, as opposed to friends’ level of perpetration and victimization, would have
stronger associations with victim status later in adolescence.
Last, consistent with the social-interactional perspective (Eron & Slaby,
1994), it is believed that aggressive victims are prone to later self-identification with
high-risk groups. Furthermore, consistent with primary socialization theory (Bynum
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
39
& Thompson, 1989) and previous findings (Sussman et al., 1999; Sussman et al.,
2000; Sussman, Unger, & Dent, 2004), it is believed that high-risk status is
associated with later aggressive victim status. Both of these associations are
depicted in the second row, in the middle portion of the conceptual model. Why
would aggressive victim status, as opposed to perpetrator or victim status, be more
strongly associated with high-risk status? Aggressive victims are unique from
perpetrators-only in that other students victimize them. Furthermore aggressive
victims are unique from victims-only (Roland & Idsoe, 2000) in that they do often
retaliate against other students with violence. Hence, aggressive victims may also
self-identify with high-risk groups in an attempt to actively cope with victimization.
Furthermore high-risk status leads to the maintenance of aggressive victimization.
The three dissertation studies will test parts of the conceptual model.
Specifically, Study 1 will examine will examine cross-sectional associations between
network characteristics and violence (perpetrator, victim, and aggressive victim
status) in the 6th grade. As such, Study 1 tests social cognitive and dominance
theories of violence. Study 2 will determine whether 7th grade high-risk status is
associated with 6th grade violence. As such, Study 2 tests the social-interactional
perspective of high-risk peer group self-identification. Study 2 will also examine
whether members who self-identify with certain peer groups are friends with
students who self-identify with those same peer groups. Finally, Study 3 will
examine whether 6th grade network characteristics and 7th grade high-risk self-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
40
identification are associated with 8th grade violence, after adjusting for 6th grade
violence. As such, Study 3 tests social cognitive, dominance, and primary
socialization theories of school violence. All of the analyses adjust for gender,
ethnicity, and school levels of violence in a nested model.
It is particularly important to account for variations in violence among
schools since school characteristics may strongly influence violence. Furlong and
Morrison (2000) note that schools with low levels of violence tend to be firm and
consistent with their policies, are smaller in size, and have lower levels of crowding
(Furlong & Morrison, 2000).
The next chapter describes the sample, research methods, and measures
assessed in the longitudinal study. Next, Chapters 3, 4 and 5 present the research
hypotheses, analytic plans, results, and brief discussions for Studies 1, 2, and 3
respectively. Finally, Chapter 6 summarizes the overall findings, limitations, and
implications for future research.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
41
Chapter II: Methods
This data analyzed for this dissertation were from one longitudinal study.
This chapter describes the sample, research procedures, and measures utilized in that
study in great detail. As such, this chapter is the Methods section for Studies 1,2,
and 3. The analyses conducted for each of the three studies will be discussed in their
respective chapters.
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 the 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 of them 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
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
42
approached, 38 did not receive administrator 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 into 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 take the survey and provided written parental
consent participated. A total of 3,190 students who provided active parental consent
and student assent completed the survey in the 6th grade. Of these, 2,822 students
completed the 7th grade survey, and 2,561 students completed the 8th grade survey.
A total of 2,292 students completed surveys in all three waves.
Students attending the 16 program schools also completed a shorter survey in
the 6th grade, which included items that asked students write 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. Eight of
these schools received a smoking prevention program that focused of cultural issues,
and the other 8 schools received the traditional program. A total of 1,562 completed
this shorter survey. The following year, all of the 16 program schools were invited
for a teacher’s training to implement the second year of the curriculum. Those
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
43
students whose schools accepted the invitation, received teacher training, and
implemented the curriculum were given the friendship nomination survey for a
second consecutive year. The other schools only completed the larger survey. A
total of 1,199 7th grade students completed this survey. Students who remained in
the study were compared to those who were lost to follow-up on the variables of
interest in each study. These comparisons will be presented in subsequent chapters.
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 the students all were attending English-language
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
Several variables of interest in this dissertation were not assessed each year.
In this section I will clarify which years each of the measures were assessed.
Perpetrator, victim, and aggressive victim status. Four items were adopted
from Olweus (1991) to assess self-reported physical and verbal forms of perpetration
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
44
and victimization during the past three months. These items were assessed in the 6th
th
grade and in the 8 grade. The perpetration 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 awhile”, and 0 = “never.” A total aggression score and a total victimization
score were calculated by summing responses on their two respective items. Hence,
perpetration scores and victimization scores ranged from 0 to 6.
Three dichotomous variables, perpetrator (yes/no), victim (yes/no), and
aggressive victim (yes/no) were created. In the 6th grade, students were classified as
“perpetrators” if they scored 4 or higher on perpetration and less than 4 on
victimization, “victims” if they scored less than 4 on perpetration and 4 or higher on
victimization, and “aggressive victims” if they scored 4 or higher on both
perpetration and victimization. I used cutoff of 4 so that students who were
moderately to frequently involved in bullying, victimization, and aggressive
victimization were identified. This coding procedure resulted in approximately half
of the sample being categorized as a perpetrator, victim, or aggressive victim.
In the 8th grade, however, scores on perpetration and victimization items were
substantially lower, and nearly 90% of the sample did not meet the criteria for
perpetrator, victim, or aggressive victim status. In other words, they were control
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
45
students. To address this issue, I used cutoff of 2 so that students who were at least
“once in awhile” involved in bullying, victimization, and aggressive victimization
were identified.
In the 8 grade an additional two items concerning rumor spreading
perpetration and victimization were assessed. The victimization item was the
following: “Did another kid spread rumors about you?” The perpetration item was
the following: “Did you spread rumors about another kid?” The response options
for these items were the same as the previous ones. The same criteria used to
identify perpetrators, victims, and aggressive victims for physical and verbal
violence was also used for rumor spreading violence. However a cutoff of 1 was
used since two items were assessed instead of four.
Friendship network variables. Friendship network variables were assessed in
the 6th and 7th grades 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’ perpetration, and friends’ victimization.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
46
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 the total number of students in the class. Reciprocity was the proportion of
nominations sent that was reciprocated with a nomination received. Since friendship
nominations and violence variables were all collected only during the 6th grade,
friends’ level of perpetration and victimization were only assessed during the 6th
grade. Friends’ perpetration was the mean perpetration score of the classmates that
he or she nominated, and ranged from 0 to 6. Friends’ victimization was the mean
victimization score of the classmates that he or she nominated, and ranged from 0 to
6 .
Peer group self-identification. Peer group self-identification was assessed in
the 7th and 8th grades with a mark-all-that-apply item: "What group of kids do you
hang out with the most?" The response options were “jocks” (athletic kids, sports
kids), “skaters/bladers”, “artistic” kids (artists, musicians, actors), “rockers”,
“paisas”, “popular” kids, “smart” kids, “gamers”, “religious” kids,
“gangsters/cholos”, and “other” (write-in your group). Response options were
determined through a literature search of previous assessments by Sussman (1990)
and pilot work with the IRP population. Our pilot investigations included a short
survey of a subset of teachers in the study schools ("We would like your help in
determining what kinds of groups or cliques exist at your school. What are the terms
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
47
that students in your classroom use? Below are examples of group names that have
been suggested to us previously. Please amend our list in any way you think
appropriate and add as many group names as you can think of."). The pilot
investigation also included survey of 370 middle-school students ("People often hang
out in different groups at school and outside of school. A group of friends could be a
group that does something together (like playing basketball), or a group that likes the
same things (like rap music). Please list all the groups that you can think of at your
school and give each group a name."). These data were examined to identify the
most frequently mentioned groups and any groups that might be specific to our
population.
Compared with previous assessments, this study included a group name that
was relatively recent in origin (“gamers”), and two group names that were specific to
Latino culture (“cholos”, “(newly-arrived) immigrants”). Pilot work suggested that
middle school students tend to identify with multiple groups and therefore the item
was written to allow respondents to endorse as many groups as applied to them. The
coding scheme utilized for this item will be described in Chapter 4.
Ethnicity. Self-reported ethnicity was assessed in the 6th and 8th grades. The
6th grade assessment consisted of 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/Hispanic, and (8)
Black/African-American”. Response options for each of these questions were 1=
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
48
“yes” and 0= “no.” Since Latinos and Asians comprised the majority of our sample
(53.8% and 22.8%, respectively) three dichotomous variables were created: Latino,
Asian (those who answered “yes” to being Chinese, Pacific Islanders, Filipinos,
Koreans, or Vietnamese), and other (those who did not identify themselves as being
either Asian or Latino). The 8th grade assessment consisted of the question stem:
“My ethnicity is.” The same ethnic categories in the 6th grade were presented in the
8th grade, but in a forced-choice format. Again, the three dichotomous ethnicity
variables were created.
Self-reported grades. Self-reported grades were assessed in the 6th grade with
the item: “What grades did you get in school last year?” Response options were
“mostly A’s”, “mostly B’s”, “mostly C’s”, “mostly D ’s”, and “mostly F’s.”
Lifetime smoking. Lifetime smoking was assessed in the 7th grade with the
question: “Have you ever tried smoking, even a few puffs?” Those students who
responded “yes” were categorized as lifetime smokers.
Depressive symptoms. Depressive symptoms were assessed in the 7th grade.
Five items were adapted from the Center for Epidemiological Studies Depression
Scale (CES-D; Radloff, 1991), a 20-item self-report scale that assesses past week
depression. In a pilot study the 20 CES-D items were factor analyzed using the
principal components method to determine which items to use in the main trial of the
longitudinal study. Consistent with suggestions from previous research, five items
were chosen that loaded the highest on the factor, labeled "depression." The question
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
49
stem was the following: “Think about how you felt during the past 7 days. On how
many of these days did you:” Items were: “Have trouble shaking off sad feelings?”
“Feel depressed?” “Think your life had been a failure?” “Feel lonely?” “Feel sad?”
Response options were: 1= “0-1 day”, 2= “2-3 days”, 3= “4-5 days”, and 4= “6-7
days”. Scores on these items were summed. Cronbach’s alpha for the entire scale
was .87.
Hostility. Hostility was assessed in the 7th grade. To assess self-reported
hostility, the following 4 questions adapted from the Buss-Durkee Hostility
Inventory (Buss & Durkee, 1957) were asked. They were the following: “I lose my
temper easily.” “Sometimes people bother me just by being around.” “I can’t help
being a little rude to people I don’t like.” “Lately, I have been kind of grouchy.”
Responses were rated on a 4-point scale: 0= “definitely no,” to 3= “definitely yes.”
Scores on these items were summed. Cronbach’s alpha for the entire scale was .73.
Parental monitoring. Parental monitoring, reports of parents’ knowledge and
control over student whereabouts, was assessed in the 7th grade. To ascertain
perceptions of parental monitoring, a 5 question series was adapted. The items were
the following: “Do you ever go places that your parents don’t want you to go?”
“When you go out with friends, do your parents ask you where you are going?” “Are
you allowed to go out with friends that your parents don’t know?” “How important
is it to your parents to know where you are at all times?” “How often do your
parents really know where you are?” Response options ranged from 1= “very often,”
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
50
to 4= “never.” Mean scores were calculated across the items. Cronbach’s alpha was
.64 for the scale.
Communication with parents. Communication with parents was assessed in
the 7th grade. Questions were adapted to assess perceived communication with
parents. They were the following: “How often do you talk with your parents about
what is on your mind?” “How often do you ask your parents for advice?” “How
often do you tell your parents your secrets?” “If you had a problem would you be
able to talk to your parents about it?” The four response options ranged from: 1=
“very often,” to 4- “never”. The Cronbach’s alpha was .83 for the scale.
School bonding. School bonding was assessed in the 7th grade and consisted
of the items: “Does your teacher really care about the students in your school?”
“Does your teacher really care about the students in your school?” “Are the things
you learn in school important?” Response options ranged from 1= “no, definitely
not”, to 4= “yes, definitely.” Items were summed. Cronbach’s alpha for the entire
scale was .63.
The analyses for each of the respective studies will be described in their
respective chapters.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
51
Chapter III: Study 1
The conceptual model described in Chapter 1 suggests that social cognitive
theory (Bandura, 2002) best explains perpetrator status, whereas dominance theory
(Hawley, 1999) best explains victim status. Study 1 tests social cognitive theory and
dominance theories of school violence with a cross-sectional design using 6th grade
baseline data. It also expands upon previous research by examining the unique
effects of each network variable, that is, adjusted for the effects of the other network
variables. Study 1 also examines whether ethnic differences exist among
perpetrators, victims, and aggressive victims, a topic that is yet to receive
considerable attention. All analyses were performed on the entire sample and among
females and males separately. A variation of this study is currently in press
(Mouttapa, Valente, Gallaher, Rohrbach, & Unger, in press).
Predictions
Predictions pertaining to associations between network characteristics,
perpetration, victim, and aggressive victim status were derived from dominance
theory and social cognitive theory. They are the following:
1. Consistent with social cognitive theory (Bandura, 2002) and previous
findings (Poulin Cilessen, Hubbard, et al., 1997; Salmivalli, Lappalainen, &
Lagerspetz, 1998), I expected that the friends of perpetrators would score
higher on self-reported perpetration and lower on self-reported victimization
relative to the friends of other students. Furthermore I predicted that the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
52
friends of aggressive victims would report higher rates of both perpetration
and victimization relative to the friends of other students.
2. Consistent with dominance theory (Hawley, 1999) and previous findings
(e.g., Pellegrini et al, 1999; Boulton et al., 1999; Hodges et al., 1999;
Salmivalli, Lappalainen, & Lagerspetz, 1998), I expected that victims would
have less social status (e.g., fewer nominations received and lower friendship
reciprocity) relative to their classmates. Previous research also suggests that
in elementary school, victimized students are likely to have more victimized
and fewer violent friends (Hodges, Malone, & Perry, 1997). Since these
students were surveyed shortly after finishing elementary school (in the
beginning of the 6th grade), it was also predicted that the friends of victims
would report higher rates of victimization and lower rates of perpetration.
3. Finally, similar to the findings of studies that examined a Latino-majority
Southern Californian sample (Graham & Juvonen, 2002; Furlong et al.,
1998), I predicted that Latinos would more frequently be perpetrators and
less frequently victims relative to Asian students and students from other
ethnic groups. It was not possible to compare African-American and
Caucasian students with other ethnic groups in this study since their numbers
were too small in this sample.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
53
Sample
A total of 1,368 students (52.9% female) had complete data on the variables
of interest in the 6th grade, 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 small numbers of non-
Hispanic Whites, African Americans, and other ethnic groups (classified as “other”).
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 Generalized Linear Models.
Regression of perpetrators, victims, and aggressive victims. The
dichotomous outcome variables in this study were physical and verbal perpetrator,
victim, and aggressive victim status. Logistic regression models were estimated to
determine whether nomination variables (friendship nominations sent, received, and
reciprocated), friends’ behavior (friends’ involvement in perpetration and
victimization), and ethnicity were associated with each of the dependent variables
separately: classification as a perpetrator, victim, and aggressive victim. GLIMMIX,
a SAS macro (Wolfinger, 1998), was used to control for the frequency of the
outcome variable within each student’s school. Hence, each of the models was
nested within schools. As mentioned in Chapter 1, it is particularly important to
account for variations in violence among schools since school characteristics may
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
54
strongly influence violence. Hence, associations in a nested model are adjusted for
potential confounders such as school policies and class size.
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 to test the main research hypotheses.
Results
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
missing data were more often perpetrators (p < 0.0001) and victims (p < 0.0001), and
less often aggressive victims (p <0 .0001) relative to those with complete data. The
mean scores of friends’ perpetration and victimization were lower among those with
complete data (p < 0.01). Those with complete data sent and received more
friendship nominations (p < 0.0001), had a higher proportion of reciprocated
friendships (p < 0.05), were more often Asian (p < 0.0001), and less often Latino (p
<0.01).
Gender differences on ethnicity and network variables
Table 1 presents gender differences on network characteristics and outcome
variables. Males had more friends who were perpetrators (3.6; p < 0.0001) and
victims (4.0; p < 0.0001) relative to females (3.1 and 3.8, respectively). Females had
a higher proportion of reciprocated friendships (58.7%) relative to males (54.3%; p <
0.01). Consistent with previous findings (e.g., Crick, Casas, & Ku, 1999), males
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
55
were more often perpetrators (9.3%; p < 0.01) and aggressive victims (36.3%; p <
0.0001) relative to females (5.8% and 26.1%, respectively).
Table 1
Gender Differences on Network Characteristics and Outcome Variables
Males fn=
Mean
645)
SD
Females fn=
Mean
723)
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
Reciprocity 0.54 0.30 0.59 0.29 7.66**
Friends’ Perpetration 3.59 1.09 3.08 0.90 86.94***
Friends’ Victimization 4.02 1 .0 2 3.77 1.03 20.50***
n % n %
x2
Outcome Variables
Perpetrator 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***
* g < 0.05, **g< 0.01,***g< 0.0001.
Ethnic differences were examined among perpetrators, victims, and
aggressive victims in univariate analyses. No ethnic differences were found in
perpetrator status. However, victims were more likely to be Asian (30.3%) relative
to Latinos (21.2%) and others (19.7%; p < 0.01). Furthermore Asians were less
likely to be aggressive victims (35.3%) relative to Latinos (43.5%) and others
(47.8%; p < 0.01).
Logistic regression analyses
This section presents the logistic regression results of perpetrators, victims,
and aggressive victims.
Perpetrator status. Table 2 presents logistic regression of perpetrator status.
As predicted, the friends of perpetrators had higher rates of perpetration (adjusted
odds ratio= 1.31 ,p<0 .0001) and lower rates of victimization (adjusted odds ratio=
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
56
0.79, p < 0.05) in the entire sample. These findings did not reach statistical
significance when the model was run for each gender separately. Among females
only, perpetrators sent a larger number of friendship nominations (adjusted odds
ratio=2.16, p < 0.05), received fewer friendship nominations (adjusted odds ratio=
0.77, p < 0.05), but had a higher proportion of reciprocated friendships (adjusted
odds ratio= 7.55, p < 0.05) relative to non-perpetrators. This finding suggests that
female perpetrators belong to smaller, more cohesive friendship groups relative to
other females. Male perpetrators did not differ from male non-perpetrators on any of
the variables in the model. Contrary to the predictions, perpetrators and non
perpetrators did not vary by ethnicity.
Table 2
Logistic Regression of Perpetrator Status
Total (n= 1368) Males (n= 645) Females (n= 723)
AOR 95% Cl AOR 95% Cl AOR 95% Cl
Friends’ Behavior
Perpetration
j 3 j ***
1.09-1.57 1 .2 0 0.95-1.53 1.35 0.98-1.87
Victimization 0.79* 0.64-0.98 0.81 0.60-1.08 0.77 0.55-1.06
Nominations
Sent 1.23 0.89-1.70 1 .0 2 0.70-1.48 2.16* 1.05-4.46
Received 0.95 0.85-1.07 1.04 0.90-1.20 0.77* 0.63-0.95
Reciprocated 2.45 0.88-6.85 1.72 0.45-6.58 7.55* 1.42-40.15
Ethnicity
Latino (vs. Other) 1.40 0.81-2.41 1.51 0.74-3.09 1.14 0.49-2.65
Asian (vs. Other) 1.58 0 .8 8 -2 .8 6 1.78 0.83-3.84 1.41 0.54-3.67
Note. All odds ratios are adjusted for all other variables in the model and the frequency of
perpetration within each student’s school.
* g <0 .05, **p<0.01, ***p <0 .0001.
Victim status. Table 3 presents logistic regression of victim status. Victims
were disproportionately Asian (adjusted odds ratio = 1.78, p < 0.0001), and this
finding was also observed among males separately (adjusted odds ratio = 2.02, p <
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
57
0.01). The friends of victims reported lower rates of perpetration (adjusted odds
ratio = 0.84, p <0 .01) relative to the friends of non-victims, and this finding was also
observed among males separately (adjusted odds ratio = 0.80, p <0 .01). Contrary to
the predictions, the friends of victims did not differ on victimization relative to the
friends of non-victims. Also contrary to the predictions, victims did not differ on
nominations received and reciprocated.
Table 3
Logistic Regression of Victim Status
Total (n= 13681 Males (n= 6451 Females (n= 7231
AOR 95% Cl AOR 95% Cl AOR 95% Cl
Friends’ Behavior
Perpetration 0.84** 0.74-0.95 0.80* 0.67-0.95 0.89 0.73-1.09
Victimization 1.05 0.93-1.18 1.14 0.95-1.36 0.96 0.82-1.13
Nominations
Sent 0.90 0.77-1.05 0.94 0.84-1.04 0.85 0.68-1.06
Received 0.95 0.89-1.02 0.94 0.75-1.16 0.97 0.88-1.06
Reciprocated 1.16 0.65-2.06 1.30 0.55-3.08 1 .0 1 0.46-2.22
Ethnicity
Latino (vs. Other) 1.23 0.90-1.67 1.25 0.79-1.97 1.17 0.76-1.78
Asian (vs. Other)
j y g * * *
1.27-2.48 2 .0 2 ** 1.25-3.27 1.56 0.98-2.50
Note. All odds ratios are adjusted for all other variables in the model and the frequency of
victimization within each student’s school.
*g < 0.05, **p< 0 .01, ***g < 0 .0001.
Ethnic differences in victim status. An additional two logistic regression
models were performed to further explore ethnic differences in victim status. The
models tested whether Asians and Latinos differed in victimization from non-Asians
and non-Latinos, respectively, in Asian majority schools and in Latino majority
schools. These models controlled for all covariates. No ethnic differences in
victimization were found in Asian majority schools. However, Asians were more
frequently victimized in Latino majority schools (adjusted odds ratio = 2.65, p <
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
58
0.01). Hence, Graham and Juvonen’s (2002) finding that ethnic minorities are
victimized more frequently was only the case in Latino majority schools in this
study.
Since the proportion of Latinos and Asians varied greatly in the schools, it is
not known whether the findings are due to numeric differences in the sizes of the
ethnic groups or to specific qualities of Latinos and Asians. For example, many
Latino-majority schools were over 90% Latino, whereas the Asians in Asian-
majority schools were never such a strong majority. Therefore the finding that
Asians were more frequently victimized in Latino-majority schools may be due to
the fact that they were severely outnumbered by Latinos. Conversely Latinos in
Asian-majority schools are not nearly as outnumbered, which may protect them from
increased victimization.
Aggressive victim status. Table 4 presents logistic regression results of
aggressive victim status. As predicted, the friends of aggressive victims were more
aggressive than the friends of other students (adjusted odds ratio = 1.19, p < 0.01).
Furthermore aggressive victims were less often Asian in the entire sample (adjusted
odds ratio = 0.75, p < 0.05) and among males only (adjusted odds ratio = 0.67, p <
0.05). Aggressive victims did not differ from other students on any of the
nomination variables.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
59
Table 4
Logistic Regression of Aggressive Victim Status
Total (n= 1368) Males (n= 645) Females fa - 723)
AOR 95% Cl AOR 95% Cl AOR 95% Cl
Friends’ Behavior
Perpetration 1.19** 1.08-1.32 1.14* 1.00-1.31 1.2 1 ** 1.02-1.43
Victimization 1 .0 0 0.91-1.11 0.89 0.76-1.03 1 .1 1 0.96-1.28
Nominations
Sent 1.05 0.91-1.21 1.05 0.87-1.27 1 .1 0 0.88-1.36
Received 0.98 0.93-1.04 0.98 0.90-1.06 0.97 0.90-1.05
Reciprocated 0 .6 8 0.42-1.11 0.60 0.29-1.24 0.93 0.46-1.87
Ethnicity
Latino (vs. Other) 0.94 0.73-1.21 0.78 0.54-1.12 1.14 0.79-1.67
Asian (vs. Other) 0.75* 0.56-0.99 0.64* 0.52-0.96 0.94 0.60-1.45
Note. All odds ratios are adjusted for all other variables in the model and the frequency of
victimization within each student’s school.
* b < o.o5, **e < o .0 0 6 , ***e < o .0 0 0 1 .
Conclusions
This section draws some conclusions based specifically on the findings of
Study 1. General findings and limitations that pertain to all three studies of this
dissertation will be addressed in subsequent chapters.
Several studies have examined the friendship network characteristics of
perpetrators and victims. However, the present study compared perpetrators,
victims, and aggressive victims with other adolescents on multiple network
characteristics. The conceptual model posits that social cognitive theory (Bandura,
2002) best explains perpetrator status, and dominance theory (Hawley, 1999) best
explains victim status. However Study 1 findings suggest that social cognitive
theory best explains violent perpetration, as well as victimization, and aggressive
victimization among 6th grade adolescents. The behaviors of one’s friends more
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
60
strongly predicted one’s own behaviors relative to one’s centrality in the social
network.
The most robust findings were observed in the amount of perpetration among
one’s nominated friends. Specifically, perpetrators and aggressive victims had more
violent friends relative to other students, and victims had less violent friends relative
to other students. These findings suggest that the presence of violent friends is
strongly associated with perpetration, whereas the absence of violent friends is
associated with victimization.
The cross-sectional association between individuals’ and friends’ violence
has been demonstrated in several studies and with several forms of violence
including fighting (Haselager, Hartup, van Lieshout, & Riksen-Walraven, 1998;
Kupersmidt, DeRosier, & Patterson, 1995), bullying (Haselager et al., 1998;
Pellegrini, Bartini, & Brooks, 1999), and aggressive behavior (Tremblay, Masse,
Vitaro, & Dobkin, 1995; Poulin & Boivin, 2000). Similarly, among victims,
previous research indicates that victimization is negatively 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).
Contrary to previous findings (Browning, Cohen, & Warman, 2003), the
friends of victims did not experience higher rates of victimization relative to the
friends of non-victims. Thus, this study suggests that some individuals from non-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
61
aggressive peer groups get victimized more frequently than others. Perhaps
victimized students in this study differed from their non-victimized friends on
intrapersonal characteristics such as self-blame, anxiety, low self-worth, and
depression. These characteristics have been associated with victimization in
previous studies (Graham & Juvonen, 1998; Hawker & Boulton, 1999). Olweus
(1991) describes an entire typology of victims that include family and personality
profiles. Such characteristics may shed light on why some students are singled out
for victimization.
Dominance theory (Hawley, 1999) posits that violence facilitates access to a
central position in the peer network. On the contrary, the findings of this study
suggest that perpetrators generally do not differ from other students on measures of
centrality. Similarly Rodkin et al. (2000) found that males who occupy central
positions in the peer network were heterogeneous in their level of aggressive
behaviors. Furlong, Morrison, and Greif (2003) suggest that future studies should
examine why some popular students take advantage of the opportunity to be
aggressive while others do not. The answer to this question may be partially
revealed by examining perceived expectations among popular students. Rigby
(1997) found that one’s belief that close friends would accept him or her bullying
others was positively associated with intentions to bully and bullying behaviors.
Despite their social prominence, popular students are disproportionately subject to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
62
group influences relative to other students (Haynie, 2001). This might be the case
for violent perpetration as well.
Dominance theory (Hawley, 1999) also posits that victims would have lower
social status relative to other students. However, Study 1 indicated that victims did
not differ from other students on nominations sent, received, and reciprocated. Why
would this be the case? Pellegrini and Pellegrini and Bartini (2001) suggest that
violence is used rather liberally in the 6th grade to establish dominance in the
classroom. Hence, many students, regardless of their social status, may be
susceptible to victimization. Later in middle school, however, violence generally
decreases (Petit, 1997). Hence, those with lower social status may be singled out for
victimization. Study 3 will examine whether lower social status is associated with
later victimization.
A noteworthy trend was found among females, such that aggressors belonged
to relatively small groups, but their friendships were more often reciprocated. This
suggests that female aggressors occupy less central network positions, but have
stronger friendships. Since physical and verbal aggression is less common among
adolescent females (e.g., Crick, Casas, & Ku, 1999), as well as in this study, it is
plausible that female violence is less accepted among adolescents. These findings
are consistent with those of Farmer, Leung, et al. (2002). They found that aggressive
girls belonged to smaller friendship groups consisting mostly of girls like
themselves. Since females tend to use relational forms of violence (e.g., rumor
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
63
spreading, social ostracism), further research is needed to determine whether
relational violence is associated with high centrality among females.
Asians were more frequently victimized in Latino majority schools, but
Latinos were not more frequently victimized in Asian majority schools. Such
findings challenge the assumption that all ethnic majority groups are protected from
victimization and all ethnic minorities are prone to victimization. Much may depend
on the social context: which particular ethnic group is the majority and which ethnic
group is the minority, differences in school policies, actual percentages of each
ethnic group, as well as many other factors specific to each ethnic group including
cultural values, immigration history, and parenting characteristics, to name a few.
More research in multiethnic settings is necessary to determine which characteristics
of specific ethnic groups are associated with aggression, victimization, and
aggressive victimization when the groups are majorities and minorities at school. It
is also of great importance to make distinctions between violence occurring between
ethnic groups and violence occurring within ethnic groups. Such distinctions will
provide further insight into the ethnic majority and minority effects of violence
found in previous studies (e.g., Graham & Juvonen, 2002; Furlong et al., 1998).
Limitations
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. Another
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
64
limitation of Study 1 is that the results are based on cross-sectional analyses.
Therefore causal relationships between friendship network characteristics and
perpetration, victimization, and aggressive victimization cannot be inferred.
Previous research suggests that similarities in aggressive behaviors precede
friendship formation, rather than the reverse (Poulin & Boivin, 2000). In this study it
is not known whether adolescents tend to select similarly aggressive (or non-
aggressive) friends, and/or whether groups of friends influence each other towards
higher (or lower) levels of aggressive behavior. Longitudinal studies would help
determine whether selection or influence models most appropriately illustrate the
behavioral trajectories of perpetrators, victims, and aggressive victims.
Implications
The findings suggest that the violent behaviors of one’s friends, as opposed to
one’s social status, are more strongly associated with perpetrator, victim, and
aggressive victim status. These findings were observed among both male and female
adolescents. Violence prevention efforts that reach highly violent students may also
effectively reduce violence among their friends, or may encourage them to befriend
less violent students. Furthermore, assertiveness training in handling violent
situations could be beneficial for both violent and non-violent students alike. Such
training may help students defend themselves and their friends effectively from
potential perpetrators using nonviolent strategies.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
65
Chapter IV: Study 2
The conceptual model described in Chapter 1 suggests that the social-
interactional perspective (Eron & Slaby, 1994) best explains why aggressive victims
later identify with high-risk peer groups. Previous studies have examined whether
high-risk status is associated with later violent perpetration (Sussman et al., 1999;
Sussman et al., 2000, Sussman, Unger, & Dent, 2004) but no known studies have
tested the social-interactional perspective by examining the opposite order of events.
Study 2 examines whether violent perpetrator, victim, and aggressive victim status in
the 6th grade is associated with high-risk status in the 7th grade, adjusting for social
network characteristics. Hence, the associations between group identification and
violence are adjusted for potential confounders such as centrality in the social
network and friends’ involvement in violence. Similar to Study 1, analyses were
also adjusted for gender, ethnicity, and school levels of violence.
Prior to examining associations between group identification and violence,
Study 2 examines cross-sectional characteristics of 7th grade high-risk and low-risk
groups in greater detail, specifically on demographic, social network, and
psychosocial characteristics, and relationships with parents and to school.
Last, Study 2 examines whether one’s peer group self-identification in the 7th
grade is linked to one’s friendship network, a question that has been addressed in few
studies. La Greca, Prinstein, and Fetter (2001) found that 82% of adolescents had at
least one of their three best friends self-identified with the same peer group.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
66
Moreover the highest-risk group, “burnouts”, had no friendships with the lowest-risk
group, “brains.” Urberg, Degirmencioglu, Tolson, and Halliday-Scher (1995) on the
other hand concluded that although some overlap was found, neither best friends nor
friendship groups were embedded in the same peer group. These studies differ from
Study 2 in that they did not allow respondents to report identifying with multiple
peer groups. Hence, Study 2 examines whether friendship groups identify with
similar constellations of peer groups as opposed to specific peer groups. For
example, a group of friends may not identify with the exact same peer groups, but
they may have considerable overlap (e.g., Friend A identifying with 3 out the 4 peer
groups that Friend B identified with). Keep in mind that these analyses are only
descriptive and are therefore preliminary.
Predictions
The predictions for Study 2 were the following:
1. Consistent with previous research (e.g., Barber, Eccles, & Stone, 2001;
Patterson & Dishion, 1985), I predicted that 7th grade youth who identify with
high-risk groups would report lower grades, school bonding, parental
monitoring, and parent-child communication, and higher rates of smoking,
hostility, and depressive symptoms relative to low-risk students.
2. The social-interactional perspective (Eron & Slaby, 1994) suggests that some
violent peers join deviant groups because other classmates reject them.
These students may identify with certain attitudes and beliefs, which lead to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
67
acts of violence that are not well accepted by classmates, which finally lead
to subsequent victimization. Hence, I predicted that 6th grade aggressive
victims would more likely identify with high-risk groups in the 7th grade
relative to other students.
3. Last, Urberg et al. (2000) suggest that peer groups channel friendships among
similar adolescents. Hence I predicted that those students who identify with a
specific peer group (e.g., “popular” students) would have a greater number of
friends who also identify with that specific peer group than with other groups.
Furthermore I predicted that high-risk students would have a higher
proportion of high-risk friends, and low-risk students would have a lower
proportion of high-risk friends.
Sample
A total of 2,822 students completed the 7 grade survey and comprised the
analytic sample. However the sample sizes for different analyses did vary depending
on which variables were being considered. Of the 2,822 students, 2,701 of them
(95.7%) were surveyed in the 6th grade, and 121 of them (4.3%) were surveyed for
the first time. The retention rate from the 6th to 7th grade was 84.7%. The analytic
sample was 50.4% female and their mean age was 12.3 years (SD = 0.50). Ethnicity
was assessed in the 6th grade but not in the 7th grade. Among those who provided
data on ethnicity in the 6th grade, nearly half (47.7%) were Latino, 21.9% were
Asian, and 30.5% belonged to other ethnic groups.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
68
Analyses
Identification of general peer groups. Exploratory factor analysis of peer-
group self-identification items was performed to identify general categories of peer
groups that adolescents identify with the most. The principal components and
Varimax rotation method options were selected. The intention was to create
orthogonal factors and identify a distinct set of peer groups, rather than identifying
clusters of groups that are highly correlated with each other. Two factors were
identified using the eigenvalue > 1 criterion. One was labeled “high risk” and the
other “low-risk.”
Differences between high-risk and low-risk groups. High-risk and low-risk
groups were compared on demographic characteristics, self-reported grades, social
network variables, psychosocial variables, and adult authority. Comparisons were
made with chi-square analyses and Generalized Linear Models.
Logistic regression of peer group self-identification. Next I examined
whether 6th grade violence and social network characteristics were associated with
i t
high-risk group identification in the 7 grade. Only those variables that were
significantly associated with high-risk group identification in univariate analyses
were included in this model. Covariates included gender and ethnicity. PROC
GLIMMIX was used to control for similar rates of high-risk students within schools
in a nested model. As mentioned before, it is particularly important to account for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
69
variations in violence among schools since school characteristics may strongly
influence violence.
Peer groups and friendships. I assessed within-group differences in the mean
number of nominated friends who self-identified with each of the 10 peer groups.
For instance, the mean number of “jocks”, “gamers”, “artists”, etc., that “smart kids”
nominated as friends were compared to each other. This was done by calculating the
mean and 95% confidence intervals for friendship nominations of each peer group,
within each peer group. Keep in mind that the 10 peer groups are not mutually
exclusive since students were allowed to identify with as many peer groups they
wanted. Nevertheless general patterns were examined.
Next, I assessed whether peer groups differed in the proportion of high-risk
students they nominated. For example, did “gamers” nominate a higher proportion
of high-risk group members as friends as did “artistic kids?” Finally, a t-test was
calculated to determine whether high-risk groups, relative to low-risk groups,
nominated a higher proportion of high-risk friends. As we will see later, high-risk
and low-risk peer groups were constructed such that they were mutually exclusive.
Results
Attrition analyses
Students who were surveyed in the 7 grade were compared to those students
who were lost to follow-up (from Grades 6 to 7) on all of the variables of interest.
Compared to those who were lost to follow-up, students who were surveyed in the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
70
7th grade more frequently reported receiving “mostly A’s” on their report cards
(39.7% vs. 29.9%, p < 0.05), nominated friends with lower rates of violent
perpetration (mean= 3.3 vs. 3.6, p < 0.001), were less hostile (mean= 2.4 vs. 2.6, p <
0.0001), and had fewer depressive symptoms (mean= 1.5 vs. 1.6). There were no
differences on gender, ethnicity, other social network variables (nominations sent,
received, reciprocated, and friends’ victimization), parental monitoring,
communication with parents, school bonding, and violence variables (perpetrator,
victim, and aggressive victim status).
Group identification
In general, students self-identified with multiple peer groups. The vast
majority (84.3%) identified with two or more peer groups, nearly two thirds (64.9%)
identified with three or more peer groups, and about half (49.7%) identified with four
or more peer groups. The most frequently self-identified group was “skaters”
(36.2%), followed by “smart kids” (34.2%), “jocks” (31.6%), “popular students”
(31.1%), “rockers” (19.7%), “gamers” (19.5%), “artists” (13.6%), “gangsters”
(11.2%), “religious students (7.4%), and “immigrants” (6.8%).
Exploratory factor analysis
Exploratory factor analysis was performed on the items that indicated
identification with each of the 10 peer crowds. Significant factor loadings are
presented in Table 5. Two factors emerged; one included self-identified “rockers”,
“skaters”, “gangsters”, and “immigrants,” and the other factor included self
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
71
identified “smart kids”, “artistic kids”, “religious kids”, “jocks”, “popular kids”, and
“gamers.” Sussman et al.’s (1999) high-risk group included “rockers” and
“gangsters.” Therefore the first factor was labeled “high-risk.” Similarly Sussman
et al.’s (1999) low-risk group included “jocks” “brains” (smart kids) and “popular
students.” Therefore the second factor was labeled “low-risk.”
Table 5
Factor Loadings of Peer Group S elf-Identification Items
Categories High-Risk Low-Risk
____________ Groups Groups
Rockers 0.70
Skaters 0.67
Gangsters 0.64
Immigrants3 0.45
Smart kids 0.68
Artistic kids 0.64
Religious kids 0.51
Jocks 0.51
Popular kids 0.49
Gamers______________________ 0.46
3 In pilot studies the category “Paisa” was identified as a peer group among this sample. Paisas refer
to newly arrived immigrants from Mexico.
Note. Factor loadings were derived from exploratory factor analysis, Principal Components, Varimax
rotation. Factor loadings of 0.45 and above were considered significant.
Creating discrete risk categories
Although exploratory analyses of the peer groups did generate two larger
factors, a substantial amount of respondents (40.7%) self-identified with at least one
group from each factor. Therefore three categories were created: (1) “high-risk”
(those who endorsed only high-risk groups), (2) “medium-risk” (those who endorsed
at least one high-risk group and at least one low-risk group), and (3) “low-risk”
(those who endorsed only low-risk groups). Comparisons were made between the
three categories on all variables of interest, which included gender, ethnicity, self-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
72
reported grades, lifetime smoking, nominations sent, received, and reciprocated,
depression, hostility, parental monitoring, communication with parents, and school
bonding. In all cases, comparisons between the three groups either: (1) failed to
demonstrate differences on the variable of interest, or (2) indicated that the medium
and high-risk groups differed significantly from the low-risk group. The only
exception was school bonding. The low-risk group had the highest rates of school
bonding (mean= 3.5), followed by medium-risk group (mean= 3.3), and the high-risk
group (mean= 3.2).
Since the high-risk and medium-risk groups were most similar across the
variables, they were collapsed into one category called “high-risk.” High-risk
students were those who self-identified with at least one of the high-risk groups:
“rockers”, “gangsters”, “skaters”, and “immigrants.” Low-risk students were those
who did not belong to any of these groups. A similar categorization scheme was
utilized by (Sussman, Stacy, et al. 1999) in a sample of continuation high school
students.
Differences between high-risk and low-risk students
Table 6 presents differences between high-risk and low-risk students on all of
the variables of interest. Clearly, several differences were found. High-risk students
were more often male (55.2% vs. 48.3%, p < 0.01), Latino (54.4% vs. 43.0%) and
less often Asian (13.6% vs. 27.7%, p < 0.0001). As predicted, high-risk students
had lower self-reported grades. They had fewer A’s (24.3% vs. 43.7%) and more
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
73
C’s, D’s, and F’s (41.2% vs. 21.7%, p < 0.0001) relative to low-risk students. High-
risk students were over twice as likely than low-risk students to have ever smoked
(21.7% vs. 8.5%, p < 0.0001). There were no differences between high-risk and
low-risk groups on nominations sent, received, and reciprocated.
Table 6
Differences Between 7th Grade High-Risk and Low-Risk Students
High
Risk
Students
Low
Risk
Students
n % n %
x 2
Demoeranhics
Gender
Male 864 55.2 590 48.3 13.0**
Female 701 44.8 631 51.7
Ethnicity
Latino 393 54.4 442 43.0
5 ***
Asian 98 13.6 285 27.7
Other 232 32.1 302 29.4
Self-reported Grades (6 grade)
Mostly A ’s 303 24.3 445 43.7 136.7***
Mostly B ’s 436 35.0 353 34.7
Mostly C ’s 369 30.0 178 17.5
Mostly D ’s 97 7.8 25 2.5
Mostly F’s 42 3.4 17 1.7
Health-Risk Behavior
Lifetime Smoking 337 21.7 147 8.5 1 1 2 7 ***
Mean SD Mean SD F
Network Nominations:
Sent 4.3 1.5 4.2 1 .6 0.3
Received 3.5 2.7 3.4 2 .6 0.4
Reciprocated 0 .6 0.3 0 .6 0.3 0 .6
Psvchosocial Variables
Depressive Symptoms 1 .6 0 .8 1.4 0.7 29.4***
Hostility 2.5 0 .8 2.4 0.7
19 4 ***
Adult Authority
Parental Monitoring 3.3 0.5 3.5 0.5
71 7 ***
Communication with Parents 2 .6 0.7 2 .8 0.7 41.8***
School bonding 3.3 0 .6 3.5 0.5 47.6***
*g < 0.05, **e< 0.01, ***£< 0.0001.
As predicted, high-risk students scored higher than low-risk students on
depressive symptoms (mean= 1.6 (SD= 0.8) vs. 1.4 (SD= 0.7), p < 0.0001) and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
74
hostility (mean= 2.5 (SD= 0.8) vs. 2.4 (SD^ 0.7), p < 0.0001). Furthermore high-
risk students scored lower than low-risk students on parental monitoring (mean= 3.3
(SD= 0.5) vs. 3.5 (SD= 0.5), p < 0.0001), communication with parents (mean= 2.6
(SD= 0.7) vs. 2.8 (SD= 0.7), p < 0.0001), and school bonding (mean= 3.3 (SD= 0.6)
vs. 3.5 (SD= 0.5), p< 0.0001).
Sixth grade violence and social network characteristics and 7th grade group
identification
Table 7 presents comparisons between 7th grade high-risk and low-risk
groups on violence and social network characteristics assessed one year earlier.
Table 7
Sixth Grade Characteristics by 7th Grade Group Identification
High Low
Risk Risk
Students Students
6 th Grade Variables n % n %
i
Violence
Perpetrator 56 7.2 64 5.7 1.9
Victim 146 18.8 265 23.4 5.8*
Aggressive Victim 373 49.4 469 41.4 8 .0 **
Friends’ Behavior: M SD M SD F
Perpetration 3.4 1.1 3.3 1 .0
jg 7***
Victimization 4.0 1.1 3.9 1.1 5.3*
Network Nominations:
Sent 4.3 1.5 4.2 1 .6 1.5
Received 4.5 2.7 4.2 2 .6 3.3
Reciprocated 0 .6 0.3 0 .6 0.3 1.4
*p < 0.05, **p < 0.01, ***p < 0.0001.
As predicted, high-risk students were more often aggressive victims one year
ago relative to low-risk students (49.4% vs. 41.4%; p < 0.01). Furthermore high-risk
students were less likely to be victims one year ago than low-risk students (18.8% vs.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
75
23.4%, p < 0.05). Thus, high-risk groups are likely to be aggressive victims, not just
victims only. The friends of high-risk students had higher self-reported rates of
perpetration (mean= 3.4 (SD= 1.1) vs. 3.3 (SD= 1.0); p < 0.0001), and victimization
(mean= 4.0 (SD= 1.1) vs. 3.9 (SD= 1.1); p < 0.05), relative to low-risk students.
There were no differences in the number of nominations sent and received and the
proportion of nominations reciprocated.
Logistic regression of high-risk status
All of the variables that were significantly associated with group
identification in the univariate analyses above were included in a logistic regression
model. This nested model predicted whether victimization and aggressive victim
status in the 6th grade were associated with group identification in the 7th grade after
adjusting for ethnicity, gender, and significant network variables. See Table 8.
Table 8
Logistic Regression of High-risk Status
AOR 95% Cl
Violent Status
Victim 0.80 0.58-1.10
Aggressive Victim 1.55** 1.16-2.06
Friends’ Behavior
Perpetration 1.13 0.98-1.30
Victimization 1.06 0.92-1.21
Covariates
Gender (Males vs. Females) 1.47** 1.14-1.90
Asians (vs. non-Asians) 0.51** 0.36-0.73
Latinos (vs. non-Latinos) 1.50* 1.07-2.09
*e <0.05, **£<0.001.
Self-reported perpetrator status was not included in the model since it was not
associated with group identification in univariate analyses. As predicted, aggressive
victim status in the 6th grade was associated with high-risk status in the 7th grade.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
76
This suggests that aggressive victim status, regardless of friends’ involvement in
violence and one’s position in the social network, is associated with later
involvement in high-risk groups. The finding does not infer causality, but it is
consistent with a causal model.
Group differences in nominating friends
The remaining analyses of Study 2 were conducted entirely on the subsample
of 1,199 students who completed an additional survey that included nominations of
th
their current five best friends in the 7 grade. Table 9 presents, for each group
separately, differences in the mean number of friends nominated from each peer
group. For example, did “smart kids” more frequently nominate other self-identified
“smart kids” than they did “jocks?”
Only 4 of the 10 peer groups nominated the greatest number or at least a
similar number of friends from their own group as opposed to other groups. These
groups included “immigrants”, “smart kids”, “jocks”, and “popular kids.” Some peer
groups were nominated more frequently than others as indicated by the column
totals. For example “jocks” and “popular students” were the most frequently
nominated groups by nearly all peer groups and across both risk categories.
Conversely “gamers” and “rockers” received relatively few nominations by all peer
groups and across both risk categories. The row totals were fairly similar, suggesting
that peer groups nominated an approximately equal number of friends.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
77
Table 9
Peer Group
Self-Identification
1. 2 . 3. 4. 5. 6 . 7. 8 . 9. 1 0 . 1 1 .
High-Risk
1. Rockers 0 .6b d 0.3d e 0.7b c 0 .8b c 0.4d 0.9a b 0.4c d 1.3a 0.9a b 0 .1 e 6.4
2. Skaters 0 .2 f 0.7c d O
oo
cr
l . l a b 0.9b c 0.7c d 0.4ef 1 .0 b c 1.5a 0.4^ 7.7
3. Gangsters 0.3ef 0.4d e 0.7b c 1.0 ab 0.5c d 0.7b c 0.3® l . l a 1 .0 a 0 .2 f 7.0
4. Immigrants1 0 .2 f 0.5d e 0.7b c 1.4 3 0 .8 b 0.7b c 0.5c d 1 .2 a 1 .3 a 0 .3 ef 7.6
Low-Risk
5. Smart 0.3° 0.5b c 0.5te 1.2 a 1 .1 a 0 .6 b 0.4te 1.4 3 1.3a 0.3C 7.6
6 . Artistic 0.3ef 0 .3 e 0 .8 c 0.9b c 0.5d l . l b 0.4d e 1.5a 0.9b c 0 .2 f 6.9
7. Religious 0.3d e 0 .2 e 0 .6 c d 1.3a 0 .6 c d 0.9a c 0.7b c 1.2 a 1 .0 a b 0.3d e 7.1
8 . Jocks 0.3ef 0.3e 0 .6 c d 0.9b 0.7b c 0.9 b 0.4d e 1.5a 0.9 b 0 .2 f 6.7
9. Popular 0.3f 0.5d e 0.7c d 1.2 b 0 .8 c 0.7c d 0.4ef l . l b 1.5a 0.3f 7.5
lO.Gamers 0 .2 f 0 .6 d e 0 .6 “ 1 .2 a b 0.9b d 0 .6 “ 0 .4 ef 1 .0 b c 1.5a 0.3ef 7.3
11. Total 3.0 4.3 6.7 1 1 .0 7.2 7.8 4.3 12.3 1 1 .8 2 .6
In pilot studies the category “Paisa” was identified as a peer group among this sample. Paisas refer
to newly arrived immigrants from Mexico.
Note. Analyses were conducted on a subsample o f 1,199 students who provided social network data
in the 7th grade.
Note. Superscript letters a through f across the rows are groupings derived from assessing means and 95%
confidence intervals.
Note. All Generalized Linear Models predicting friendships for each of the 10 peer groups were
significant (p < 0.05). Possible values ranged from 0 to 5.
Group differences in nominating high-risk friends
Next, group differences in the proportion of high-risk friends nominated were
assessed. For example, did self-reported “jocks” nominate more friends who were
high-risk? This was done by calculating the mean and 95% confidence intervals for
the mean proportion of high-risk friends nominated for each peer group separately,
as is shown in Table 10. Although there was a general trend for high-risk groups to
nominate a higher proportion of high-risk groups, no significant group differences
were found. All of the 10 peer groups were equally likely to nominate high-risk
friends.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
78
Finally, a t-test was calculated to determine whether high-risk students,
relative to low-risk students, nominated a higher proportion of high-risk friends. This
was found to be true. High-risk students nominated a higher proportion of friends
belonging to high-risk groups (0.40, 95% CI= 0.38-0.42) than did low-risk students
(0.32, SD= 0.29-0.34; p < 0.0001).
Table 10
Proportion of High-Risk Friends by Peer Group
Self-Identification Proportion 95% Confidence
Interval
Hieh-Risk Groups
Rockers 0.41 0.35-0.47
Skaters 0.41 0.36-0.45
Gangsters 0.43 0.39-0.47
Immigrants1 0.38 0.36-0.41
Low-Risk Groups
Smart 0.33 0.29-0.36
Artistic 0.34 0.31-0.37
Religious 0.35 0.30-0.39
Jocks 0.32 0.30-0.35
Popular 0.37 0.34-0.39
Gamers 0.39 0.33-0.44
1 In pilot studies the category “Paisa” was identified as a peer group among this sample. Paisas refer
to newly arrived immigrants from Mexico.
Note. Analyses were conducted on a subsample o f 1,199 students who provided social network data
in the 7th grade.
Note. No group differences were found.
Conclusions
This section draws some conclusions based specifically on the findings of
Study 2. General findings and limitations of this dissertation will be addressed in the
final chapter.
Study 2 provides valuable contribution to the literature in several different
ways. First, group identification was assessed with a mark-all-that applies rather
than a forced choice item. This method made it possible to determine whether
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
79
individuals tend to identify with specific “clusters” of groups, rather than one group
alone, and whether these clusters are similar to “high-risk” groups and “low-risk”
groups id e n tif ie d in previous literature on demographic, psychological, and
behavioral characteristics.
Second, previous research indicates that high-risk status is associated with
later violent perpetration and victimization. The present study also examines
whether aggressive victims, those who are both perpetrators and victims of violence,
are more likely to identify with high-risk groups, and whether involvement in
violence is associated with later high-risk group identification as the social-
interactional perspective suggests. Third, no known studies have determined
whether peer group self-identification corresponds with the peer groups that
nominated friends identify with. Study 2 tests exactly this.
Study 2 findings suggest that high-risk students, namely, self-identified
“rockers”, “skaters”, “gangsters”, and “immigrants” had similar characteristics to
high-risk students identified in previous studies. Specifically high-risk students more
frequently engaged in smoking, were more hostile and depressed, communicated less
and were monitored less by their parents, received lower grades, and were less
bonded to their school relative to low-risk students. As such, these findings validate
the methods used to classify high-risk and low-risk students in this study.
The univariate analyses also indicated that high-risk students did not differ
from low-risk students on the number of friendship nominations sent, received, and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
80
reciprocated in both the 6th and 7th grades. Furthermore the friends of high-risk
students were more frequently violent and victimized. This suggests that despite
their lower levels of psychosocial functioning, poorer parent-child relationships, and
school performance, high-risk students are not isolated by their classmates. They are
friends with students who are violent, and, to a lesser extent, victimized.
Study 2 indicated that 6th grade aggressive victims, relative to other students,
more often identified with high-risk groups in the 7th grade. This suggests that
aggressive victims may have a unique set of characteristics and experiences with
peers that predispose them towards high-risk status. Dodge (1983) observed early
elementary schoolboys in playgroups. He found that boys who used violence in the
form of insults, threats, excluding others from play, and physical aggression were
often rebuffed and socially isolated later. In a cross-sectional study of middle school
students, Haynie, Nansel, et al. (2001) found that aggressive victims, relative to
bullies-only and victims-only, were involved with more deviant friends. The
findings from this study and previous research are consistent with the notion that
aggressive victims identify with high-risk peers and/or identify with the image of
high-risk youth, possibly as an attempt to ward off their perpetrators and/or to
achieve higher social status by adopting a tougher image.
The univariate analyses indicated that friends’ level of aggression and
victimization in the 6th grade were associated with one’s own high-risk group
a f f ilia tio n in the 7th grade. However this was not the case in a model that included
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
81
one’s own involvement in violence, gender, and ethnicity. This suggests that one’s
own experiences with violence and individual characteristics more strongly predict
later identification with high-risk groups, as opposed to possible influences that
violent and victimized friends may have on the individual.
Study 2 found that only 4 of the 10 peer groups (e.g., “jocks”) nominated
students who self-identified with their own groups the most, or at least as much as
other peer groups. Since the vast majority of students self-identified with multiple
peer groups, it is possible that friendships are based upon matching group
identification for certain peer groups but not others. Conversely it could mean that
certain self-identified peer group members do not choose their friends based upon
matching peer group identification.
Seeing the larger picture, students categorized as high-risk and those
categorized as low-risk did not form mutually exclusive friendship groups. Rather,
over half (60%) of the friends of high-risk students were low-risk, and nearly one
third (32%) of the friends of low-risk students were high-risk.
These findings indicate that friendship groups are relatively heterogeneous in
regards to self-identification with specific peer groups as well as with high-risk and
low-risk categories. Nevertheless some peer groups befriend each other slightly
more than other peer groups, and high-risk students did have some more high-risk
friends compared to low-risk students. As such, one’s peer group self-identification
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
82
does not indicate well-defined friendship boundaries but does to some extent indicate
relative friendship preferences.
Finally, Study 2 found that certain peer groups, specifically self-identified
“popular students” and “jocks” received the most friendship nominations. This
finding is consistent with LaFontana and Cilessen’s (2002) finding that peer-
identified “popular” students have a larger number of social contacts relative to other
students. This suggests that “popular students” and “jocks” may be effective leaders
for school-based violence prevention programs, as they are frequently befriended
across peer groups.
Limitations
th
Since students who were surveyed in the 7 grade differed from those lost to
follow-up, the results may not generalize as well to students with low self-reported
grades, high levels of depression and hostility, and those whose friends have high
rates of perpetration. Second, both peer-group self-identification and violence
variables were not assessed at both years. The former was assessed only in the 7th
grade and the latter was assessed only in the 6th grade. Hence it was not possible to
determine whether experiences with violence actually led to increases in high-risk
self-identification, or whether violence and high-risk self-identification generally
occur simultaneously. Future studies should examine whether baseline aggressive
victimization as well as increases in aggressive victimization are associated with
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
83
identification with a high-risk group among those students who are low-risk at
baseline.
Last, although the mark-all-that-applies methodology may provide a more
accurate picture of peer group self-identification, it is difficult to determine the
characteristics of individual peer groups. For example high-risk and low-risk
students did not differ in the number of nominations sent, received, and reciprocated.
This may be because several high-risk students also identify with one or more
popular low-risk groups, or that some high-risk groups by themselves are popular
among other students. Future research should examine whether forced-choice
measures of peer group self-identification are associated with social network
characteristics.
Implications
In sum, Study 2 findings suggest that self-perceptions of group identification
are associated with real-life characteristics and behaviors. Adolescents who self-
identify with “rockers”, “skaters”, “gangsters”, and “immigrants” are at increased
risk of poor psychosocial functioning, poor academic performance, and health risk
behaviors such as smoking. This may be the result of poor parent-child
relationships, as high-risk students report less parental monitoring and
communication. Future studies should examine whether characteristics of parent-
child relationships during elementary school are associated with high-risk peer group
self-identification in adolescence in a prospective study. Such research would help
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
84
determine whether family-based interventions that train parents to better monitor and
communicate with their children may prevent high-risk group affiliation during
adolescence. Second, many self-identified high-risk adolescents may benefit from
clinical services to effectively cope with hostility and depressive symptoms and
possibly prevent future involvement in high-risk behaviors such as violence.
Sixth grade aggressive victims were more likely to identify with high-risk
peer groups the following year relative to other students. Why would this be so?
The social-interactional perspective (Eron & Slaby, 1994) suggests that violent
students who also have problems with victimization may be inclined to join deviant,
high-risk groups. These groups may consist of actual peers and/or self-identification
with a theoretical peer group. Study 1 demonstrated that during the 6th grade,
aggressive victims did not differ from other students on social status, but did have
more violent friends. Hektner, August, and Realmuto (2000) also found that
aggressive victims tend to befriend similar others by early adolescence.
If aggressive victims have comparable social status to other students, why do
they still get victimized? Previous research does demonstrate that relational and
physical forms of victimization do occur among preadolescent and adolescent
friendship pairs (Crick & Nelson, 2002). Aggressive victims may often be groups of
friends who victimize each other. Future research should examine whether this is
the case by assessing both friendship nominations and information about who is
violent against whom.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
85
Last, the finding that friendships are not highly embedded within peer groups
suggests that peer group self-identification may be more an autonomous choice as
opposed to an indicator of friendship selection or influence. Hence violence
prevention efforts should not target high-risk peer groups alone. Friendship groups
often contain both high-risk and low-risk students who may influence each other’s
behaviors. Specifically “popular students” and “jocks” may have strong influences
over other students regardless of group identification and may serve as ideal peer
leaders.
Although some overlap between peer group self-identification and one’s
nominated friends exist, both may be independently associated with violent
perpetration, victimization, and aggressive victim status. Study 3 tests exactly that.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
86
Chapter V: Study 3
Study 1 demonstrated in cross-sectional analyses of 6th graders that social
cognitive theory (Bandura, 2002) as opposed to dominance theory (Hawley, 1999)
best explains school violence among adolescents. Specifically perpetrators and
aggressive victims were friends with relatively violent students, and victims were
friends with less violent students. Study 2 was consistent with the social-
interactional perspective (Eron & Slaby, 1994) in that aggressive victims more often
self-identified with high-risk groups relative to other students. Last, Study 3 tests
three of the four theories embedded in the conceptual model described in Chapter 1.
These theories include social cognitive theory (Bandura, 2002), dominance theory
(Hawley, 1999), and the theory of secondary deviation (See Bynum & Thompson,
1989).
Studies 1 and 2 examined physical and direct verbal forms of violence.
However these overt behaviors tend to decrease in the later school years and a more
psychological form of violence begins to emerge (Petit, 1997). This violence is less
overt and involves the manipulation of relationships to harm another’s feelings or
reputation. Spreading rumors, an relational form of violence, was assessed in the 8th
grade in the longitudinal study. Therefore longitudinal associations between
predictor variables and rumor spreading violence will also be examined.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
87
Predictions
The predictions for Study 3 were the following:
1. Consistent with the findings of Study 1, other previous findings, (e.g.,
Espelage, Holt, and Henkel, 2003), and the conceptual model, I predicted that
violent perpetrator status in the 8th grade would be best explained by social
cognitive theory (Bandura, 2002). Specifically 8th grade violent perpetrator
status would be associated with high levels of friends’ perpetration and low
levels of friends’ victimization in the 6th grade.
2. The conceptual model posits that aggressive victimization is best explained
by the theory of secondary deviation (See Bynum & Thompson, 1989), which
posits that self-perceptions based upon past experiences with violence
influence subsequent participation in violence. Consistent with previous
findings (Sussman et al., 1999; Sussman et al., 2000; Sussman, Unger, &
Dent, 2004), I predicted that high-risk status in the 7th grade would be
associated with aggressive victim status in the 8th grade.
3. The conceptual model posits that victimization, especially during
adolescence, is best explained by dominance theory (Hawley, 1999).
Consistent with previous findings (e.g., Pellegrini et al., 1999; Boulton et al.,
1999; Hodges et al., 1999; Salmivalli, Lappalainen, & Lagerspetz, 1998), I
predicted that 8th grade victim status would be associated with fewer
nominations received and lower friendship reciprocity in the 6th grade.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
88
4. Finally, previous findings suggest that violent students utilize overt behaviors
such as hitting and threats less frequently, and utilize relational behaviors
such as spreading rumors more frequently (Petit, 1997). I attempted to link
earlier predictions of physical and verbal violence to that of rumor spreading
violence. Similar to the predictions of with that of physical and verbal
violence and consistent with theory of secondary deviation (See Bynum and
Thompson, 1989), I predicted that 7th high-risk self-identification would be
th
associated with aggressive victim status in the 8 grade. I did not predict that
low 6th grade social status would be associated with 8th grade rumor
spreading victim status since popular students are often victims of rumor
spreading (Prinstein & Cillessen, 2003). Rather I predicted that 7th grade
high-risk students would more likely be victims of rumor spreading in the 8th
grade because of their non-conventional image. Last, consistent with social
learning theory (Bandura, 2002), I did predict that friends’ level of physical
and verbal perpetration in the 6th grade would be associated with 8th grade
rumor spreading perpetrator status.
Sample
A total of 3,190 students completed the 6th grade survey, 2,822 students
completed the 7th grade survey, and 2,561 students completed the 8th grade survey.
A total of 2,292 students completed surveys in all three waves and comprised the
analytic sample in Study 3. The retention rate from the 6th to 8th grades was 71.8%.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
89
The analytic sample was 52.3% female and their mean age was 13.3 years (SD =
0.50) in the 8th grade, and was 60.2% Latino, 23.5% Asian, and 16.4% belonged to
other ethnic groups.
Analyses
Differences among violent subgroups. First, chi-square analyses and Generalized
th
Linear Models were calculated to determine whether 8 grade perpetrators, victims,
and aggressive victims of physical and verbal violence differed on gender and
ethnicity, 6th grade violence, 6th grade network characteristics, 7th grade peer group
self-identification, and the proportion of 7th grade friends self-identified with high-
risk groups.
Logistic regression of perpetration, victimization, and aggressive victim
status. Next I examined whether 6th grade social network characteristics and 7th
grade high-risk group identification were associated with 8th grade perpetrator,
victim, and aggressive victim status separately. Only those predictor variables that
were significantly associated with 8th grade violence in univariate analyses were
included in these models. Covariates included gender, ethnicity, and baseline levels
of violence. PROC GLIMMIX was used to control for similar rates of violence
within schools in a nested model. As mentioned before, it is particularly important
to account for variations in violence among schools since school characteristics may
strongly influence violence.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
90
The same univariate and logistic regression analyses were performed to
determine whether network characteristics and high-risk status were associated with
mm nr spreading violent perpetration, victimization, and aggressive victim status.
Results
Attrition analyses
Students who completed all three waves of the longitudinal study were compared
to those students who were lost to follow-up on all of the variables of interest.
Asians had a lower attrition rate (15.0%) relative to Latinos (26.1%; p < 0.0001).
Furthermore the analytic sample had a lower proportion of aggressive victims
(21.1%) relative to those with missing data (25.2%, p < 0.05). There were no
differences between the analytic sample and those lost to follow-up on perpetrator
status, and all social network variables.
Differences among violent subgroups
Table 11 presents differences between perpetrators, victims, aggressive
victims, and controls of physical and verbal violence on all variables of interest.
Several differences were found among groups. Aggressive victims and perpetrators
were more often male (61.2% and 55.5% respectively, p < 0.0001). A higher
proportion of perpetrators were Latino (70.2%) and a lower proportion was Asian
(15.9%) and other ethnic groups (13.9%, p < 0.0001).
Although changes in perpetration, victimization, and aggressive victim status
were observed from the 6th to the 8th grades, each of these violent subgroups were
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
91
more likely to remain in the same subgroup (e.g., victims remaining victims) rather
than transitioning to other subgroups (e.g., victims becoming aggressive victims, p <
0.0001). Hence I adjusted for 6th grade perpetration in the logistic regression model
predicting 8th grade perpetrator status, 6th grade victimization for the model
predicting 8th grade victim status, and 6th grade aggressive victimization for the
model predicting 8th grade victim status. This parsimonious method was chosen
because there was not enough power in the models to adjust for 6th grade
participation in all three forms of violence.
Eighth grade aggressive victims and perpetrators more often identified with
high-risk groups in the 7th grade (43.5% and 42.8%, respectively) relative to victims
and controls (31.8% and 24.2%, respectively). Furthermore the 6th grade friends of
8th grade perpetrators and aggressive victims were more violent (mean= 3.6 (SD=
1.1), mean= 3.4 (SD= 1.1), respectively) relative to the friends of victims (mean= 3.2
(SD= 1.0)). Last, 8th grade victims received the fewest friendship nominations in the
6th grade relative to all other violent subgroups (mean= 3.8 (SD- 2.4)). No
differences were found among 8th grade violent subgroups on friends’ level of
victimization and friendship nominations sent and reciprocated in the 6th grade, and
the proportion of 7th grade high-risk friends.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
92
Table 11
Differences among Physically and Verbally Violent Subgroups
Perpetrators Victims Aggressive
Victims
Controls
' 7 .........
n % n % n % n %
Demographics
Gender: Male 162 55.5 190 52.9 375 61.2 1392 49.7 28.2***
Ethnicity
Latino 177 70.2 149 47.9 284 53.7 854 63.8
4 9 3 ***
Asian 40 15.9 89 27.7 146 27.6 8 6 27.7
Other 99 13.9 76 24.4 99 18.7 76 24.4
6 t h Grade Violence
Perpetrator 25 18.7 7 4.2 15 4.7 73 5.7 92.1***
Victim 29 2 1 .6 51 30.5 75 23.5 256 19.8
Aggressive Victim 49 36.6 70 42.0 182 57.1 541 41.9
Control 31 23.1 39 23.4 47 14.7 421 32.6
7th Grade Hieh-Risk 125 42.8 114 31.8 267 43.5 1059 24.2 140.1***
Identification
Mean SD Mean SD Mean SD Mean SD F-value
6 th Grade
Friends’ Behavior:
Perpetration 3.6a 1.1 3.2 c 0.9 3.4 a b 1.1 3.3 b c 1.1 3.4*
Victimization 4.0 1.0 3.9 1 .0 4.0 1 .0 3.9 1.1 0.5
6 th Grade
Network Nominations:
Sent 4.3 1.5 4.2 1.5 4.3 1.5 4.2 1 .6 0.5
Received 4 .6 a 2.4 3.8b 2.4 4 .6 a 2 .8 4 .3 a 2 .6
3 9 **
Reciprocated 0 .6 0.3 0.5 0.3 0 .6 0.3 0 .6 0.3 1 .2
7th Grade Proportion 0.4 0.3 0.4 0 .2 0.4 0 .2 0.4 0.3 0 .6
o f hish-risk friends
*p < 0.05, **p < 0.01, ***p < 0.0001.
Note. Superscript letters across the rows are groupings derived from the Duncan test.
Logistic regression of perpetrators, victims, and aggressive victims
All of the variables that were significantly associated with 8th grade violence
in the univariate analyses presented above were included in three logistic regression
models. The nested models predicted whether 6th grade network characteristics and
friends’ behaviors, as well as 7th grade peer group self-identification, were associated
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
93
with 8th grade perpetration, victimization, and aggressive victim status, respectively.
tVi
The models adjusted for 6 grade outcomes, gender, and ethnicity.
Table 12 presents logistic regression of physical and verbal perpetrator,
victim, and aggressive victim status. As predicted and consistent with Study 1, level
of perpetration among one’s 6th grade friends was positively associated with
perpetrator status two years later (Adjusted odds ratio= 1.20, p < 0.01). As predicted
and consistent with Study 2 findings, high-risk status in the 7th grade was associated
with aggressive victim status one year later (Adjusted odds ratio= 1.40, p < 0.05).
As predicted, those with fewer friendship nominations in the 6th grade were more
likely to be victims two years later (Adjusted odds ratio= 0.92, p < 0.05).
Furthermore, level of perpetration among one’s 6th grade friends was not associated
with victim status two years later. This suggests that social status, as compared to
one’s friends, is more predictive of later victim status.
Table 12
Logistic Regression of Physical and Verbal Violence
8th Grade
Perpetration
AOR 95% Cl
8th Grade
Victimization
AOR 95% Cl
8th Grade
Aggressive
Victimization
AOR 95% Cl
6 th Grade
Network Nominations
Received 1 .0 0 0.93-1.07 0.92* 0.86-0.98 1.06* 1.0 1 - 1 .1 2
Friends’ Perpetration 1.2 0 * 1.02-1.42 0.87 0.73-1.05 1 .0 0 0.88-1.14
7th Grade
Groun Identification
High (vs. Low-risk) 1.31 0.89-1.93 1.14 0.80-1.62 1.40* 1.06-1.85
*£ < 0.05.
Note. Odds ratios adjusted for 6 th grade violence, ethnicity, gender, and frequency o f violence within
each student’s school.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
94
An unexpected result was found. The number of nominations received in the
6th grade was positively associated with aggressive victim status two years later
(Adjusted odds ratio= 1.06, p < 0.05), which may suggest that aggressive
victimization may be employed to increase social status. This implication of this
finding will be addressed in the discussion section of this chapter.
Difference scores on social network variables and group self-identification
The analyses presented above examined associations between 6th grade social
network variables and friends’ behaviors, 7th grade high-risk status, and 8th grade
violence. However friendship nominations were also assessed in the 7th grade and
high-risk status was also assessed in the 8th grade. Therefore I examined whether
difference scores on network nominations from the 6th to 7th grade were associated
with perpetrator, victim, and aggressive victim status. I also did the same for
changes in high-risk status from the 7th to 8th grade. None of these difference scores
were associated with the violence, by themselves and after adjusting for the first
assessment. Hence difference scores were excluded from the logistic regression
models presented above.
Perpetrators, victims, and aggressive victims of spreading rumors
Finally, logistic regression analyses of rumor spreading were performed to
determine whether 6th grade network characteristics and 7th grade high-risk status
were associated with 8th grade perpetrator, victim, and aggressive victim status.
First, univariate analyses were performed to determine whether the violent subgroups
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
95
differed on all of the variables of interest. Only 7th grade high-risk status was
associated with 8th grade violence. It was included in each of three logistic
regression models that were nested within schools and adjusted for gender and
ethnicity. High-risk groups in the 7th grade were more likely to be victims (Adjusted
odds ratio= 1.43, p < 0.01) and aggressive victims (Adjusted odds ratio= 1.51 »P<
0.01) relative to low-risk students. See Table 13.
Table 13
Logistic Regression of 8th Grade Rumor Spreading
Perpetration Victimization Aggressive
Year 3 Year 3 Victimization
Year 3
AOR 95% Cl AOR 95% Cl AOR 95% Cl
High (vs. Low-risk) 1.38 0.98-1.95 1.43** 1.10-1.87 1.51** 1.12-2.04
*p< 0.01.
Note. Odds ratios adjusted for 6* grade violence, ethnicity, gender, and frequency of violence within
each student’s school.
Which peer groups are more involved in rumor spreading?
Means and 95% confidence intervals were calculated for each peer group to
further explore which specific peer groups were more often victims and aggressive
victims of rumor spreading. “Popular students” (16.7%) were more often victims of
rumor spreading relative to “gangsters” (10.8%). No other group differences were
found. “Popular students” were also more often aggressive victims (14.0%) relative
to “immigrants” (7.1%). Again, no other group differences were found. To a much
lesser extent, these findings are consistent with that of Prinstein and Cillessen
(2003). They found that popularity was positively associated with reputational
victimization, classmates attempting to destroy one’s social reputation.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
96
Next, popular students were divided into two categories: those who also
belonged to at least one high-risk group and those who belonged only to low-risk
groups. Both categories of popular students were compared on rumor spreading
victimization and aggressive victim status. No differences were found, indicating
that high-risk and low-risk popular students were equally likely to be victims and
aggressive victims. The pattern of findings suggests that rumor spreading victim and
aggressive victim status is more strongly associated with high-risk status as opposed
to self-identification with specific groups.
Conclusions
This section draws some conclusions based on the findings of Study 3
integrated with those of Study 1. Other general findings and limitations of this
dissertation will be addressed in the final chapter.
Study 3 provides a unique contribution to the literature by examining the
relative influence of social network characteristics, friends’ behaviors, and high-risk
status on physical, verbal, and relational violence in a longitudinal design. No
known study of violence has included all of these predictor variables in one model.
Study 3 also expands upon Study 1 and 2 findings. Specifically Study 3 examines
whether the same associations found in Study 1, a cross-sectional study, can be also
observed in a longitudinal model. It also examines whether high-risk status, an
outcome of aggressive victim status in Study 2, also predicts aggressive victim status
one year later.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
97
The findings from Studies 1 and 3 indicate that friends’ level of perpetration
is both cross-sectionally and longitudinally associated with perpetrator status. Poulin
and Boivin (2000) suggest that similarities in violent behaviors among friends are the
result of friendship selection rather than friends influencing subsequent perpetration.
If the friendship selection model does apply to this sample, it may suggest that
although affiliations with specific friends may change, violent students generally
seek the friendship of other violent students. Hence friends’ level of perpetration
could be associated with one’s own perpetrator status across multiple years. On the
other hand the experience of associating with violent students for a given amount of
time may have lasting effects on one’s own perpetration, regardless of who he or she
befriends later.
Study 1 demonstrated that during the 6th grade, victims do not differ from
other students on social status, but do have friends who are less violent. Other
studies (Kochenderfer & Ladd, 1997; Hodges, Malone, & Perry, 1997) suggest that
the presence of violent friends may fulfill a protective function, especially among
males. This may be especially the case in the 6th grade, a period when violence is
more frequently used to establish social dominance among classmates during the
transition into middle school (Pellegrini & Long, 2002). Study 3, however, indicated
that a lower number of nominations received most strongly predicted victim status in
the 8th grade. The presence of violent friends may a lesser role in v ic tim iz a tio n
through adolescence as physical and verbal violence declines after the 6th grade
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
98
(Pellegrini & Long, 2002). In these later years, students who have been socially
isolated from their classmates over time may be more susceptible to victimization.
Pellegrini and Long (2002) suggest that perpetrators may prefer to target lower-status
students because they are be less likely to suffer from peer retribution and loss of
social reputation as a result.
In Study 1, aggressive victims, like perpetrators, had more violent friends and
did not differ from other students on social status. Study 3, however, indicated that a
higher number of nominations received in the 6th grade and high-risk status in the 7th
grade were associated with aggressive victim status in the 8th grade. Why is this? It
is possible that at earlier ages, one’s participation in violence is strongly determined
by the behaviors of actual friends who serve as models and reinforcers. By
adolescence, however, personal choices may factor into involvement in aggressive
victimization. The finding that high-risk group identification is associated with later
aggressive victimization after adjusting for friends’ level of violence supports
Sussman, Unger, and Dent’s (2004) assertion that high-risk group self-identification
may be an indicator of personal life choices.
Why would some students who have many friends become aggressive
victims? Pellegrini and Long (2002) suggest that some adolescents engage in
conflicts with each other to draw attention to them. For example males may like to
showcase their physical prowess and females may like to showcase their smarts in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
99
verbal conflict. Such conflict displays may be particularly important later in
adolescence as romantic relationships develop.
Last, Study 3 indicated that high-risk status in the 7th grade was positively
associated with rumor spreading aggressive victimization one year later. Being
talked about and talking about other people may be more common among those
students who identify with “gangsters”, “skaters”, “rockers”, and “immigrants”
because they do engage in violent behavior and are frequently victims of violence as
well. Owens, Shute, and Slee (2000) found that gaining attention and social
acceptance among classmates was a primary reason why 15-year old Australian girls
engaged in rumor spreading. This may also be the case among high-risk groups.
Furthermore they may be targets for rumor spreading possibly because of their less
conventional appearance, beliefs, and behaviors.
Limitations
Since students who were surveyed in the 7th grade differed from those lost to
follow-up, the results may not generalize as well to Latino adolescents and
aggressive victims. Second, friendship nominations were not assessed in the 8th
grade. Hence it is not known whether 8th grade perpetrators, victims, and aggressive
victims differed at that time on network characteristics and friends’ level of
perpetration and victimization. Since physical and verbal violence decreases through
adolescence, it is feasible that the network characteristics distinguishing each of
these three groups do change.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
100
Last, rumor spreading was not assessed in the 6th grade. Hence, it was not
possible to adjust for the e f f e c ts of 6th grade involvement in rumoring spreading in
logistic regression models predicting 8th grade involvement in rumor spreading.
Furthermore it was not possible to determine whether friends’ level rumor spreading
in the 6th grade was associated with rumor spreading violence in the 8th grade.
Nevertheless 6th physical and verbal violence were significantly associated with 8th
grade rumor spreading, and were used as covariates in these models.
Implications
Violence prevention programs in school are often comprehensive and may
include the participation of entire student body, several teachers, staff members, and
parents (e.g., Mellor, 1997; Roland & Munthe, 1997; Peterson & Rigby, 1999).
However the findings from Studies 1, 2, and 3 indicate that perpetrators, victims, and
aggressive victims each have unique characteristics that may predispose them
towards their current status. Furthermore these characteristics may change over
time. This suggests that violence prevention programs may be most effective if they
also include some activities that target perpetrators, victims, and aggressive victims.
Friends’ level of violence in the 6th grade was associated with perpetrator
status at that time and two years later. Dishion, McCord, and Poulin’s (1999)
deviancy training concept suggests that the repetition of contact among violent
students leads to subtle but powerful reinforcement for continued violent behaviors.
On the other hand it has been argued that substance use prevention programs are
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
101
most effective when students are grouped together with friends than with non-friends
during the program (See Valente, Gallaher, & Mouttapa, in press). Furthermore
grouping violent and non-violent students together may lead to violence initiation
among non-violent students. Hence it is unclear whether or not separating violent
students from each other may be an effective strategy to reduce violence, as Dishion,
McCord, and Poulin (1999) suggest. Future research is needed to determine whether
deviancy training for violence may occur among non-violent youth who are exposed
to violent students in an intervention program.
Sixth grade victims had less violent friends during that time. However fewer
nominations in the sixth grade was associated with victim status two years later.
Hence, violence prevention programs may best address the current victimization
problems among 6th graders with social skills training on how to handle violent
situations. Such training may help students defend themselves and their friends
effectively from potential perpetrators using nonviolent strategies. Furthermore
befriending solutions may also prevent later victimization. Bemdt (2002) suggests
that even having a few good friends may lead children to make positive contacts with
several other classmates and make them less vulnerable to victimization.
Friends’ level of violence in the 6th grade was associated with aggressive
victim status at that time. Hence, similar to perpetrators, separating violent students
from each other may be an effective strategy to prevent aggressive victimization in
the 6th grade. Two years later, however, a larger number of nominations in the 6th
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
102
grade and high-risk status were associated with aggressive victimization. Prevention
implications may include alternative strategies to aggressive victimization to increase
one’s social status. For example popular students are generally high on prosocial
characteristics including cooperation, sociability, and assertiveness (See Rodkin et
al, 2000). Such characteristics can be encouraged and reinforced in violence
prevention programs.
Furthermore Sussman, Unger, and Dent (2004) speculate that encouraging
students to not identify with high-risk peer groups may reduce violence. One
strategy may include teaching students that high-risk peer groups are images created
by corporate sponsors to make profit. Adopting this belief may lead students to look
unfavorably upon high-risk groups. No known studies have examined such
strategies to reduce high-risk peer group self-identification.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
103
Chapter VI: Discussion
This dissertation attempts to provide further insight into violence among
adolescents. Previous research suggests that the onset of serious violent offenses
begins as early as age 12 and rises sharply through age 16, but is nearly zero after
age 20 (Elliot, 1994). Petit (1997) states in his review that approximately half of all
aggressive children become aggressive adolescents and half of all aggressive
adolescents become aggressive adults. Although not all violent youth are violent in
adulthood, the vast majority of serious violent offenders engaged in minor forms of
delinquent and violent behaviors during their youth. Hence, examining what may be
relatively mild forms of violence during adolescence, including pushing, hitting,
teasing, verbal threats, and spreading rumors, is of great importance. Preventing
such types of violence at a young age may prevent violence later in life for a
substantial number of students.
Specifically it is important to examine school violence since schools provide
explicit and/or implicit norms for appropriate behavior, as well as the opportunity for
large numbers of same-aged peers to socialize with each other. Although causes of
violence may originate from sources outside of school, what happens in school may
largely determine one’s involvement in violence later in life (See Furlong, Paige, &
Osher, 2003).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
104
In this chapter I will briefly discuss the general implications of this
dissertation, implications for everyday school practices, limitations of this
dissertation, and future directions.
General implications
The findings of this dissertation supported the conceptual model described in
Chapter 1. Hence, this dissertation demonstrates that different theories best describe
perpetrator, victim, and aggressive victim status at school among adolescents. First,
perpetrator status is best described by social learning theory (Bandura, 2002). That
is, those students who are friends with perpetrators are more likely to be perpetrators
themselves. Hence, the stability of friendships with violent students may play the
strongest role in perpetration as opposed to social status or high-risk peer group self-
identification.
Next, victim status is best described with dominance theory (Hawley, 1999).
jL
Specifically friends’ level of perpetration was negatively associated with 6 grade
victimization, but fewer friendship nominations in the 6th grade were positively
associated with 8th grade victimization. Such findings suggest that violent students
th
and non-violent students remain in separate friendship groups in the 6 grade, and
therefore may leave some non-violent students vulnerable to victimization. However
the stability of low social status from the 6th to the 8th grades may most strongly
predict victimization, as opposed to the behaviors of one’s friends or high-risk peer
group self-identification.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
105
Last, aggressive victim status is best described by the theory of secondary
deviation (See Bynum & Thompson, 1989). Specifically high-risk peer group self-
identification in the 7th grade was associated with 8th grade aggressive victim status.
Furthermore the social-interactional perspective (Eron & Slaby, 1994) best describes
high-risk peer group self-identification. Such findings suggest that aggressive
victims identify with high-risk groups in an attempt to reduce victimization.
Identification with high-risk groups then leads to the stability of aggressive victim
status as opposed to the violent behaviors of one’s friends. The finding that the
number of friendship nominations received in the 6th grade was positively associated
with 8th grade aggressive victim status suggests that engaging in aggressive
victimization in 8th grade is a way of gaining the attention of other students and
maintaining one’s social status.
In general, the findings from this dissertation suggest that perpetrators,
victims, and aggressive victims should be examined separately, as each of them have
unique network characteristics and vary in their identification with high-risk groups.
Perpetration, victimization, and aggressive victimization may be most effectively
reduced if characteristics of each of these three groups are targeted in intervention
programs. For example, a prevention program that provides opportunities for
victims to increase their social connections with classmates and attempts to prevent
identification with high-risk peer groups may effectively reduce both victimization
and aggressive victimization. More research is needed to determine whether
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
106
reducing exposure to violent friends reduces perpetration among perpetrators but
increases perpetration among non-violent students.
Next I discuss how the findings of this dissertation may have implications for
everyday practices for teachers in the classroom.
Recommendations for everyday school practices
Throughout this dissertation I have discussed how the study findings may be
utilized in violence prevention programs. However such programs cannot be
implemented everywhere, as they do involve large amounts of materials, training,
time, and money. There is also no guarantee that short-term programs produce long-
lasting results. In light of this I present recommendations to teachers for everyday
practices based upon the dissertation findings. Such practices are intended not to be
laborious but still effective.
First, attempting to reduce the risk factors associated with high-risk status
may prevent future involvement in aggressive victimization. Hence, teachers should
be cognizant of their students’ moods and relationship to school life and implement a
plan if necessary. For example those students who appear hostile or depressed could
be referred to the school psychologist. Those students who have low grades and are
disconnected from school activities may benefit from a one-on-one meeting or
tutoring.
Integrating violent and non-violent students in workgroups may prevent
fixture perpetration and victimization for a few reasons. First, perpetrators would be
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
107
more distanced from other perpetrators who reinforce each other’s violent behaviors.
Second, they would be in the presence of other students who may not approve of
their violent behavior. Third, increased contact with victims in a structured setting
where violence is not tolerated requires perpetrators to familiarize themselves with
victims in a respectful manner. This increased familiarity may make perpetrators
less inclined to victimize students outside of the classroom. Last, victims would
have the opportunity to interact with friends outside of their friendship groups. Such
interactions may lead to increased social connections, provided that the non-violent,
non-victimized classmates in the workgroup are supportive.
Limitations
This section addresses a few of the limitations that pertain to the entire
dissertation. Strategies to address these limitations are discussed.
Limited generalizabilitv. First, since the study sample of this dissertation
consisted primarily of Latino and Asian adolescents, the findings may not generalize
as well to Caucasian adolescents. Future research is needed to determine whether
the associations found in this study are also found among a primarily Caucasian
sample. Future research is also needed to determine whether associations between
network characteristics, group self-identification, and school violence vary by
ethnicity or cultural values. Examining interaction terms of the predictor variables
with ethnicity, immigration status, and cultural values can do this. Since ethnicity
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
108
may be confounded with individual differences in socioeconomic status, it is also
important to adjust for socioeconomic status in such models.
Individual perpetrator-victim pairs were not identified. The measures
described in this dissertation measured the extent to which a student has been violent
and been victimized. They did not, however, assess the qualities of individual
relationships between pairs of students. For example, two students may report high
scores on violent perpetration and victimization measures because they are
aggressive with each other. Other students may report similar scores for different
reasons. For example a student may score high on perpetration and victimization
because he or she tends to be violent against one student but is victimized by
another. Without data on individual relationships, one cannot distinguish between
bullying and other forms of violence. Vermande, Van den Oord, Goudena, and
Rispens (2000) had kindergartners describe characteristics of their aggressors
without naming them. Such methods could possibly be used to assess characteristics
of adolescent perpetrator-victim pairs in cases where network data cannot be
collected.
Information about perpetrator-victim pairs would be particularly useful to
understand the social boundaries of violent behaviors. In this dissertation it is not
known whether aggressive victims are violent against some students and victimized
by others, or whether aggressive victims primarily victimize each other. Information
about perpetrator-victim pairs will also help us determine whether perpetrators and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
109
victims are also friends with each other, as well as self-identify with the same or
different peer groups. Such research will help determine whether violence
prevention programs should attempt to increase cohesiveness and cooperation within
friendship groups and peer groups, or to build bridges between separate friendship
groups and peer groups.
Collecting data on perpetrator-victim pairs would also provide insight into
whether individual or group attributes of other groups leads to violent perpetration.
For example Pitner, Astor, Benbenishty, Haj-Yahia, and Zeira (2003) found that
Arab and Jewish students were more likely to endorse more negative moral, social
conventional, and personal attributions about the other ethnic group compared to
their own ethnic group. Gathering data about who perpetrates against whom would
help identify whether negative attributions towards one group is associated with
violence towards those group members. The groups may consist of actual friendship
groups or self-perceived groups such as “jocks” and “smart kids.”
Friendship nominations were limited to the classroom. Friendship
nominations were limited only to those students in the particular class period that
they were surveyed in. Although many middle school students may share several
class periods, it is very possible that some of the respondents’ friends were in other
classes. Hence, measures of nominations sent and received in this dissertation may
underestimate respondents’ true social status within the school, and may not
accurately reflect their friends’ level of perpetration and victimization. Hence,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
110
associations between social network characteristics and violence may have been
attenuated. If possible, asking students to nominate the best friends within the
school, as opposed to the particular class period, may be an ideal method to assess
network characteristics. This is especially the case in middle school and high school,
when students attend classes in different classrooms.
The situational context of violent events was not assessed. Contextual
variations are rarely included in studies of school violence (Furlong & Morrison,
2000), as was not in this study. Some students may more frequently engage in either
proactive or reactive forms of violence. Some violent incidents may stem from
boyfriend/girlfriend disputes, whereas others may be the result of ethnic conflict.
Some students, especially boys, may have engaged in innocuous bouts of rough-and-
tumble play, and indicated on the survey that they did hit, push, or hurt another
student. Such behaviors are not, by definition, acts of violence since there was
mostly likely no intent to deliberately harm another student.
It is important to know the extent to which students engage in which forms of
violence, with whom, and for what reasons. For example, Furlong & Morrison
(2000) contend that institutional practices at school may adversely impact their
students and cause aggression and victimization. Such practices include
exclusionary practices, an overly competitive learning environment, toleration of
abuse, and discriminatory guidance policies. Violence arising from such practices
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I l l
may be more effectively prevented by addressing school policies than by addressing
individual students.
The situational context of violence can be assessed in different ways.
Espelage and Swearer (2003) suggest that videotaping and other observational
measures may be useful, as Coie, Cillessen, and colleagues (1999) did with
American 3rd graders. Observational measures, however, may not work as well with
adolescents, as they may be more subtle and inconspicuous with how they victimize
others. Another approach is to ask the respondent specific questions about why
specific instances of violence occur and the social circumstances, as Owens (2000)
did with high school girls. Such questions could be open-ended or forced-choice in
an interview or on a survey. Finally multiple informants, for example, the
individual, ones’ peers, and one’s teachers may provide unique perspectives on why
some students are perpetrators and other students are victims.
Last, it is not known whether parents and the school largely influence peer
group self-identification. For example it is highly feasible that “smart kids” are often
those students who are selected for advanced classes, and “jocks” are those students
whose parents place a high value on sports participation. Such findings may explain
ethnic and gender differences found among high-risk and low-risk groups. Future
research should examine whether identification with specific peer groups is
associated with participation in specific school activities as well as the attitudes,
beliefs, and behaviors of parents.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
112
Violence prior to adolescence was not assessed. The data utilized in this
dissertation did not include assessments of violent perpetration and victimization
th
prior to the 6 grade. Therefore violence prior to adolescence is not known in this
sample. Previous research indicates that violent preadolescent children are at
increased risk for violence as adults relative to those who begin violence during
adolescence. It is believed that late-onset violence is the result of issues that emerge
and eventually dissipate during adolescence (e.g., sexual maturation), whereas early-
onset violence represents the beginning of a lifelong persistent trajectory of violence
(Petit, 1997). Hence, future studies should attempt to gather data on violence prior to
adolescence, whether it is through earlier assessments, parental reports, or school
reports.
Future directions
Studies of social network characteristics, peer group self-identification, and
violence have provided insights into social factors that influence violence and have
recommended effective strategies for violence prevention programs. This section
provides further recommendations for future research.
The influence of popular students. In their review Hodges and Rodkin (2003)
suggest that popular students, assessed by “like most” nominations, have a
disproportionate amount of influence over the classroom as a whole. Although
popular students have lower rates of violence relative to other students (Dubow &
Cappas, 1988; Coie & Kupersmidt, 1993), whom they dislike may largely influence
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
113
who gets victimized in that classroom. In other words, those students who are
disliked by popular classmates may be more susceptible to victimization. Hence,
future research should examine whether victimization is more strongly influenced by
who dislikes them (e.g., self-identified “popular students” or those who received
many friendship nominations), relative to the number of people who dislike them.
Such research would help determine whether violence prevention programs should
attempt to forge amicable relationships between victimized students and popular
classmates.
Classroom centralization. Centralization is the extent to which the social
network, for example, a classroom, is organized around a few focal individuals
(Scott, 2000). Pellegrini and Bartini (2001) suggest that classroom levels of violence
decrease once positions of dominance are established, and, presumably, the
classroom social network is relatively centralized. No known studies have examined
associations between classroom centralization and violence. Such studies may find
that an optimal amount of centralization in friendship networks does exist such that
violence occurs less frequently. On the other hand, associations between centrality
and violence may largely depend upon the number of violent and non-violent
students occupy central network positions.
Network-based interventions. No known network-based violence
interventions have been implemented. In their review Valente, Gallaher, and
Mouttapa (in press) mention that there has been a long tradition of using peer
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
114
leaders, those frequently nominated as leaders, to assist in the delivery of substance
abuse prevention programs. Programs that have matched students with the leaders
they nominate have been more successful relative to those where students are
randomly assigned to peer leaders (See Valente, Gallaher, & Mouttapa, in press).
This might also be the case for violence prevention programs. As mentioned earlier,
it may be best to assign violent students to separate intervention groups to minimize
their influences over each other’s behavior (Dishion, McCord, & Poulin, 1999). It
also may be particularly promising to designate the peer leader role to peer-
nominated defenders of victims. Not only are defenders real-life role models of anti-
bullying behavior; they are also one of the most well liked students among
classmates (Salmivalli, Lagerspetz, Bjorkqvist, Osterman, & Kaukiainen, 1996).
Similarly, considering peer group self-identification for assigning students to
intervention groups may also be effective. Self-reported “popular students” and
“jocks” may be influential peer leaders since they receive several friendship
nominations across nearly all peer groups. Future research is needed to determine
whether separating students categorized as “high-risk” from each other during the
intervention leads to increases or decreases in school violence.
As previous research demonstrates, school violence is prevalent and is
associated with many adverse consequences. Violence prevention programs that
take the classroom social network and group identification into consideration may
effectively reduce violence and improve the quality of life among many students.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
115
References
Andreou, E. (2000). Bully/victim problems and their association with psychological
constructs in 8-12-year-old Greek schoolchildren. Aggressive Behavior. 26.
49-56.
Aronson, E. (1980). The Social Animal (3rd Ed.). San Francisco: W.H. Freeman and
Company.
Bagwell, C.L., Coie, J.D., Terry, R.A., & Lochman, J.E. (2000). Peer clique
participation and social status in preadolescence. Merrill-Palmer Quarterly.
46* 280-305.
Baldry, A., & Farrington, D. (1999). Types of bullying among Italian school
children. Journal of Adolescence, 22, 423-426.
Bauman, K.E., & Ennett, S.T. (1996). On the importance of peer influence for
adolescent drug use: Commonly neglected considerations. Addiction. 91.
185-198.
Bandura, A. (2002). Social cognitive theory in cultural context. Applied
Psychology: An International Review. 51, 269-290.
Barber, B.L., Eccles, J.S., & Stone, M.R. (2001). Whatever happened to the jock,
the brain, and the princess? Young adult pathways linked to adolescent
activity involvement and social identity. Journal of Adolescent Research. 16,
429-455.
Bemdt, T.J. (2002). Friendship quality and social development. Current Directions
in Psychological Science. 11, 7-10.
Bemdt, T.J., Hawkins, J.A., & Jiao Z. (1999). Influences on friends and friendships
on adjustment to junior high school. Merrill-Palmer Quarterly. 45. 13-41.
Bond, L., Carlin J.B., Thomas, L., Rubin, K., & Patton, G. (2001). Does bullying
cause emotional problems? A prospective study of young teenagers. British
Medical Journal. 323, 480-484.
Brown, B.B. (1996). Visibility, vulnerability, development, and context: Ingredients
for a fuller understanding of peer rejection in adolescence. Journal of Early
Adolescence, 16, 27-36.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
116
Brown, B.B. (1989). The role of peer groups in adolescents’ adjustment to
secondary school. In T.J. Bemdt & G.W. Ladd (Eds.), Peer relationships in
child development (pp. 188-215). New York: Wiley.
Boulton, M. J. (1995). Parents of bully/victim problems in mixed race groups of
children. Social Development. 4. 277-293.
Boulton, M. J., Tmeman, M., Chau, C., Whitehand, C., & Amatya, K. (1999).
Concurrent and longitudinal links between friendship and peer victimization:
implications for befriending interventions. Journal of Adolescence, 22. 461-
466.
Browning, C., Cohen, R., & Warman, D.M. (2003). Peer social competence and the
stability of victimization. Child Study Journal. 33. 73-89.
Buss, A. H., & Durkee, A. (1957). An inventory for assessment different kinds of
hostility. Journal of Counseling Psychology. 21. 343-349.
Bynum, J.E., & Thompson, W.E. (1989). Juvenile Delinquency: A sociological
approach. Needham, MA: Allyn & Bacon.
Caims, R.B., Neckerman, H.J., & Cairns, B.D. (1989). Social networks and the
shadows of synchrony. In G.R. Adams & R. Montemayor (Eds.), Biology of
adolescent behavior and development (pp. 275-305).
Centers for Disease Control and Prevention (2003). YRBSS: Youth online
comprehensive results. Retrieved June 12, 2004 from
http://apps.nccd.cdc.gov/yrbss/
Coie, J.D., Cillessen, A.H.N., Dodge, K.A., Hubbard, J.A., Schwartz, D., Lemerise,
E.A., & Bateman, H. (1999). It takes two to fight: A test of relational factors
and a method for assessing aggressive dyads. Developmental Psychology.
35, 1179-1188.
Coie, J.D., & Kupersmidt, J.B. (1993). A behavioral analysis of emerging social
status in boys’ groups. Child Development. 54. 1400-1416.
Connolly, J., Pepler, D., Craig, W., & Taradash, A. (2000). Dating experiences of
bullies in early adolescence. Child Maltreatment. 5. 299-310.
Crick, N., Casas, J., & Ku, H. (1999). Relational and physical forms of peer
victimization in preschool. Developmental Psychology. 35. 376-385.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
117
Crick, N.R., & Nelson, D.A. (2002). Relational and physical victimization within
friendships: Nobody told me there’d be friends like these. Journal of
Abnormal Child Psychology. 30. 599-607.
Dishion, T.J., McCord, J., & Poulin, F. (1999). When interventions harm: Peer
groups and problem behavior. American Psychologist. 9. 755-764.
Dodge, K.A. (1983). Behavioral antecedents of peer social status. Child
Development. 54,1386-1399.
Dubow, E.F., & Cappas, C.L. (1988). Peer social status and reports of children’s
adjustment by their teachers, by their peers, and by their self-ratings. The
Journal of School Psychology. 26. 69-75.
Dupper, D.R., & Meyer-Adams, N. (2002). Low-level violence: A neglected aspect
of school culture. Urban Education. 37. 350-364.
Elliot, D.S. (1994). Serious violent offenders: Onset, developmental course, and
termination- The American Society of Criminology 1993 Presidential
Address. Criminology. 32.1-21.
Eron, L.D., & Slaby, R.G. (1994). Reason to hope: A psychosocial perspective on
violence and youth. Washington, D.C.: American Psychological Association.
Espelage, D.A. (2002). Bullying in Early Adolescence: The Role of the Peer Group.
Retrieved February 12th, 2004 from
http://athealth.com/Consumer/disorders/Bullying%20in%20Adolescence.htm
1 .
Espelage, D.L., & Holt, M.K. (2001). Bullying and victimization during early
adolescence: Peer influences and psychosocial correlates. In R.A. Geffher,
M. Loring, et al. (Eds.), Bullying behavior: Current issues, research, and
interventions (pp. 123-142). New York: Haworth Maltreatment and Trauma
Press/The Haworth Press, Inc.
Espelage, D.A., Bosworth, K., & Simon, T.R. (2001). Short-term stability and
prospective correlates of bullying in middle-school students: An examination
of potential demographic, psychosocial, and environmental influences.
Violence and Victims. 16. 411-426.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
118
Espelage, D.L., Holt, M.K., & Henkel, R.R. (2003). Examination of peer-group
contextual effects on aggression during early adolescence. Child
Development. 74. 205-220.
Espelage, D.A., & 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. 365-383.
Evans, I.M., Heriot, S.A., & Friedman, A.G. (2002). A behavioural pattern of
irritability, hostility, and inhibited empathy in children. Clinical Child
Psychology and Psychiatry, 7, 211-224.
Espelage, D.L., & Holt, M.K. (2001). Bullying and victimization during early
adolescence: Peer influences and psychosocial correlates. In R.A. Geffher,
M. Loring, et al. (Eds.), Bullying behavior: Current issues, research, and
interventions (pp. 123-142). New York: Haworth Maltreatment and Trauma
Press/The Haworth Press, Inc.
Farmer, T.W., Leung, M-C, Rodkin, P.C., Cadwallader, T.W., Pearl, R., & Van
Acker, R. (2002). Deviant or diverse peer groups? The peer affiliations of
aggressive elementary students. Journal of Educational Psychology. 94. 611-
620.
Farmer, T.W., Estell, D.B., Bishop, J.L., O’Neal, K.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. 992-1004.
Furlong, M., & Morrison, G. (2000). The school in school violence: Definitions and
facts. Journal of Emotional and Behavioral Disorders, 8. 71-82.
Furlong, M., Morrison, G., Bates, M., & Chung, A. (1998). School violence
victimization among secondary students in California: Grade, gender, and
racial-ethnic group incidence patterns. The California School Psychologist.
3, 71-87.
Furlong, M.J., Morrison, G.M., & Greif, J.L. (2003). Reaching an American
Consensus: Reactions to the special issue on school bullying. School
Psychology Review. 32, 456-470.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
119
Furlong, M.J., Paige, L.Z., & Osher, D. (2003). The Safe Schools/Healthy Students
(SS/HS) Initiative: Lessons learned from implementing comprehensive youth
development programs. Psychology in the Schools. 40, 447-456.
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.
Gest, S.D., Graham-Bermann, S.A., & Hartup, W.W. (2001). Peer experience:
Common and unique features of number of friendships, social network
centrality, and sociometric status. Social Development. 10. 23-40.
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.
Hawkins, D.L., Pepler, D.J., & Craig, W.M. (2001). Naturalistic observations of
peer interventions in bullying. Social Development. 10. 512-527.
Hektner, J.M., August, G.J., & Realmuto, G.M. (2000). Patterns and temporal
changes in peer affiliation among aggressive and nonaggressive children
participating in a summer school program. Journal of Clinical Child
Psychology, 29. 603-614.
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.
Hawker, D.S.J., & Boulton M.J. (1999). Twenty years' research on peer
victimization and psychosocial maladjustment: A meta-analytic review of
cross-sectional studies. Journal of Child Psychology & Psychiatry & Allied
Disciplines. 41. 441-455.
Hawley, P. H. (1999). The ontogenesis of social dominance: A strategy-based
evolutionary perspective. Developmental Review. 19. 97-132.
Haynie, D.L., Nansel, T., Eitel, P., Crump, A.D., Saylor, K., Yu, K., & Simons-
Morton, B. (2001). Bullies, victims, and bully/victims: Distinct groups of at-
risk youth. Journal of Early Adolescence. 21. 29-49.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
120
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, 3 3 .1032-1039.
Hodges, E.V., & Perry, D.G. (1999). Personal and interpersonal antecedents and
consequences of victimization by peers. Journal of Personality and Social
Psychology. 76. 677-685.
Huttunen, A., Salmivalli, C., & Lagerspetz, K.M.J. (1996). Friendship networks and
bullying in schools. Annals of the New York Academy of Sciences. 794.
355-359.
Kasen, S., Cohen, P., & Brook, J.S. (1998). Adolescent school experiences and
dropout, adolescent pregnancy, and young adult deviant behavior. Journal of
Adolescent Research. 13. 49-72.
Kochenderfer-Ladd, B.J. (2003). Identification of aggressive and asocial victims and
the stability of their peer victimization. Merrill-Palmer Quarterly. 49.401-
425.
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
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
121
and withdrawn behavior, academic achievement and demographic
characteristics. Journal of Social and Personal Relationships. 12. 439-452.
LaFontana, K.M., & Cillessen, A.H.N. (2002). Children’s perceptions of popular
and unpopular peers: A multimethod assessment. Developmental
Psychology, 38. 635-647.
La Greca, A.M., Prinstein, M.J., & Fetter, M.D. (2001). Adolescent peer crowd
affiliation: Linkages with health-risk behaviors and close friendships.
Journal of Pediatric Psychology. 2 6 ,131-143.
Lewis, M., & Feiring, C. (1989). Early predictors of childhood friendship. In T.J.
Bemdt & G.W. Ladd (Eds), Peer Relationships in Child Development (pp.
246-273). Oxford, England: John Wiley & Sons.
Ma, X. (2002). Bullying in middle school: Individual and school characteristics of
victims and offenders. School Effectiveness and School Improvement, 13,
63-89.
Mellor, A. (1997). Bullying: The Scottish experience. Irish Journal of Psychology.
18, 248-257.
Menesini, E., Codecasa, E., Benelli, B., & Cowie, H. (2003). Enhancing children’s
responsibility to take action against bullying: Evaluation of a befriending
intervention in Italian middle schools. Aggressive Behavior, 29,1-14.
Merten, D.E. (1996). Information versus meaning: Toward a further understanding
of early adolescent rejection. Journal of Early Adolescence, 16, 37-45.
Michell, L. (1997). Loud, sad or bad: Young people’s perceptions of peer groups
and smoking. Health Education Research. 12, 1-14.
Mouttapa, M., Valente, T.W., Gallaher, P., Rohrbach, L., & Unger, J.B. (in press).
Social network predictors of bullying and victimization. Adolescence.
Olweus, W. (1991). Bully/victim problems among schoolchildren: basic facts and
effects of a school based intervention program (pp. 411-448). In D. Pepler &
K. Rubin (Eds.), The development and treatment of childhood aggression.
Hillsdale New York: Lawrence Erlbaum and Associates.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
122
Owens, L., Shute, R., & Slee, P. (2000). “Guess what I just heard!”: Indirect
aggression among teenage girls in Australia. Aggressive Behavior. 26. 67-
83.
Pakaslahti, L., & Keltikangas-Jarvinen, L. (2000). Comparison of peer, teacher, and
self-assessments on adolescent direct and indirect aggression. Educational
Psychology, 20.177-190.
Pakiz, B., Reinherz, H.Z., & Giaconia, R.M. (1997). Early risk factors for serious
antisocial behavior at age 21: A longitudinal community study. American
Journal of Orthopsychiatry. 67. 92-101.
Patterson, G.R., & Dishion, T.J. (1985). Contribution of families and peers to
delinquency. Criminology. 23. 63-79.
Pellegrini, 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., & Bartini, M. (2001). Dominance in early adolescent boys:
Affiliative and aggressive dimensions and possible functions. Merrill-Palmer
Quarterly, 47, 142-163.
Pellegrini, A.D., Bartini, M., & Brooks, F. (1999). School bullies, victims, and
aggressive victims: Factors relating to group affiliation and victimization in
early adolescence. Journal of Educational Psychology, 91,216-224.
Pellegrini, A.D., & Long, J.D. (2002). A longitudinal study of bullying, dominance,
and victimization during the transition from primary school through
secondary school. British Journal of Developmental Psychology. 20.259-
280.
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.
Peterson, L., & Rigby, K. (1999). Countering bullying at an Australian secondary
school with students as helpers. Journal of Adolescence, 22, 481-492.
Petit, G.S. (1997). The developmental course of violence and aggression:
Mechanisms of family and peer influence. Anger. Aggression, and Violence.
20,283-299.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
123
Poulin, F., & Boivin, M. (2000). The role of proactive and reactive aggression in the
formation and development of boys’ friendships. Developmental
Psychology. 36. 233-240.
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.
Prinstein, M.J., & Cillessen, A.H.N. (2003). Forms and functions of adolescent peer
aggression associated with high levels of peer status. Merrill-Palmer
Quarterly. 49. 310-341.
Prinz, R.J., Blechman, E.A., & Dumas, J.E. (1994). An evaluation of peer coping-
skills training for childhood aggression. Journal of Clinical Child
Psychology. 2 3 .193-203.
Pitner, R.O., Astor, R.A., Benbenishty, R., Haj-Yahia, M.M., & Zeira, A. (2003).
The effects of group stereotypes on adolescents’ reasoning about peer
retribution. Child Development. 74, 413-425.
Radloff, L. S. (1991). The use of the Center for Epidemiologic Studies Depression
Scale in adolescents and young adults. Journal of Youth and Adolescence.
20. 149-166.
Reiss, A.J., & Roth, J.A. (1993). Understanding and preventing violence.
Washington D.C.: National Academy Press.
Rigby, K. (1997). Attitudes and beliefs about bullying among Australian school
children. The Irish Journal of Psychology. 18. 202-220.
Rivers, I., & Smith, P. K. (1994). Types of bullying behaviour 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. 3 6 .14-24.
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. 384-400.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
124
Roland, E., & Idsoe, T. (2001). Aggression and bullying. Aggressive Behavior. 27.
446-462.
Roland, E., & Munthe, E. (1997). The 1996 Norwegian program for preventing and
managing bullying in schools. Irish Journal of Psychology. 18. 233-247.
Roth, D. A., Coles, M.E., & Heimberg, R.G. (2002). The relationship between
memories for childhood teasing and anxiety and depression in adulthood.
Journal of Anxiety Disorders. 16. 149-164.
Rowe, D. C., Almeida, D. M., & Jacobson, K. C. (1999). School context and genetic
influences on aggression in adolescence. Psychological Science. 10. 277-
280.
Rys, G.S., & Bear, G.G. (1997). Relational aggression and peer relations: Gender
and developmental issues. Merrill-Palmer Quarterly, 43. 87-106.
Salmivalli, C., Huttunen, A., & Lagerspetz, K.M.J. (1997). Peer networks and
bullying in schools. Scandinavian Journal of Psychology. 38. 305-312.
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., & Lagerspetz (1996). Bullying in schools: Main results of the
research project. In C.F. Ferris & T. Grisso (Eds.) Understanding aggressive
behavior. USA: Annals of the New York Academy of Sciences. 794. 401-
404.
Salmivalli, C., Lagerspetz, K., Bjoerkqvist, K., & Oesterman, K. (1996). Bullying as
a group process: Participant roles and their relations to social status within
the group. Aggressive Behavior. 22. 1-15.
Salmivalli, C., Lappalainen, & 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.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
125
Schwartz, D., Dodge, K.A., Pettit, G.P., & Bates, I.E. (1997). The early
socialization of aggressive victims of bullying. Child Development. 68. 66-
675.
Scott, J. (2000). Social network analysis: A handbook. (2n d ed.). London: Sage
Publications.
Shakib, S., Hoffman, B., Ortega, E., Mouttapa, M., Unger, J.B., Fosados, R., Fuqoa,
J., Sussman, S., & Ashmore, R. (in preparation). Peer-group affiliations and
substance use: A review of the literature.
Sharp, S., Thompson, D., & Arora, T. (2000). How long before it hurts? An
investigation into long-term bullying. School Psychology International. 21.
37-46.
Sussman, S., Simon, T.R., Stacy, A.W., Dent, C.W., Ritt, A., Kipke, M.D.,
Montgomery, S.B., Burton, D., & Flay, B.R. (1999). The association of
group self-identification and adolescent drug use in three samples varying in
risk. Journal of Applied Social Psychology, 29, 155-1581.
Sussman, S., Dent, C.W., & McCullar, W.J. (2000). Group self-identification as a
prospective predictor of drug use and violence in high-risk youth.
Psychology of Addictive Behaviors. 14. 192-196.
Sussman, S., Unger, J.B., & Dent, C.W. (2004). Peer group self-identification
among alternative high school youth: A predictor of their psychosocial
functioning five years later. International Journal of Clinical and Health
Psychology. 4, 9-25.
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.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
126
Urberg, K.A.., Degirmencioglu, S., Tolson, J.M., & Halliday-Scher, K. (1995). The
structure of adolescent peer networks. Developmental Psychology. 31. 540-
547.
Urberg, K.A., Degimencioglu, S.M., Tolson, J.M., & Halliday-Scher, K. (2000).
Adolescent social crowds: Measurement and relationship to friendships.
Journal of Adolescent Research. 15.427-445.
Valente, T.W., Gallaher, P., & Mouttapa, M. (in press). Using social networks to
understand and prevent substance use: A transdisciplinary perspective.
Substance Use and Misuse.
Vermande, M.M., Van den Oord, E.J.C.G., Goudena, P.P., & Rispens, J. (2000).
Structural characteristics of agressor-victim relationships in Dutch school
classes of 4- to 5-year olds. Aggressive Behavior. 26. 11-31.
Warman, D.M., & Cohen, R. (2000). Stability of aggressive behaviors and
children’s peer relationships. Aggressive Behavior, 26. 277-290.
Wasserman, S., & Faust, K. (1994). Social networks analysis: Methods and
applications. Cambridge, UK: Cambridge University Press.
Wolfinger, R. (1998). GLIMMIX: A SAS macro for fitting generalized linear mixed
models using PROC MIXED and the output delivery system (ODS). SAS
Institute, Inc.
Zeira, A., Astor, R.A., & Benbenishty, R. (2003). School violence in Israel:
Findings of a national survey. Social Work, 48, 471-483.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Abstract (if available)
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Aggression and victimization as predictors of drug use in early adolescence
PDF
Implicit cognition and dissociative experiences as predictors of adolescent substance use
PDF
Correlates of compliance with mammography screening guidelines among low-income Latinas: An exploratory study
PDF
Self -care diabetes behaviors among Latinos with type 2 diabetes
PDF
Correlates of smoking behaviour among Egyptian adolescents
PDF
The influence of future orientation on the relationship between condom use and violence among homeless/runaway youth: Implications for the prevention of HIV and violence
PDF
Patterns and risk factors for adolescent smoking progression
PDF
Adoption and implementation of evidence-based tobacco use prevention curricula and programming
PDF
Emotional intelligence and smoking risk factors in early adolescents
PDF
Alternatives to condom use and not sharing needles among young injection drug users, their sex partners, and the social network members
PDF
Adolescent exposure to alcohol advertising: A prospective extension of Strickland's model
PDF
Understanding visceral fat, subcutaneous abdominal fat, and total body fat in children: Longitudinal changes and relations to metabolic risks of type 2 diabetes
PDF
Contributions of interpersonal communication within a mass media environment: Exploring the role of social influence in a statewide tobacco control media campaign
PDF
The effect of potential stressors on early atherosclerosis: The Los Angeles Atherosclerosis Study
PDF
Cigarette smoking, metabolic syndrome, and subclinical atherosclerosis
PDF
Development and validation of the Group Supervision Impact Scale
PDF
Vocational rehabilitation among colorectal cancer survivors
PDF
Predictors of adolescent depressive symptomatology
PDF
Adaptation to a life -threatening diagnosis: Dispositional optimism and pessimism and posttraumatic growth among patients with HIV
PDF
Self-help and problem-proneness as predictors of weapon carrying among adolescents
Asset Metadata
Creator
Mouttapa, Michele
(author)
Core Title
Social network and group self-identification predictors of school violence
Degree
Doctor of Philosophy
Degree Program
Preventive Medicine
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
health sciences, public health,OAI-PMH Harvest,psychology, social
Language
English
Contributor
Digitized by ProQuest
(provenance)
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c16-411171
Unique identifier
UC11340362
Identifier
3145250.pdf (filename),usctheses-c16-411171 (legacy record id)
Legacy Identifier
3145250.pdf
Dmrecord
411171
Document Type
Dissertation
Rights
Mouttapa, Michele
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA
Tags
health sciences, public health
psychology, social