Joint Trajectories of Bullying and Peer Victimization Across Elementary and Middle School and Associations With Symptoms of Psychopathology

Abstract

The joint development of trajectories of bullying perpetration and peer victimization from Grade 5 to Grade 8 and concurrent and predictive associations with parent- and child-reported symptoms of psychopathology (anxiety, depression, attention-deficit/hyperactivity disorder, and somatization) were examined in a large sample (N = 695) of Canadian children. Dual trajectory modeling revealed four distinct subgroups of children: (a) those low in both bullying perpetration and peer victimization (low/limited involvement); (b) those with moderately increasing levels of involvement in bullying perpetration and low levels of victimization (bullies); (c) those with low levels of bullying perpetration and moderate/decreasing levels of peer victimization (initial/declining victims); and (d) a victim-to-bully group characterized by increasing bullying perpetration and moderate decreasing victimization. Conditional probability results suggest that a pathway from peer victimization to involvement in bullying is more likely than a pathway from bullying perpetration to peer victimization. Children classified in the victim-to-bully and initial/declining victim groups showed more pervasive elevations in parent- and child-reported symptoms of psychopathology across elementary and middle school and in Grade 9 than individuals with limited involvement in bullying or peer victimization. Most associations with Grade 9 parent- and child-reported symptoms of psychopathology remained even after controlling for initial symptoms of psychopathology. Results are discussed in the context of extant taxonomies of involvement in bullying, the temporal relationship between bullying and victimization, and the increased mental health risk associated with both pure victims and bully-victims. (PsycINFO Database Record (c) 2014 APA, all rights reserved).

Full text

Joint Trajectories of Bullying and Peer Victimization Across Elementary
and Middle School and Associations With Symptoms of Psychopathology
John D. Haltigan and Tracy Vaillancourt
University of Ottawa
The joint development of trajectories of bullying perpetration and peer victimization from Grade 5 to
Grade 8 and concurrent and predictive associations with parent- and child-reported symptoms of
psychopathology (anxiety, depression, attention-deficit/hyperactivity disorder, and somatization) were
examined in a large sample (N⫽695) of Canadian children. Dual trajectory modeling revealed four
distinct subgroups of children: (a) those low in both bullying perpetration and peer victimization
(low/limited involvement); (b) those with moderately increasing levels of involvement in bullying
perpetration and low levels of victimization (bullies); (c) those with low levels of bullying perpetration
and moderate/decreasing levels of peer victimization (initial/declining victims); and (d) a victim-to-bully
group characterized by increasing bullying perpetration and moderate decreasing victimization. Condi-
tional probability results suggest that a pathway from peer victimization to involvement in bullying is
more likely than a pathway from bullying perpetration to peer victimization. Children classified in the
victim-to-bully and initial/declining victim groups showed more pervasive elevations in parent- and
child-reported symptoms of psychopathology across elementary and middle school and in Grade 9 than
individuals with limited involvement in bullying or peer victimization. Most associations with Grade 9
parent- and child-reported symptoms of psychopathology remained even after controlling for initial
symptoms of psychopathology. Results are discussed in the context of extant taxonomies of involvement
in bullying, the temporal relationship between bullying and victimization, and the increased mental health
risk associated with both pure victims and bully–victims.
Keywords: bullying, peer victimization, trajectory modeling, symptoms of psychopathology
Bullying is aggressive behavior that is intentional, is repetitive
in nature, and involves an imbalance of power between the ag-
gressor and his or her target. At some point in their lives, many
children are involved in the perpetration of bullying or experience
bullying as a victim (Pepler, Jiang, Craig, & Connolly, 2008;
Vaillancourt et al., 2010). Indeed, bullying is a widespread prob-
lem. Most often, researchers have identified three distinct groups
or classes of individuals involved in bullying. Children who bully
others (i.e., pure bullies) are thought to represent approximately
5%–17% of school-age children (Jansen et al., 2012;Nansel et al.,
2004;Vaillancourt et al., 2010). Children who are victimized by
their peers (i.e., pure victims) constitute approximately 4%–12%
of school-age children (Nansel et al., 2004;Vaillancourt et al.,
2010). Finally, children who are characterized as bully–victims
(i.e., children who bully others but also get bullied) are thought to
represent between 4% and 13% of school-aged children (Jansen et
al., 2012;Nansel et al., 2004;Vaillancourt et al., 2010). Although
children who are victimized by their peers (pure victims) are at
heightened risk for experiencing mental health and adjustment
problems, bully–victims are considered to be at risk for the most
diverse range of adjustment problems, with symptoms of malad-
aptation associated with both bullies and victims (for reviews, see
Arseneault, Bowes, & Shakoor, 2010;Lereya, Samara, & Wolke,
2013).
Despite a growing number of longitudinal studies investigating
the developmental course and mental health functioning of both
bullies and victims (Arseneault et al., 2006;Barker, Arseneault,
Brendgen, Fontaine, & Maughan, 2008;Jansen et al., 2012;Pel-
legrini & Long, 2002;Pepler et al., 2008;Wolke, Copeland,
Angold, & Costello, 2013), there is a general lack of consensus in
the literature examining bullying and peer victimization regarding
an organizing conceptual or theoretical framework (although see
Pepler et al., 2006;Swearer, Espelage, Vaillancourt, & Hymel,
2010). In view of the likelihood that the developmental origins of
involvement in bullying and susceptibility to peer victimization are
diverse and operate in transaction with contextual and ecological
influences, a developmental psychopathology perspective may be
an especially promising framework from which to organize re-
search efforts aimed at understanding the development and course
of involvement in bullying and peer victimization.
Within a developmental psychopathology perspective, spe-
cial attention is given to individual differences in the develop-
mental origins, course, and correlates of both normal and ab-
normal developmental processes (Cicchetti, 1984;Sroufe,
This article was published Online First October 13, 2014.
John D. Haltigan and Tracy Vaillancourt, Counselling, Faculty of Edu-
cation, University of Ottawa, and School of Psychology, Faculty of Social
Sciences, University of Ottawa.
This study was supported by grants from the Canadian Institutes for
Heath Research. We thank Heather Brittain and Amanda Krygsman for
their coordination of the study.
Correspondence concerning this article should be addressed to Tracy Vail-
lancourt, Counselling, Faculty of Education, University of Ottawa, 145 Jean-
Jacques-Lussier, Ottawa, ON K1N 6N5, Canada. E-mail: tracy.vaillancourt@
uottawa.ca
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Developmental Psychology © 2014 American Psychological Association
2014, Vol. 50, No. 11, 2426–2436 0012-1649/14/$12.00 http://dx.doi.org/10.1037/a0038030
2426

2013). In addition, there is a sensitivity to the importance of
understanding ecological and contextual influences on children
and families (Cicchetti & Aber, 1998). Central to the develop-
mental psychopathology perspective are the concepts of equi-
finality, in which multiple pathways eventuate in single devel-
opmental outcome, and multifinality, in which a single
developmental input or pathway may precipitate a diverse set of
developmental outcomes (Cicchetti & Rogosch, 1996). Using
these concepts, key research questions can be more clearly
organized and systematically investigated. For example, the
concept of multifinality naturally organizes research aimed at
examining how pure and co-occurring or joint trajectories of
bullying and peer victimization are related to later developmen-
tal outcomes. Precisely, it provides a framework from which to
test the proposition that individuals with co-occurring involve-
ment in bullying and peer victimization (bully– victims) will, in
fact, show the most diverse array of elevated mental health
symptomatology relative to pure bullies or pure victims.
Many of the longitudinal studies investigating involvement in
bullying perpetration and victimization (e.g., Barker, Arse-
neault, et al., 2008;Pepler et al., 2008) have adopted a growth
mixture modeling or related semiparametric group-based mod-
eling approach to better understand heterogeneity in bullying
perpetration or victimization, its developmental origins and
correlates, and its growth over time (for a discussion of group-
based modeling approaches, see Nagin & Tremblay, 1999).
These group-based trajectory studies have generally found ev-
idence for three to five distinct developmental trajectories of
involvement in bullying and victimization across time, which
typically include a low/stable trajectory group and a high/
chronic trajectory group, along with other groups that are
defined by increased or decreased involvement over time (see
Barker, Boivin, et al., 2008;Biggs et al., 2010;Goldbaum,
Craig, Pepler, & Connolly, 2003;Pepler et al., 2008;Reijntjes
et al., 2013).
To our knowledge, only one published study has examined the
joint development of bullying perpetration and peer victimization
using a statistical group-based trajectory approach (although see
Arseneault et al., 2006;Haynie et al., 2001, and Swearer, Turner,
Givens, & Pollack, 2008 as examples of studies that have exam-
ined the joint development of bullying and victimization using a
priori grouping methods other than formal group-based trajectory
modeling techniques). Barker, Arseneault, et al. (2008) used
growth mixture modeling to identify clusters of individuals who
followed unique trajectories of bullying and victimization between
13 and 16 years of age. In addition to these trajectories, they also
modeled joint trajectories of bullying and victimization and exam-
ined joint trajectory groups (e.g., bully–victims) on delinquency
and self-harm in midadolescence (age 16). Barker, Arseneault, et
al. (2008) found that over time, peer victimization increased the
likelihood of bullying perpetration to a greater extent than bullying
perpetration increased the likelihood of peer victimization. Addi-
tionally, both boys and girls in the bully and bully-victim joint
trajectory groups were highest in midadolescent delinquency,
while both boys and girls in the bully–victim joint trajectory group
showed elevations in self-harm behavior (e.g., cutting or burning
oneself) at age 16.
The Current Study
Given that bully–victims are likely to carry putative risk factors
that are common to both bullies and victims, it is expected that
they are not only at heightened risk for a diverse range of mental
health problems but also show different emergent patterns over
time of involvement in bullying and experiences with peer victim-
ization. Indeed, evidence suggests that both bullying perpetration
and peer victimization show considerable association from one
year to the next, suggesting important developmental relationships
(Barker, Arseneault, et al., 2008). Key among the questions that
need to be better understood with respect to the bully–victim group
are whether (a) bully–victims were initially victimized which then
precipitated their involvement in bullying or (b) they were initial
bullies who subsequently experienced victimization due to retal-
iatory behavior on the part of their peers. Although some evidence
favors support of the first explanation (Barker, Arseneault, et al.,
2008) more work is needed to understand the developmental
course of the bully–victim group.
To this end, the current study sought to build upon the Barker,
Arseneault, et al. (2008) joint trajectory analyses of bullying and
victimization and investigated the co-occurrence between bullying
and victimization using a developmental psychopathology concep-
tual framework (Cicchetti, 1984;Sroufe & Rutter, 1984). Follow-
ing work in the bullying literature that has either drawn upon a
developmental psychopathology perspective implicitly with re-
spect to the development and problem of bullying (e.g., a
developmental-contextual view, Pepler et al., 2006; a social-
ecological model, Swearer, Espelage, Vaillancourt, & Hymel,
2010) or explicitly with respect to the impact of peer victimization
on later development (e.g., McDougall & Vaillancourt, 2014), we
examined developmental trajectories of bullying perpetration and
peer victimization as well as their dynamic co-occurrence over
time (i.e., joint trajectories; Nagin & Tremblay, 2001) across
middle childhood (Grades 5–8) using data drawn from a large
sample of Canadian children. In addition, given the heightened
mental health risk thought to be associated with the bully–victim
group, we examined relations of joint trajectory group membership
with concurrent measures of both parent- and child-reported symp-
toms of psychopathology including depression, anxiety, attention-
deficit/hyperactivity disorder (ADHD), and somatoform symptom-
atology. Last, heeding the point articulated by Barker, Arseneault,
et al. (2008) that continuity in mental health symptomatology itself
may account for developmental outcomes (Costello, Mustillo, Er-
kanli, Keeler, & Angold, 2003), we conducted prospective analy-
ses evaluating associations between joint trajectory group mem-
bership during the elementary and middle school years and
measures of parent- and child-reported symptoms of psychopathol-
ogy in the first year of high school (Grade 9) while controlling for
prior symptoms of psychopathology.
Method
Participants
Data were drawn from the McMaster Teen Study, an on-going
study examining the relations between bullying, mental health, and
academic achievement. The study began in the spring of 2008
(Grade 5, age 10; Time 1). Data collected annually from T1 (Grade
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2427
TRAJECTORIES OF BULLYING AND PEER VICTIMIZATION

5) to T5 (Grade 9) were available for the present study. Partici-
pants were initially recruited from a random sample of 51 schools
within a large Southern Ontario public school board. At Time 1, a
total of 875 participants were recruited (80% participation rate) to
take part in the longitudinal arm of the study. At each time point,
children completed self-report measures about their behavior, peer
experiences, and mental health functioning (described below).
Parents (87% mothers) were also interviewed about their child’s
behavior, peer experiences, and mental health functioning at each
time point. At T1, most of the parent reporters were over 40 years
of age (59%), with a median household income of $70,000–
$80,000, and well over half (74%) had postsecondary education.
The median income for the city from which participants were
recruited at Time 1 was $76, 222 and for the province it was $70,
910. At T1 child participants were on average 10.91 (SD ⫽0.36)
years of age and approximately half were girls (53%). Most
children (71%) were Caucasian; the rest identified as Middle
Eastern Canadian (2%); African/West Indian Canadian (Black;
3%); South Asian Canadian (3%); Asian Canadian (2%); Native
Canadian (1%); South/Latin American Canadian (1%); other (4%);
did not know (12%).
Data for the current report are based on 695 participants who
had reports of bullying and peer victimization for at least one time
point across T1–T4, thus permitting inclusion in bullying and
victimization trajectory analyses across the middle school period
(i.e., Grades 5–8; see below). The exclusion of T5 (Grade 9)
bullying and victimization data in trajectory modeling was done
for two reasons: (a) to limit trajectory analyses to the middle
school period; and (b) to allow for the ability to predict T5 parent-
and child-reported psychopathology independent of bullying and
victimization data (i.e., temporal priority).
Of the analytic sample, 91.5% (n⫽638) had complete bullying
and victimization data at T1, 86.4% (n⫽602) had complete data
at T2, 78.6% (n⫽548) had complete data at T3, and 73.0% (n⫽
509) had complete data at T4. Complete bullying and victimization
data at all four time points were available for 57.4% (n⫽400). We
conducted attrition analyses examining the analytic sample (N⫽
695) and the full sample recruited for the longitudinal arm of the
study at T1 (N⫽875) on child sex, maternal education, and
household income. Participants in the analytic sample were more
likely to have parents with higher levels of education, ␹
2
(4) ⫽
52.96, p⬍.01, and more likely to come from families with higher
levels of income, ␹
2
(7) ⫽50.10, p⬍.01. The full sample and the
analytic sample did not differ by child sex.
Procedures
Data were collected using paper/pencil surveys from children in
their classrooms at T1; in ensuing years (T2–T5), data were
collected in the privacy of children’s homes with the option of
completing either a paper/pencil or an online version. Parents were
interviewed over the telephone by a trained research assistant.
Parental consent and child assent were obtained at each time point.
The study has maintained consistent yearly approval status from
the pertinent university research ethics boards.
Measures
Measures for bullying perpetration and peer victimization are
presented first, followed by parent- and child-reported measures of
psychopathology. All available psychopathology data pertaining to
the measures used in the current report (see description below)
from T1 (Grade 5) through T5 (Grade 9) were used. Paralleling the
bullying/victimization trajectory data, psychopathology data be-
tween T1 and T4 (Grades 5 and 8) were averaged to provide
composite measures of these constructs during the elementary and
middle school period. Mental health symptomatology data from T5
(Grade 9) were used in separate predictive analyses that allowed us
to control for stability of psychopathology when assessing the
predictive significance of the bullying and victimization trajectory
groups for mental health symptomatology during the first year of
high school (T5, Grade 9).
Involvement in bullying perpetration and peer victimization.
An adaptation of the Olweus Bully/Victim Questionnaire (Olweus,
1996) was used to measure involvement in bullying and peer
victimization (Vaillancourt et al., 2010). Students were first pro-
vided with a standard definition of bullying that differentiated
bullying from fighting, aggression, and teasing, and then were
asked to respond to the question “Since the start of the school year
(September)...Howoften have you bullied/been bullied at
school?” using a 5-point scale (0 ⫽not at all, 1⫽only a few times
this year, 2⫽every month, 3⫽every week, 4⫽many times a
week). In addition to this general question, students were also
provided with examples of physical, verbal, social, and cyber
bullying/victimization and were asked to report about their expe-
riences both as perpetrator and victim along a 5-point frequency
scale identical to the one above. Responses across the five items
for both perpetrator and victim status were averaged within re-
spective time points. Higher scores indicate higher levels of bul-
lying perpetration and/or victimization. Internal reliability for the
bullying perpetration items was good at each time point (T1 ␣⫽
.77; T2 ␣⫽.71; T3 ␣⫽.72; T4 ␣⫽.78). Similarly, internal
reliability for the peer victimization items was good at each time
point (T1 ␣⫽.79; T2 ␣⫽.81; T3 ␣⫽.81; T4 ␣⫽.82).
Parent-reported symptoms of psychopathology. Parent re-
ported symptoms of psychopathology were assessed from T1 to T5
using the Brief Child and Family Phone Interview Version 3
(BCFPI-3; Cunningham, Pettingill, & Boyle, 2000) and the Behavior
Assessment Scale for Children (BASC–II; Reynolds & Kamphaus,
2004) parent-reported somatization subscale. The BCFPI is a 30-min
structured phone interview used to screen for emotional and behav-
ioral issues in children ages 3–18 years. The managing anxiety,
depression, and ADHD subscales of the BCFPI were selected for
examination so as to parallel child-reported measures of psychopa-
thology used in the current study (see below). Example statements on
these scales include “Do you notice that your child worries about
doing the wrong thing?” (anxiety), “Do you notice that your child is
unhappy, sad, or depressed?” (depression), and “Do you notice that
your child is distractible, has trouble sticking to an activity?”
(ADHD). Items are rated on a 3-point Likert scale ranging from
0⫽never to2⫽often. Responses are summed within each scale
(items reversed where necessary) to yield total scores.
The BASC–II is a multimethod, multidimensional measure of
behavior and self-perceptions of people between the ages of 2 and
25. Data on the somatization subscale of the parent-report form
were only collected from T3 to T5 as the adolescent form of the
parent-report BASC–II begins at age 12 (T3, Grade 7). The som-
atization subscale assesses the tendency of the child or adolescent
to be overly sensitive and complain about relatively minor physical
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2428 HALTIGAN AND VAILLANCOURT

problems or ailments, and to overreport the occurrence of various
physical complaints. It is composed of items that cover topics
including doctors’ visits and physical ailments (e.g., visits school
nurse, gets sick, has headaches, is afraid of getting sick, makes
frequent visits to the doctor).
Parent-reported BCFPI-3 data were averaged to provide com-
posite measures of parent-reported symptoms of anxiety, depres-
sion, and ADHD symptomatology across the elementary and mid-
dle school years (T1–T4). Internal consistency (Cronbach’s ␣)of
the BCFPI subscales across T1–T5 ranged from .81 to .86 (anxi-
ety), .83 to .87 (depression), and .83 to .85 (symptoms of ADHD).
Parent-reported BASC–II somatization data were standardized and
averaged across T3–T4 to provide a composite measure of parent-
reported somatization symptomatology in middle school. Internal
consistency (Cronbach’s ␣) of the BASC–II parent-reported som-
atization subscale across T3–T5 ranged from .73 to .75.
Child-reported symptoms of psychopathology. Child-reported
symptoms of anxiety, depression, and hyperactivity symptoms
were assessed from T1 to T5, and child-reported attention problem
and somatization symptomatology from T3 to T5, using the Self-
Report of Personality (SRP) form of the BASC–II (Reynolds &
Kamphaus, 2004). Child self-reported attention problem and so-
matization symptomatology were assessed only from T3 (age 12,
Grade 7) forward as these scales are included in the adolescent
version of the BASC–II only. For each scale, responses to items
were answered in both a 2 ⫽true or 0 ⫽false format, as well as
on a 4-point scale of frequency, ranging from 0 ⫽never to3⫽
almost always. Responses were reversed where appropriate and
summed to yield an overall score.
In the current study, the anxiety, depression, and hyperactivity
scale scores from Grade 5 to Grade 8 (T1–T4) were averaged to
provide composite measures of anxiety, depression and hyperac-
tivity symptomatology. Similarly, the attention problems and so-
matization scales were averaged across Grades 7 and 8 (T3–T4) to
provide a composite measure of attention problem and somatiza-
tion symptomatology. Last, the hyperactivity and attention prob-
lems composites (r⫽.68, p⬍.001) were themselves standardized
and averaged to yield a child-reported ADHD symptomatology
composite to parallel the ADHD subscale of the BCFPI. Internal
consistency of the BASC subscales across relevant time points
ranged from .86 to .91 (anxiety), .87 to .89 (depression), .80 to .82
(hyperactivity), .83 to .85 (attention problems), and .68 to .74
(somatization problems).
Results
Descriptive Statistics
Means and standard deviations of bullying/victimization across
the elementary and middle school years by sex are presented in
Table 1. Overall rates for bullying perpetration appear to show a
marginal increase, while overall rates for peer victimization
showed a clearer pattern of decline across the elementary and
middle school period. To ensure that these overall trends were
accurate, we examined linear change. For bullying perpetration,
both boys and girls did not show significant linear change in
bullying perpetration over time. In contrast, for peer victimization,
both boys (slope –.22, SE 0.05, z⫽–4.34, p⬍.001) and girls
(slope –.23, SE 0.05, z⫽–4.84, p⬍.001) showed significant
decreasing linear trends. Overall associations between consecutive
time points for both bullying perpetration (range r⫽.40–.56) and
peer victimization (range .52–.56) suggested relative stability in
personal involvement in bullying and victimization. The correla-
tions between bullying perpetration and peer victimization were
.22 (Grade 5), .36 (Grade 6), .40 (Grade 7), and .38 (Grade 8),
suggesting a tighter coupling of involvement in bullying and peer
victimization across the elementary and middle school period. All
of these correlations were statistically significant at p⬍.01.
Developmental Trajectories
Semiparametric group-based methods were used to identify the
number and shape of distinct trajectories of bullying perpetration
and peer victimization across the elementary school years using
four cycles of data (T1–T4). With this method, individual variation
over time is considered to be normally distributed within groups
which themselves have distinct growth patterns. Models were
estimated in Mplus 7.11(Muthén & Muthén, 2012) via latent class
growth analysis (LCGA). Estimation of trajectories proceeded in
two steps: (a) selecting the number of trajectory groups, followed
by (b) estimating the shape (slope) of each trajectory. Models were
first estimated independently for boys and girls to confirm that no
Table 1
Mean Levels of Bullying Perpetration and Peer Victimization Across Elementary and Middle School
Measure
Boys Girls Total
MSD N MSD N MSD N
Bullying perpetration
G5 .24 .39 294 .19 .29 345 .22 .34 639
G6 .23 .33 281 .20 .30 321 .21 .31 602
G7 .23 .32 251 .23 .35 297 .23 .33 548
G8 .30 .40 223 .27 .39 286 .28 .39 509
Peer victimization
G5 .77 .75 297 .91 .76 348 .84 .76 645
G6 .63 .70 281 .69 .68 321 .66 .69 602
G7 .49 .59 251 .67 .67 297 .59 .65 548
G8 .51 .63 223 .64 .64 286 .59 .64 509
Note. G⫽grade.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2429
TRAJECTORIES OF BULLYING AND PEER VICTIMIZATION

differences in trajectory shape by sex were identified (data avail-
able from the authors). As no differences were identified, data
were combined across boys and girls, and sex differences in the
proportion of the sample in each trajectory were calculated.
1
On the basis of prior work modeling the developmental trajec-
tories of bullying perpetration and peer victimization (Barker,
Boivin, et al., 2008;Pepler et al., 2008), a series of models was
fitted beginning with a one-trajectory model and moving to a
four-trajectory model. Evaluation of the best fitting models was
based on the following criteria: (a) the Bayesian information
criterion (BIC); (b) the Lo-Mendell-Rubin likelihood ratio test
(LMR–LRT); (c) entropy; (d) a conceptually clear model; and (e)
a model with a sufficient number of members in each group to be
able to examine group differences. The BIC is a commonly used fit
index in which lower values indicate a more parsimonious model.
LMR–LRT provides a k– 1 likelihood ratio-based method for
determining the ideal number of trajectories; a low pvalue (p⬍
.05) indicates that the ktrajectory model is a better fit to the data
compared to the k– 1 trajectory model. Finally, entropy is a
measure of classification accuracy with values closer to 1 indexing
greater precision (range 0–1).
Bullying perpetration. Although the BIC values indicated
that a four-group solution best fit the data, a two-group solution
was selected because the three- and four-group solutions did not
add substantially to the understanding of group patterns. In addi-
tion, entropy was highest for the two-group solution and the
LMR–LRT indicated that a two-group solution was superior to a
three group solution. The two-group solution and the fit indices for
the one- to four-group solutions are presented in the left panel of
Figure 1. In the two-group trajectory model, the majority of
children followed a consistently low/stable involvement in bully-
ing (87% of the sample; n⫽605; 284 boys, 321 girls) with the
remainder following a moderate/increasing pattern of bullying
(13%; n⫽90; 46 boys, 44 girls). Examination of posterior
probabilities indicated that children were well matched to their
group (.98 for the low/stable group and .91 for the moderate
increasing group).
Peer victimization. A two-group solution was also selected
from among one–four class trajectories of peer victimization.
Similar to the bullying perpetration models, the three- and four-
group solutions did not add substantially to the understanding of
group patterns, and both the LMR–LRT and entropy suggested that
a two-group solution was superior to a three-group solution. More-
over, the additional subgroups in the four-group solution consisted
of groups with 3% or less of the sample, which would not have
yielded enough children to make meaningful group comparisons.
The two-group solution and the fit indices for the one- to four-
group solutions are presented in the right panel of Figure 1.Inthe
two-group trajectory model, the majority of children followed a
consistently low/declining pattern of victimization (85.5% of the
sample; n⫽594; 292 boys, 302 girls), and the remainder followed
a moderate/declining pattern of victimization (14.5%; n⫽101; 38
boys, 63 girls). Examination of the posterior probabilities indicated
that children were well matched to their group (.98 for the low/
declining group and .93 in the moderate/declining group).
Joint trajectories of bullying perpetration and peer
victimization. The best fitting models for involvement in bully-
ing and victimization described above were used as the starting
point for the joint models. The joint trajectory analyses identified
four groups of children with distinct developmental patterns of
bullying and victimization. Group 1 represents children who were
classified in trajectories of low/stable bullying and low declining
victimization; these children, 73% of the sample (n⫽508; 244
boys, 264 girls), had little or no involvement in bullying as either
bullies or victims across the elementary and middle school years
(i.e., a limited involvement group). Group 2, the bullies, repre-
sented 11% of the sample (n⫽79; 45 boys, 34 girls) and were
characterized by trajectories of low/stable victimization and mod-
erate/increasing bullying. Group 3, the initial victimization chil-
dren, represented 10% of the sample (n⫽69; 31 boys, 38 girls)
and were characterized by trajectories of moderate declining vic-
timization and low/stable bullying. Finally, 6% of the sample (n⫽
39; 10 boys, 29 girls) were in Group 4, which was characterized by
those in moderate/declining victimization and moderate/increasing
bullying trajectories (i.e., those who transitioned from victim to
bullies; the “victim-to-bully” group).
Sex differences in proportion of sample in single and joint
trajectories trajectory. Sex differences in both the single and
joint trajectory groups were tested by comparing the proportions of
boys and girls in each trajectory group (vs. all others). Although
the shapes of the trajectories did not differ across sex, there were
differences in the proportion of boys and girls across victimization,
␹
2
(1) ⫽4.61, p⫽.032, and joint trajectory groups, ␹
2
(3) ⫽10.55,
p⫽.014. There were significantly more girls in the moderate
declining victimization group (9.9% girls vs. 5.5% boys) and in the
victim-to-bully joint trajectory group (4.2% girls vs. 1.4% boys),
while there were more boys in the bully joint trajectory group
(Group 2; 6.5% boys vs. 4.9% girls) based on examination of
standardized residuals in the differences in proportions across each
trajectory group.
Probabilities of bullying perpetration conditional on peer
victimization. Children in the low/stable and moderate/declin-
ing victimization trajectories were most likely to be classified in
the low stable bullying trajectory (probability ⫽.86 and .64,
respectively). Compared to the low victimization trajectory, chil-
dren in the moderate declining victimization trajectory (probability
.36 versus .14, OR 3.46, 95% CI [2.18, 5.47]) were more likely to
be classified in the moderate increasing bullying trajectory.
Probabilities of peer victimization conditional on bullying
perpetration. Children low/stable in bullying were most likely
to be classified in the low/declining victimization trajectory (prob-
ability ⫽.89) as were those moderate/increasing in bullying (prob-
ability ⫽.61). Compared to children in the moderate/increasing
bullying trajectory however, those in the low bullying trajectory
group were more likely to be classified in the moderate declining
victimization group (probability .39 versus .12, 4.69, 95% CI
[2.93, 7.42]).
1
Additional trajectory analyses inclusive of T5 (Grade 9) data were
conducted to explore whether there was evidence that bullying or victim-
ization trajectories would appear different in form with the inclusion of
these data (e.g., we might find evidence for an increase or spike in
victimization following the transition from middle school to high school).
These analyses yielded no evidence to suggest that victimization (or
bullying) trajectories would be altered with the inclusion of these data. In
fact, the best fitting overall model continued to be a two-group model in
which the same moderate/declining and low/stable victimization trajecto-
ries as reported in the current analyses were apparent.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2430 HALTIGAN AND VAILLANCOURT

Covariates Associated With Trajectory
Group Membership
Analyses of variance (ANOVA) procedures were used to examine
associations between trajectory group membership and parent- and
child-reported symptoms of psychopathology assessed concurrently
with the involvement in bullying perpetration and peer victimization
data used to create the trajectory groups (T1–T4). We initially eval-
uated models with participant sex and trajectory group, as well as their
interaction, as between-subjects factors. Because only one such sex
interaction emerged (Sex ⫻Trajectory group for parent-reported
somatization), we note in Table 3 that the effect of trajectory group on
parent-reported somatization was significantly larger for girls; all
model results presented below are those for a single factor (i.e.,
trajectory group) between-groups design.
To minimize significant results due merely to chance, in addition to
uncorrected pvalues, we report the Benjamini-Hochberg (BH; Ben-
jamini & Hochberg, 1995) correction for multiple testing (likewise,
below we interpret differences as significantly different if the BH-
corrected p⬍.05). In addition to providing tstatistics and their
associated BH-corrected pvalues in group comparison analyses, we
provide estimates of effect size (Cohen’s d). Cohen’s (1992) criteria
were adopted in interpreting d(small effect ⫽0.2, medium effect ⫽
0.5, and a large effect ⫽0.8). Prior to all substantive analyses, parent-
and child-reported outcome measures were standardized into zscores.
Concurrent bivariate analysis. The means and standard devi-
ations of parent- and child-reported symptoms of psychopathology
across the elementary and middle school years for each trajectory
group are presented in Table 2 and the results of contrast tests between
each of the trajectory groups and composite measures of parent- and
child-reported symptoms of psychopathology are presented in Table
3. During the elementary and middle school years, children classified
in the bullying, initial/declining victimization, or victim-to-bully
groups were distinguished from children in the limited involvement
group on 17 of 24 relevant contrasts (BH corrected), suggesting that
the groups characterized by bullying, peer victimization, or both
bullying and peer victimization experienced higher levels of parent-
and child-reported symptoms of psychopathology than did children in
the limited involvement group (effect sizes were generally large in
magnitude for significant contrasts). A notable exception to this
general pattern was that the bully group was indistinguishable from
the limited involvement group on parent-reported measures of symp-
toms of anxiety, depression, ADHD, and somatization. Empirically
defined members of the victim-to-bully group were also distinguished
from the bully group on four of eight indicators (BH corrected).
Specifically, those in the victim-to-bully group evidenced more ele-
vated levels of both parent- and child-reported depression and ADHD
symptomatology than those in the bully group (BH corrected; again,
effect sizes were large in magnitude). In contrast, only one of eight
possible contrasts was significant when comparing the victim-to-bully
group with children in the initial/declining victimization group: chil-
dren in the victim-to-bully group self-reported higher ADHD symp-
tomatology than did children in the initial/declining victimization
group.
Associations between group trajectories and symptoms of
psychopathology in high school (Grade 9). As with the con-
current analyses, children classified in classified in the bullying,
initial victimization, or victim-to-bully groups were distinguished
from children in the limited involvement group on 15 of 24
Figure 1. Left panel: Developmental trajectories of bullying perpetration across elementary and middle school
(G5–G8). Fit indices for one-trajectory model (Bayesian information criterion [BIC] ⫽1,628.34; entropy, n.a.;
Lo-Mendell-Rubin likelihood ratio test [LMR–LRT], n.a.), two-trajectory (BIC ⫽1,081.90; entropy ⫽.89;
LMR–LRT, p⬍.005), three-trajectory (BIC ⫽916.18; entropy ⫽.86; LMR–LRT, p⫽.44), four-trajectory
(BIC ⫽766.36; entropy ⫽.89; LMR–LRT, p⫽.36). Right panel: Developmental trajectories of peer
victimization across elementary and middle school (G5–G8). Fit indices for one-trajectory model (BIC ⫽
4,850.89; entropy, n.a.; LMR–LRT, n.a.), two-trajectory (BIC ⫽4,250.54; entropy ⫽.89; LMR–LRT, p⬍
.001), three-trajectory (BIC ⫽4,143.08; entropy ⫽.80; LMR–LRT, p⫽.18), four-trajectory (BIC ⫽4,036.02;
entropy ⫽.85; LMR–LRT, p⫽.06).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2431
TRAJECTORIES OF BULLYING AND PEER VICTIMIZATION

relevant contrasts (BH corrected), suggesting that each of these
groups experienced higher levels of parent- and child-reported
symptoms of psychopathology in the first year of high school than
did adolescents in the limited involvement group (effect sizes were
all moderate to large in magnitude; see Table 4). That said, and
again similar to concurrent analyses, adolescents in the bullying
subgroup only differed on two of eight contrasts from the limited
involvement group in Grade 9. Specifically, adolescents in the
bullying group evidenced higher levels of self-reported anxiety
and ADHD symptomatology than did those in the limited involve-
ment group. Also similar to analyses during the elementary school
years, the victim-to-bully group was also distinguished from the
bully group on two of eight indicators, with the victim-to-bully
group faring worse on measures of parent- and child-reported
depression than the bully group. Last, the victim-to-bully group
was indistinguishable from the initial victim group with none of
eight contrasts significant (BH corrected).
In a final set of ANOVA models, we repeated the above anal-
yses predicting Grade 9 (T5) parent- and child-reported symptoms
of psychopathology while controlling for initial symptoms of
psychopathology during middle school (i.e., the first time point at
which a given outcome measure was collected). Ten of 17 origi-
nally significant contrasts (unadjusted models, BH corrected) re-
mained significant at a nominal alpha level of .05 in models
controlling for initial symptoms of psychopathology. Specifically,
those in the initial victim group continued to show elevations in
both parent- and child-reported ADHD and depression symptoms,
and child-report anxiety symptoms relative to those in the limited
group, while individuals in the victim-to-bully group continued to
show elevations in child-reported anxiety, depression, and soma-
tization symptomatology relative to those in the limited group. In
addition, those in the bully group continued to show elevations in
child-reported anxiety relative to those in the limited group, and
those in the victim-to-bully group continued to show elevations in
child-reported depression symptomatology relative to those in the
bully group. Effect sizes for these effects remained moderate to
large in magnitude (mean d⫽0.68).
Discussion
The primary objective of the current study was to build upon the
existing work examining the joint development of bullying and
victimization trajectories and their associations with both concur-
rent and prospective symptoms of psychopathology within a de-
velopmental psychopathology framework. We extended the work
of Barker, Arseneault, et al. (2008) in two notable ways. First, we
modeled the joint development of peer victimization and involve-
ment in bullying across elementary and middle school period
(participant age range 9–12) when levels of bullying and victim-
ization have been shown to be higher (Arseneault et al., 2006;
Wolke, Woods, Bloomfield, & Karstadt, 2000) and may be less
stable and more dynamic (Biggs et al., 2010) than during the high
school years. Second, we examined associations between joint
trajectory group membership and both parent- and child-reported
measures of symptoms of psychopathology. In particular, we used
the developmental psychopathology concept of multifinality as a
guiding heuristic to test the assumption that individuals whose
joint growth patterns were characterized by both bullying and
victimization (i.e., bully–victims) would show the most diverse
array of elevated concurrent and prospective symptoms of psycho-
pathology.
Consistent with Barker, Arseneault, et al. (2008), we found two
distinct bullying trajectories consisting of a low/stable involve-
ment trajectory and a moderate increasing trajectory. Our peer
victimization data revealed two distinct trajectory groups consist-
ing of a low/declining trajectory and a moderate/declining trajec-
tory. That both victimization trajectories were characterized by
declining patterns of victimization is consistent with reported
general decreases in victimization as children mature (Smith, Mad-
sen, & Moody, 1999;Smith, Shu, & Madsen, 2001). However, we
did not find a moderate increasing trajectory of peer victimization
as did Barker, Arseneault, et al. (2008). One possibility for this
discrepancy is that the Barker, Arseneault, et al. (2008) study
assessed peer victimization across the ages of 13–16, a transitional
period between middle and high school, during which the relative
lack of stability in social hierarchies might be expected to lead to
Table 2
Joint Bullying-Victimization Trajectory Group Membership Means and Standard Deviations on Parent and Child-Reported Measures
of Psychopathology Across Elementary and Middle School
Measure
Limited
involvement
(n⫽508; 73%) Bully
(n⫽79; 11%) Victim
(n⫽69; 10%) Victim-to-bully
(n⫽39; 6%)
M SD M SD M SD M SD
Parent Report
BCFPI-3 Anxiety ⫺.03 0.84 ⫺.10 0.72 .09 0.89 .25 0.87
BCFPI-3 Depression ⫺.13 0.71 ⫺.09 0.69 .29 0.85 .51 1.05
BCFPI-3 ADHD ⫺.17 0.83 ⫺.03 0.88 .49 1.01 .52 0.91
BASC-II Somatization ⫺.09 0.85 ⫺.07 0.82 .37 1.16 .28 1.09
Child Report
BASC-II Anxiety ⫺.14 0.76 .26 0.74 .54 0.80 .44 0.60
BASC-II Depression ⫺.19 0.64 .17 0.76 .46 0.84 .76 0.92
BASC-II ADHD Composite ⫺.18 0.69 .54 0.87 .39 0.87 .72 0.75
BASC-II Somatization ⫺.13 0.83 .15 1.03 .56 1.13 .27 0.92
Note. As standardized scores, the means can also be interpreted as percentiles: Values of 0 are at the 50th percentile; positive scores are above the 50th
percentile; and negative scores are below the 50th percentile. Scores of 1 represent approximately the 85th percentile.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2432 HALTIGAN AND VAILLANCOURT

Table 3
Planned Comparisons on Parent- and Child-Reported Measures of Psychopathology in Elementary and Middle School With Empirically Derived (Semiparametric Group
Modeling) Groups
Measure
Trajectory group contrasts
Bully (n⫽79; 11%) vs.
Limited
(n⫽508; 73%)
Victim (n⫽69; 10%) vs.
Limited
(n⫽508; 73%)
Victim-to-bully (n⫽39;
6%) vs. Limited
(n⫽508; 73%)
Victim-to-bully (n⫽39;
6%) vs. Bully
(n⫽79; 11%)
Victim-to-bully (n⫽39;
6%) vs. Victim
(n⫽69; 10%)
B(SE)pdB(SE)pdB(SE)pdB(SE)pdB(SE)pd
Parent report
BCFPI-3 Anxiety ⫺.01 (.10) .953 0.01 .15 (.11) .171 0.18 .14 (.14) .315 0.17 .15 (.17) .375 0.19 ⫺.01 (.17) .959 0.01
BCFPI-3 Depression ⫺.07 (.10) .488 0.09 .39 (.11) ⬍.001
ⴱ
0.47 .51 (.14) ⬍.001
ⴱ
0.60 .58 (.16) ⬍.001
ⴱ
0.72 .12 (.17) .486 0.13
BCFPI-3 ADHD .10 (.11) .350 0.12 .61 (.12) ⬍.001
ⴱ
0.68 .63 (.15) ⬍.001
ⴱ
0.72 .53 (.18) ⬍.004
ⴱ
0.59 .02 (.18) .898 0.02
BASC-II Somatization .00 (.12) .997 0.00 .40 (.12) .001
ⴱ
0.44G.28 (.16) .078 0.32 .28 (.19) .138 0.32 ⫺.12 (19) .530 0.11
Child report
BASC-II Anxiety .38 (.10) ⬍.001
ⴱ
0.49 .80 (.10) ⬍.001
ⴱ
1.00 .69 (.13) ⬍.001
ⴱ
0.89 .31 (.16) ⬍.045 0.42 ⫺.11 (.16) .508 0.12
BASC-II Depression .28 (.09) .002
ⴱ
0.41 .83 (.10) ⬍.001
ⴱ
1.12 1.07 (.13) ⬍.001
ⴱ
1.47 .79 (.15) ⬍.001
ⴱ
0.92 .24 (.15) .111 0.24
BASC-II ADHD Composite .67 (.09) ⬍.001
ⴱ
0.89 .60 (.10) ⬍.001
ⴱ
0.79 1.03 (.13) ⬍.001
ⴱ
1.38 .35 (.15) .020
ⴱ
0.44 .43 (.16) .006
ⴱ
0.49
BASC-II Somatization .27 (.12) .021
ⴱ
0.32 .73 (.12) ⬍.001
ⴱ
0.83 .39 (.17) .020
ⴱ
0.47 .12 (.55) .546 0.12 ⫺.39 (.20) .085 0.32
Note. G indicates effect significantly larger for girls. Analysis of variance Fstatistics were significant for each measure except BCFPI-3 Anxiety. BCFPI-3 Depression: F(3, 684) ⫽8.85, p⬍.001;
BCFPI-3 ADHD: F(3, 684) ⫽13.93, p⬍.001; BASC-II parent-reported Somatization: F(3, 603) ⫽4.36, p⫽.005; BASC-II Anxiety: F(3, 691) ⫽29.66, p⬍.001; BASC-II Depression: F(3, 691) ⫽
44.02, p⬍.001; BASC-II ADHD Composite: F(3, 691) ⫽42.16, p⬍.001; BASC-II Somatization: F(3, 585) ⫽13.64, p⬍.001.
ⴱ
Significant at the Benjamini-Hochberg adjusted pvalue.
Table 4
Planned Comparisons on Parent and Child-Reported Measures of Psychopathology at Grade 9 (T5) With Empirically Derived (Semiparametric Group Modeling) Groups
Measure
Trajectory group contrast
Bully (n⫽79; 11%) vs.
Limited (n⫽508; 73%) Victim (n⫽69; 10%) vs.
Limited (n⫽508; 73%)
Victim-to-Bully (n⫽39;
6%) vs. Limited
(n⫽508; 73%)
Victim-to-Bully (n⫽39;
6%) vs. Bully
(n⫽79; 11%)
Victim-to-Bully (n⫽39;
6%) vs. Victim
(n⫽69; 10%)
B(SE) p d B(SE) p d B(SE) p d B(SE) p d B(SE) p d
Parent report
BCFPI-3 Anxiety ⫺.08 (.15) .566 0.08 ⫺.02 (.14) .871 0.02 .15 (.19) .433 0.15 .23 (.23) .306 0.25 .17 (.23) .446 0.18
BCFPI-3 Depression ⫺.02 (.14) .910 0.02 .45 (.14) .002
ⴱ
0.47 .58 (.19) .002
ⴱ
0.62 .60 (.22) .008
ⴱ
0.53 .13 (.22) .556 0.10
BCFPI-3 ADHD .19 (.14) .185 0.20 .58 (.14) ⬍.001
ⴱ
0.60 .57 (.19) .002
ⴱ
0.63 .38 (.22) .085 0.38 ⫺.01 (.22) .970 0.01
BASC-II Somatization .05 (.14) .727 0.05 .36 (.14) .013
ⴱ
0.36 .21 (.19) .278 0.22 .16 (.23) .493 0.15 ⫺.15 (.23) .502 0.14
Child report
BASC-II Anxiety .37 (.14) .008
ⴱ
0.39 .50 (.14) ⬍.001
ⴱ
0.51 .63 (.18) .001
ⴱ
0.67 .26 (.22) .239 0.27 .13 (.22) .557 0.13
BASC-II Depression .24 (.14) .091 0.26 .54 (.15) ⬍.001
ⴱ
0.56 .81 (.19) ⬍.001
ⴱ
0.88 .58 (.22) .009
ⴱ
0.59 .28 (.22) .217 0.22
BASC-II ADHD Composite .47 (.13) ⬍.001
ⴱ
0.54 .67 (.13) ⬍.001
ⴱ
0.79 .57 (.16) .001
ⴱ
0.68 .10 (.20) .611 0.10 ⫺.10 (.20) .629 0.11
BASC-II Somatization .25 (.14) .080 0.26 .40 (.14) .006
ⴱ
0.42 .60 (.19) .003
ⴱ
0.59 .31 (.22) .162 0.28 .16 (.22) .470 0.14
Note. Analysis of variance Fstatistics were significant for each measure except BCFPI-3 Anxiety and BASC-II parent-reported Somatization. BCFPI-3 Depression: F(3, 487) ⫽6.17, p⬍.001;
BCFPI-3 ADHD: F(3, 487 ⫽8.08, p⬍.001; BASC-II Anxiety: F(3, 484) ⫽8.52, p⬍.001; BASC-II Depression: F(3, 479 ⫽10.28, p⬍.001; BASC-II ADHD Composite; F(3, 484) ⫽14.68, p⬍
.001; BASC-II Somatization; F(3, 484 ⫽5.48, p⫽.001.
ⴱ
Significant at the Benjamini-Hochberg adjusted pvalue.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2433
TRAJECTORIES OF BULLYING AND PEER VICTIMIZATION

more dynamic patterns of bullying and victimization as children
attempt to establish their position in the social dominance hierar-
chy (Pellegrini & Long, 2002).
Although the moderate increasing bullying trajectory may seem
at odds with declines in both victimization trajectories, it may be
that with experience, children who were initially victims learned to
avoid, ignore, and retaliate against bullies (Smith et al., 2001).
Additionally, as children get older they tend to bully the same
targets. Specifically, in the early stages of group formation (i.e., at
the beginning of middle school), bullies are more likely to direct
negative and aggressive behavior toward a variety of targets. As
they learn the reactions of their victims, the field of victims then
becomes increasingly restricted (Kochenderfer & Ladd, 1996).
Our joint trajectory modeling identified distinct groups of bul-
lies, victims, and a pattern characterized by moderate/increasing
bullying and moderate/decreasing victimization (i.e., victim-to-
bully) in children. Our joint trajectory analysis also allowed us to
address whether victimization was more likely to increase the risk
for future involvement in bullying (e.g., a retaliatory perspective)
or if bullying was more likely to result in victimization. Similar to
Barker, Arseneault, et al.’s (2008) findings, our conditional prob-
ability results supported the former view, suggesting that at least
for a subgroup of children, being victimized leads to involvement
in bullying perpetration, perhaps as a way to defend themselves
and/or affiliate with other bullies in an effort to become a member
of the dominant social group (Pellegrini & Bartini, 2001; see also
Vaillancourt, Hymel, & McDougall, 2003).
We found consistent evidence that joint trajectory group mem-
bership was associated with both parent- and child-reported symp-
toms of psychopathology in theoretically expectable ways. Specif-
ically, those in the initial victim or victim-to-bully joint trajectory
groups fared worse on measures of parent- and child-reported
symptoms of psychopathology than those in the low involvement/
limited group. The only measure on which these groups did not
differ from the limited group was parent-reported anxiety symp-
tomatology. Somewhat similarly, those in the pure bully group
evidenced higher levels of self-reported symptoms of anxiety,
depression, and ADHD symptomatology than those in the limited
group. That said, bullies were not distinguishable from the limited
group on parent-reported measures of anxiety, depression, ADHD,
and somatization symptomatology. There are a number of potential
reasons why this may be the case. One more provocative possi-
bility, consistent with prior work (see Vaillancourt et al., 2003)is
that pure bullies are not as socially and emotionally maladjusted as
has been previously thought. An alternative possibility is that
parents of children who have demonstrated chronic patterns of
pure bullying are themselves less attuned to their children’s be-
havior, a possibility consistent with research showing parenting
deficits, particularly in parenting involvement and support, among
parents of children who bully (Haynie et al., 2001). Additional
research using multiple informants of children’s mental health
symptomatology, including teachers, will help clarify this issue.
Consistent with prior work (Arseneault et al., 2010;Nansel et
al., 2001;Wolke, Copeland, Angold, & Costello, 2013), absolute
mean levels of parent- and child-reported psychopathology were
generally highest in the victim-to-bully group in the current study.
However, in some contrast to prior work (e.g., Arseneault et al.,
2006;Haynie et al., 2001), we did not find that this group was
distinguishable from those in the initial victim group (i.e., pure
victims) on parent- and child-reported symptoms of psychopathol-
ogy, with the exception of elevated levels of child-reported ADHD
symptomatology across the middle school period. One possible
reason for these differences is that these earlier studies (Arseneault
et al., 2006;Haynie et al., 2001) did not use group-based modeling
techniques to form their bully/victim group classifications. Instead,
a priori decision rules were used, which do not take into account
between group differences in developmental patterns (i.e., trajec-
tories) of bullying and victimization. Another possibility is that the
victim-to-bully group may be unlikely to evidence greater mental
health symptomatology than the pure victims in the elementary
and middle school years. However, accumulated experience with
both peer victimization and bullying perpetration in conjunction
with the transition to high school and the biological and social
changes that occur in adolescence (e.g., puberty) may provide the
necessary conditions for the increased mental health risk associ-
ated with experiences of both bullying and being bullied to more
fully emerge (Pepler et al., 2006). Regardless of the potential
reasons for differences between studies regarding the long-term
outcomes of bully–victims, considered from the perspective of
multifinality, the current findings raise the possibility that it may
be the degree or severity of the experience of peer victimization
(e.g., chronic vs. episodic), rather than being a bully–victim per se,
that is associated with the most diverse range of elevated symp-
toms of psychopathology. This possibility is congruent with cross-
sectional analytic work suggesting that peer victimization is em-
pirically defined by severity rather than type (Nylund, Bellmore,
Nishina, & Graham, 2007;Wang, Iannotti, Luk, & Nansel, 2010).
Our findings regarding the predictive significance of trajectory
groups with respect to parent- and child-reported symptoms of
psychopathology in the first year of high school (Grade 9) largely
paralleled our concurrent results. Noteworthy to these predictive
analyses is that even after controlling for initial levels of parent-
and child-reported symptoms of psychopathology, most predictive
associations remained significant. In general, individuals in the
initial/declining victim and victim-to-bully trajectory groups con-
tinued to show elevations in parent- and child-reported symptoms
of psychopathology relative to those in the limited group. These
results converge with work that has shown unique effects of
victimization on mental health and social relationship functioning
after accounting for preexisting mental health difficulties (Arse-
neault et al., 2006;Wolke et al., 2013). Additionally, to our
knowledge, our results are the first to demonstrate unique effects
of peer victimization when using an empirically based framework
to form trajectory groups. Nevertheless, future work is necessary
to more definitively determine whether pure victim and bully–
victim trajectory classes (such as the victim-to-bully class in the
current study) can be discriminated on prospectively measured
mental health indicators. This undertaking will assist in illuminat-
ing whether these groups are actually statistical heuristics to aid in
classifying individuals at risk, or whether they empirically corre-
spond with truly distinct subpopulations (Bauer & Curran, 2003;
Muthén, 2004).
Limitations
There are a number of limitations to the current study. First, our
sample size, although still large, was smaller than previous work
that has examined trajectories of bullying and/or victimization,
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2434 HALTIGAN AND VAILLANCOURT

which may have reduced our ability to find additional trajectory
patterns of bullying and victimization during the elementary and
middle school period. Second, despite our multi-informant ap-
proach to measuring symptoms of psychopathology, our measures
of involvement in bullying (i.e., both bullying and peer victimiza-
tion) were limited to child-report questionnaires. Part of the more
robust association between children’s self-reported mental health
symptomatology and joint trajectory group membership could be
because children overstated and/or overgeneralized their experi-
ences with bullying and victimization, as well as their mental
health problems. Additional methods of assessing involvement in
bullying and peer victimization, such as direct observational,
teacher-report and peer-nomination methods (e.g., Pellegrini &
Long, 2002) were not available in the McMaster Teen Study.
Future multireporter and multimethod work will have the potential
to militate social desirability effects of child-report measures and
allow for separation between the methods used to measure trajec-
tory group indicators and those used to measure developmental
outcome, thus sidestepping the shared method variance limitation
of the current study.
In addition, our concurrent analyses did not allow for causal
predictions to be made with respect to group membership and
parent- and child-reported symptoms of psychopathology. For
example, it may be that internalizing symptomatology precedes
victimization (i.e., membership in the pure-victim group; see Arse-
neault et al., 2006). As Reijntjes et al. (2013) point out, while joint
trajectory studies afford the examination of the nature of dynamic
overlap of two constructs of interest over time, they do not permit
strong causal inference. Last, it is also important to point out that
our analytic strategy did not allow us to determine whether bully–
victim joint trajectory groups might account for any within-person
changes in mental health symptomatology over time.
Conclusion
The findings from the current study highlight the utility of a
developmental psychopathology framework in the study of bully-
ing perpetration and peer victimization and have important impli-
cations for both basic and applied research investigating the prob-
lem of bullying. As empirical evidence continues to mount
demonstrating a reactionary pattern in which victimization pre-
cedes involvement in bullying, the identification of children at risk
for victimization (i.e., the provocative victim) and the development
of intervention strategies designed to target them should begin
during the preschool and early elementary school years. Addition-
ally, research into compensatory and potentiating factors that ac-
count for the heterogeneity of peer victimization outcomes (Hanish
& Guerra, 2002) may be best organized via a multifinality frame-
work which underscores that there are multiple and diverse path-
ways leading from peer victimization to both adaptive and mal-
adaptive endpoints (McDougall & Vaillancourt, 2014).
Longitudinal research examining the developmental origins,
course, and correlates of involvement in bullying should also
continue to adopt more rigorous analytic approaches grounded in
a developmental psychopathology framework (see Fraley, Rois-
man, & Haltigan, 2013;Haltigan, Roisman, & Fraley, 2013 for an
example) that control for stability of psychopathology (as well as
other relevant third variables such as family hardship) in their
investigations of bullying and peer victimization. By doing so, the
field will be better positioned to address the crucial questions of
whether involvement in bullying and peer victimization have en-
during effects on mental health functioning into adulthood as well
as whether they represent a marker of present and later psychopa-
thology or are a causal factor in the longer term development of
mental health psychopathology.
References
Arseneault, L., Bowes, L., & Shakoor, S. (2010). Bullying victimization in
youths and mental health problems: “Much ado about nothing”? Psycho-
logical Medicine, 40, 717–729. doi:10.1017/2FS0033291709991383
Arseneault, L., Walsh, E., Trzesniewski, K., Newcombe, R., Caspi, A., &
Moffitt, T. E. (2006). Bullying victimization uniquely contributes to
adjustment problems in young children: A nationally representative
cohort study. Pediatrics, 118, 130–138. doi:10.1542/peds.2005-2388
Barker, E. D., Arseneault, L., Brendgen, M., Fontaine, N., & Maughan, B.
(2008). Joint development of bullying and victimization in adolescence:
Relations to delinquency and self-harm. Journal of the American Acad-
emy of Child and Adolescent Psychiatry, 47, 1030–1038. doi:10.1097/
CHI.ObO13e31817eec98
Barker, E. D., Boivin, M., Brendgen, M., Fontaine, N., Arseneault, L.,
Vitaro, F.,...Tremblay, R. E. (2008). The predictive validity and early
predictors of peer victimization trajectories in preschool. Archives of
General Psychiatry, 65, 1185–1192. doi:10.1001/archpsyc.65.10.1185
Bauer, D. J., & Curran, P. J. (2003). Distributional assumptions of growth
mixture models: Implications for over-extraction of latent trajectory
classes. Psychological Methods, 8, 338–363. doi:10.1037/1082-989X.8
.3.338
Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery
rate: A practical and powerful approach to multiple testing. Journal of
the Royal Statistical Society. Series B (Methodological), 57, 289–300.
Retrieved from http://onlinelibrary.wiley.com/journal/10.1111/
(ISSN)1467-9868
Biggs, B. K., Vernberg, E., Little, T. D., Dill, E. J., Fonagy, P., &
Twemlow, S. W. (2010). Peer victimization trajectories and their asso-
ciation with children’s affect in late elementary school. International
Journal of Behavioral Development, 34, 136–146. doi:10.1177/
0165025409348560
Cicchetti, D. (1984). The emergence of developmental psychopathology.
Child Development, 55, 1–7. doi:10.2307/1129830
Cicchetti, D., & Aber, J. L. (1998). Contextualism and developmental
psychopathology. Development and Psychopathology, 10, 137–141. doi:
10.1017/S0954579498001540
Cicchetti, D., & Rogosch, F. A. (1996). Equifinality and multifinality in
developmental psychopathology. Development and Psychopathology, 8,
597–600. doi:10.1017/S0954579400007318
Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155–159.
doi:10.1037/0033-2909.112.1.155
Costello, E. J., Mustillo, S., Erkanli, A., Keeler, G., & Angold, A. (2003).
Prevalence and development of psychiatric disorders in childhood and
adolescence. Archives of General Psychiatry, 30, 837–844. doi:10.1001/
archpsyc.60.8.837
Cunningham, C. E., Pettingill, P., & Boyle, M. H. (2000). The Brief Child
and Family Phone Interview (BCFPI). Hamilton, Canada: Canadian
Centre for Studies of Children at Risk.
Fraley, R. C., Roisman, G. I., & Haltigan, J. D. (2013). The legacy of early
experiences in development: Formalizing alternative models of how
early experiences are carried forward over time. Developmental Psy-
chology, 49, 109–126. doi:10.1037/a0027852
Goldbaum, S., Craig, W. M., Pepler, D., & Connolly, J. (2003). Develop-
mental trajectories of victimization: Identifying risk and protective fac-
tors. Journal of Applied School Psychology, 19, 139–156. doi:10.1300/
J008v19n02_09
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2435
TRAJECTORIES OF BULLYING AND PEER VICTIMIZATION

Haltigan, J. D., Roisman, G. I., & Fraley, R. C. (2013). The predictive
significance of early caregiving experiences for symptoms of psycho-
pathology through midadolesence: Enduring or transient effects? Devel-
opment and Psychopathology, 25, 209–221. doi:10.1017/
S0954579412000260
Hanish, L. D., & Guerra, N. G. (2002). A longitudinal analysis of patterns
of adjustment following peer victimization. Development and Psycho-
pathology, 14, 69–89. doi:10.1017/S0954579402001049
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.
doi:10.1177/0272431601021001002
Jansen, P. W., Verlinden, M., Dommisse-van Berkel, A., Mieloo, C., van
der Ende, J., Veenstra, R.,...Tiemeier, H. (2012). Prevalence of
bullying and victimization among children in early elementary school:
Do family and school neighbourhood socioeconomic status matter?
BMC Public Health, 12.doi:10.1186/1471-2458-12-494
Kochenderfer, B. J., & Ladd, G. W. (1996). Peer victimization: Cause or
consequence of school maladjustment. Child Development, 67, 1305–
1317. doi:10.2307/1131701
Lereya, S., Samara, M., & Wolke, D. (2013). Parenting behavior and the
risk of becoming a victim and a bully/victim: A meta-analysis study.
Child Abuse & Neglect, 37, 1091–1108. doi:10.1016/j.chiabu.2013.03
.001
McDougall, P., & Vaillancourt, T. (2014). Long-term outcomes of peer
victimization in school: Pathways to adjustment and maladjustment.
Manuscript submitted for publication.
Muthén, B. (2004). Latent variable analysis: Growth mixture modeling and
related techniques for longitudinal data. In D. Kaplan (Ed.), The Sage
handbook of quantitative methodology for the social sciences (pp. 345–
368). New York, NY: Sage. doi:10.4135/9781412986311.n19
Muthén, K., & Muthén, B. O. (1998–2012). Mplus user’s guide (7th ed.).
Los Angeles, CA: Múthen & Múthen.
Nagin, D. S., & Tremblay, R. E. (1999). Trajectories of boys’ physical
aggression, opposition, and hyperactivity on the path to physically
violent and non-violent delinquency. Child Development, 70, 1181–
1196. doi:10.1111/1467-8624.00086
Nagin, D. S., & Tremblay, R. E. (2001). Analyzing developmental trajec-
tories of distinct but related behaviors: A group-based method. Psycho-
logical Methods, 6, 18–34. doi:10.1037/1082-989X.6.1.18
Nansel, T. R., Craig, W., Overpeck, M. D., Saluja, G., Ruan, W. J., &
Health Behaviour in School-aged Children Bullying Analyses Working
Group. (2004). Cross-national consistency in the relationship between
bullying behaviors and psychosocial adjustment. Archives of Pediatric
and Adolescent Medicine, 158, 730–736. doi:10.1001/archpedi.158.8
.730
Nansel, T. R., Overpeck, M., Pilla, R. S., Ruan, W. J., Simons-Morton, B.,
& Scheidt, P. (2001). Bullying behaviors among US youth: Prevalence
and association with psychosocial adjustment. Journal of the American
Medical Association, 285, 2094–2100. doi:10.1001/jama.285.16.2094
Nylund, K., Bellmore, A., Nishina, A., & Graham, S. (2007). Subtypes,
severity, and structural stability of peer victimization: What does latent
class analysis say? Child Development, 78, 1706–1722. doi:10.1111/j
.1467-8624.2007.01097.x
Olweus, D. (1996). The Revised Olweus Bully/Victim Questionnaire. Ber-
gen, Norway: Research Center for Health Promotion, University of
Bergen.
Pellegrini, A. D., & Bartini, M. (2001). Dominance in early adolescent
boys: Affiliative and aggressive dimensions and possible functions.
Merrill-Palmer Quarterly, 47, 142–163. doi:10.1353/mpq.2001.0004
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 Psychol-
ogy, 20, 259–289. doi:10.1348/2F026151002166442
Pepler, D. J., Craig, W. M., Connolly, J., Yuile, A., McMaster, L., & Jiang,
D. (2006). A developmental perspective on bullying. Aggressive Behav-
ior, 32, 376–384. doi:10.1002/ab.20136
Pepler, D., Jiang, D., Craig, W., & Connolly, J. (2008). Developmental
trajectories of bullying and associated factors. Child Development, 79,
325–338. doi:10.1111/j.1467-8624.2007.01128.x
Reijntjes, A., Vermande, M., Goossens, F. A., Olthof, T., van de Schoot,
R., Aleva, L., & van der Meulen, M. (2013). Developmental trajectories
of bullying and social dominance in youth. Child Abuse & Neglect, 37,
224–234. doi:10.1016/j.chiabu.2012.12.004
Reynolds, C. R., & Kamphaus, R. W. (2004). Behavior Assessment System
for Children (2nd ed.) manual. Minneapolis, MN: Pearson.
Smith, P. K., Madsen, K., & Moody, J. C. (1999). What causes the age
decline in reports of being bullied at school? Toward a developmental
analysis of risks of being bullied. Educational Research, 41, 267–285.
doi:10.1080/0013188990410303
Smith, P. K., Shu, S., & Madsen, K. (2001). Characteristics of victims of
school bullying: Developmental changes in coping strategies and skills.
In J. Juvonen & S. Graham (Eds.), Peer harassment in school: The plight
of the vulnerable and victimized (pp. 332–352). New York, NY: Guil-
ford Press.
Sroufe, L. A. (2013). The promise of developmental psychopathology: Past
and present. Development and Psychopathology, 25, 1215–1224. doi:
10.1017/S0954579413000576
Sroufe, L. A., & Rutter, M. (1984). The domain of developmental psycho-
pathology. Child Development, 55, 17–29. doi:10.2307/1129832
Swearer, S. M., Espelage, D. L., Vaillancourt, T., & Hymel, S. (2010).
What can be done about school bullying? Linking research to educa-
tional practice. Educational Researcher, 39, 38–47. doi:10.3102/
0013189X09357622
Swearer, S. M., Turner, R. K., Givens, J. E., & Pollack, W. S. (2008).
“You’re so gay!”: Do different forms of bullying matter for adolescent
males? School Psychology Review, 37, 160–173.
Vaillancourt, T., Hymel, S., & McDougall, P. (2003). Bullying is power:
Implications for school-based intervention strategies. Journal of Applied
School Psychology, 19, 157–176. doi:10.1300/J008v19n02_10
Vaillancourt, T., Trinh, V., McDougall, P., Duku, E., Cunningham, L.,
Cunningham, C.,...Short, K. (2010). Optimizing population screening
of bullying in school-aged children. Journal of School Violence, 9,
233–250. doi:10.1080/15388220.2010.483182
Wang, J., Iannotti, R. J., Luk, J. W., & Nansel, T. R. (2010). Co-occurrence
of victimization from five subtypes of bullying: Physical, verbal, social
exclusion, spreading rumors, and cyber. Journal of Pediatric Psychol-
ogy, 35, 1103–1112. doi:10.1093/jpepsy/jsq048
Wolke, D., Copeland, W. E., Angold, A., & Costello, E. J. (2013). Impact
of bullying in childhood on adult health, wealth, crime, and social
outcomes. Psychological Science, 24, 1958–1970. doi:10.1177/
0956797613481608
Wolke, D., Woods, S., Bloomfield, L., & Karstadt, L. (2000). The asso-
ciation between direct and relational bullying and behavior problems
among primary school children. Journal of Child Psychology and Psy-
chiatry, 41, 989–1002. doi:10.1111/1469-7610.00687
Received January 20, 2014
Revision received August 21, 2014
Accepted August 27, 2014 䡲
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2436 HALTIGAN AND VAILLANCOURT