Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/13845662

Individual Risk and Social Risk as Interacting Determinants of Victimization


in the Peer Group

Article  in  Developmental Psychology · December 1997


DOI: 10.1037/0012-1649.33.6.1032 · Source: PubMed

CITATIONS READS

776 2,704

3 authors, including:

Ernest Hodges David Perry


St. John's University Florida Atlantic University
55 PUBLICATIONS   5,609 CITATIONS    68 PUBLICATIONS   8,181 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Ernest Hodges on 31 May 2014.

The user has requested enhancement of the downloaded file.


Developmental Psychology Copyright 1997 by the American Psychological Association, Inc.
1997, Vol. 33, No. 6, 1032-1039 0012-1649/97/S3.00

Individual Risk and Social Risk as Interacting Determinants


of Victimization in the Peer Group
Ernest V. E. Hodges, Maurice J. Malone, and David G. Perry
Florida Atlantic University

This study evaluated the hypothesis that the behavior problems that place children at risk for victim-
ization by peers are associated with victimization primarily when children are also at social risk for
victimization. Social risk was defined as lacking supportive friends or as being rejected by the peer
group. Participants were 229 boys and girls in the 3rd through 7th grades (M age = 1 1 years 2
months). As predicted, behavior problems (internalizing problems, externalizing problems, and
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

physical weakness) were more strongly related to victimization when children had few friends, had
This document is copyrighted by the American Psychological Association or one of its allied publishers.

friends who were incapable of fulfilling a protective function (e.g., were physically weak), or were
rejected by peers than when children had more friends, had friends capable of defending them, or
were better liked by peers. Results illustrate the principle that individual risk variables depend on
social context for expression.

Recent research has drawn attention to the existence and the manifestly anxious, lack humor, lack self-confidence and self-
plight of children who serve habitually as targets of peer aggres- esteem, and use ineffectual persuasion tactics; moreover, they
sion (Olweus, 1978; Perry, Kusel, & Perry, 1988). Victimiza- reward their attackers by being submissive and by relinquishing
tion, like aggression, is highly stable over the school years (Ep- resources (Olweus, 1978; Patterson, Littman, & Bricker, 1967;
stein, 1990; Khatri, Kupersmidt, & Patterson, 1994; Olweus, Perry et al., 1988; Perry, Williard, & Perry, 1990; Pierce, 1990;
1978) and is associated with a variety of negative adjustment Schwartz, Dodge, & Coie, 1993). Many of the foregoing attri-
outcomes across a wide age span. For example, kindergartners butes are consistent with a picture of the victimized child as
who are victimized in the fall of the school year are likely to having "internalizing difficulties." In addition, victimized chil-
avoid school and to report feelings of loneliness in the spring dren tend to be physically weak (Olweus, 1978). Longitudinal
(Kochenderfer & Ladd, 1996). Elementary schoolers who are work indicates that internalizing difficulties and physical weak-
victimized show gains in depression over time (Hodges, Ma- ness contribute to gains in victimization over time; they are not
lone, & Perry, 1995; Khatri et al., 1994; Kupersmidt & Khatri, simply correlates or consequences of it (Hodges et al., 1995;
1995). "Young adolescents who are abused by peers report ele- Schwartz et al., 1993). It should be noted that some victimized
vated depression and low self-esteem 10 years later in adulthood children also display ' 'externalizing problems,'' such as disrup-
(Olweus, 1992). tiveness, ineffectual aggression, and argumentativeness (Ol-
In an effort to understand the factors that place children at risk weus, 1978; Perry et al., 1988; Perry, Perry, & Kennedy, 1992).
for victimization, several investigators have focused on proximal These behaviors very likely serve to irritate and provoke other
influences operating within the peer group, especially forms of children, especially bullies; victimized children with these attri-
social impairment exhibited by children that might contribute butes are sometimes called "provocative victims" (Olweus,
to their victimization. The guiding assumption has been that 1978).
victimized children behave in ways that invite or reinforce at-
Like the foregoing studies, the present investigation is also
tacks against them. A growing literature confirms the validity
premised on the idea that children's behavioral attributes con-
of this premise. Many victimized children exhibit behaviors
tribute to their victimization, but it is built on the additional idea
that very probably signal that they will be unlikely to defend
that the relation of behavioral vulnerabilities to victimization
themselves successfully against attacks: They cry easily, are
depends on social context factors residing in the peer group. In
particular, the present investigation tested a theory of victimiza-
tion that holds that chronic victimization occurs mainly when a
Ernest V. E. Hodges, Maurice J. Malone, and David G. Perry, Depart- child not only possesses a behavior problem that places the
ment of Psychology, Florida Atlantic University. child at risk for victimization but also occupies a social position
We express appreciation to William M. Bukowski, who shared ideas in the peer group that invites, disinhibits, or permits aggressive
and data that inspired certain analyses reported in this article. This attacks toward the child by other children. We propose that a
research was supported by National Science Foundation Grant BNS 89- child who possesses a behavioral vulnerability (e.g., internaliz-
07558. We gratefully acknowledge the cooperation of the faculty and
ing problems, physical weakness) or who is behaviorally irritat-
students of the A. D. Henderson University School.
Correspondence concerning this article should be addressed to David ing to peers (e.g., displays externalizing behaviors) is likely to
G. Perry, Department of Psychology, Florida Atlantic University, Boca become chronically victimized primarily when the child either
Raton, Florida 33431. Electronic mail may be sent via Internet to perrydg (a) lacks friends who can support and protect the child against
@fau.edu. aggressors or (b) is generally rejected by the peer group. We
1032
VICTIMIZATION BY PEERS 1033

suggest that both of these aspects of a child's social relations, nalizing problems, externalizing problems, and physical weak-
which we refer to as social risk factors, are visible and known ness to predict victimization more strongly for peer-rejected
to other children in the peer group, including aggressive chil- children than for children who are not widely disliked by peers.
dren, and serve to elicit and to excuse chronic aggression toward
children with behavioral difficulties. In contrast, children who
Method
have behavior problems but are fortunate to have supportive
friends or to be better accepted generally by the peer group Participants
should be protected from chronic victimization. The roles of
these two social risk factors—the lack of allies in the form of Participants were 119 boys and 110 girls in the third through seventh
grades of a university school serving a middle-class, predominantly
friends and peer rejection—are discussed in turn.
White community (M age = 11 years 2 months). This age range was
Children's friendships serve many important developmental chosen for study because it has been suggested that victimization tends
functions (Hartup, 1993), but one function that has received to become stable during this age period (Perry et al., 1992). Written
relatively little attention is a protective function. Several investi- parental consent had been obtained for participants, who represented
gators (Bukowski, Sippola, & Boivin, 1995; Rizzo, 1989) have 92% of the children in their grades. The children participating in this
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

proposed that having one or more supportive friends protects study also took part in an unrelated study by Finnegan, Hodges, and
This document is copyrighted by the American Psychological Association or one of its allied publishers.

children against victimization. Children know very well who is Perry (1996).
friends with whom (Cairns, Cairns, Neckerman, Gest, &
Gariepy, 1988), and aggressive children probably prefer to at- Procedure
tack peers who lack supportive and protective friends because
they can do so without worrying about retaliation or ostracism Two to 3 months before the end of the school year, children responded
from the children's friends. Support for the hypothesis that to two instruments administered in their homerooms. The first was a
friendship protects against victimization was offered by Bukow- 53-item modified Peer Nomination Inventory designed to measure chil-
dren's behavioral characteristics. Because of its length, the instrument
ski et al. (1995) and Malone and Perry (1995), who found that
was administered in two sessions on consecutive school days. At the
friendless children were indeed more likely to be victimized
end of the second session, the second instrument was given. This was a
than were children who had friends. brief sociometric measure asking children to name preferred and nonpre-
In the present study, we tested the hypothesis that internalizing ferred peers. Each instrument is described below.
problems, externalizing problems, and physical weakness are
more strongly associated with victimization when children have
Modified Peer Nomination Inventory
few friends than when they have more friends. Such a finding
would support the idea that friendship buffers or protects the This inventory was a modification of Wiggins and Winder's (1961)
child against the potential contribution of behavioral risk factors Peer Nomination Inventory (PNI). The modified PNI contained 53 items
to victimization. We also tested the hypothesis that when chil- and was designed to measure victimization, physical strength, and 10
dren do have one or more friends, their friends' behavioral other behavioral dimensions relevant to social competence with peers.
qualities also affect the relation of behavioral risk to victimiza- The behavioral dimensions and the items were culled from previous
studies of peer relations (Coie & Dodge, 1988; Parkhurst & Asher, 1987;
tion. In particular, we reasoned that if a child's friends are
Perry et al., 1988; Pierce, 1990; Pope, Bierman, & Mumma, 1991).
physically weak, characterized by internalizing problems, or Children checked off the names of same-sex classmates who manifested
victimized by peers, the friends will be unlikely to fulfill a the behavior described in each item.
protective function very effectively; thus children who exhibit The behavioral scales (and the items for each, from the boys' form)
behaviors that place them at risk for victimization will be more were as follows (the girls' form was identical to that for boys, except
likely to be victimized if their friends possess these qualities "she" and "her" replaced " h e " and " h i m " ) : (a) victimization—
than if they do not. seven items: "Kids make fun of him"; "He gets beat up"; "He gets
The second social risk factor that we investigated as a possible called names by other kids"; "Kids do mean things to him"; "He gets
picked on by other kids"; "He gets hit and pushed by other kids";
moderator of the relation of behavioral risk to victimization was
"Kids try to hurt his feelings"; (b) aggression—seven items: "When
peer rejection, or the degree to which a child is generally disliked
he doesn't get his way he gets real mad"; "He's just plain mean"; "He
by peers. Children who are rejected by classmates may be per- makes fun of people"; "He says he can beat everybody u p " ; "He tries
ceived as fair game by aggressive children because the knowl- to get other people in trouble"; "He hits and pushes others around";
edge that a child is widely devalued by peers may legitimize "He tries to pick fights with people"; (c) argumentativeness—two
the child's status as a target of abuse. Even nonaggressive, main- items: "He argues a lot"; "He always has to have his own way"; (d)
stream peers tend to express negative attitudes toward rejected dishonesty—two items: "He sometimes tells lies"; "Sometimes he
classmates (Dodge, 1986; Hymel, 1986); this-may signal to takes things that belong to someone else"; (e) pushy peer entry style—
aggressive children that attacks on rejected children will go one item: "He tries to get other kids to play with him even when they
unpunished. Previous studies have shown that victimized chil- don't want to"; (f) disruptiveness—two items: "He makes noise or
bothers you in class"; "He doesn't follow rules"; (g) immaturity —
dren are indeed rejected and devalued by peers (Perry et al.,
three items: "He acts like a baby"; "He gets upset when called on to
1988); moreover, when victimized children are attacked, they
answer questions in class"; "He complains a lot, and nothing makes
elicit little empathy from peers, even from nonaggressive ones him happy"; (h) withdrawal—two items: "On the playground he just
(Perry et al., 1990). In the present study, we examined whether stands around"; "He doesn't talk much"; (i) anxiety I depression—two
peer rejection might be a social condition that serves to disin- items: "He seems unhappy and looks sad often"; "He is afraid to do
hibit aggression toward children who possess behavior problems things"; (j) hovering peer entry style—one item: "When other kids
that put them at risk for victimization. Thus, we expected inter- are playing he just hangs around watching them but doesn't join in";
1034 HODGES, MALONE, AND PERRY

(k) prosocial behavior—two items: "He is good to have in a group tors that we labeled Externalizing Problems and Internalizing
because he shares things and gives other people a turn"; "He is always Problems, we caution that the labels of externalizing and inter-
friendly"; and (1) physical strength—three items: "He is very strong"; nalizing are used for descriptive convenience only and do not
"He would win an arm wrestling contest"; "He is good at sports." imply clinical relevance. We have not validated our peer-report
The remaining items were "filler" items (e.g., "He is good at art").
measures against standardized parent- or teacher-report mea-
A score on each scale was determined for each child by calculating
sures of behavior problems, and it is unlikely that many, if
the percentage of same-sex classmates who checked the child's name
on each item of the scale and then totaling these percentages for the any, of our participants exhibited a level of externalizing or
scale. Scores for each variable were then standardized by sex and grade internalizing symptoms that would qualify for clinical diagnosis.
groupings. Intercorrelations among the four peer nomination measures
Reliability of the peer nomination scales is uniformly high. Because are given in Table 1. Although this table indicates that victimiza-
the scoring of peer nomination items involves aggregating across multi- tion was moderately correlated with each of the three behavioral
ple respondents, even single-item peer nomination scales tend to be risk variables (internalizing problems, externalizing problems,
remarkably reliable (Coie, Dodge, & Kupersmidt, 1990). Indeed, test- and physical strength), caution must be exercised in interpreting
retest coefficients (rs) over a 1-year period for the same measures as these correlations because the use of common respondents
those used here were reported by Hodges et al. (1995) to be as follows:
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(peers) likely inflated the relations (e.g., perceptual biases may


This document is copyrighted by the American Psychological Association or one of its allied publishers.

victimization, .84; immaturity, .77; withdrawal, .80; anxiety/depression,


be operating). However, the main hypotheses of the study in-
.74; hovering peer entry style, .77; aggression, .73; argumentativeness,
volved tests of whether the relations of behavioral risk variables
.74; dishonesty, .72; pushy peer entry style, .58; disruptiveness, .74;
prosocial behavior, .69; and physical strength, .80. to victimization differ significantly as a function of level of
social risk, and tests of these hypotheses should be less compro-
mised by the use of common respondents to measure behavioral
Sociometric Measures risk variables and victimization. We turn now to tests of our
Children were asked to name the 3 same-sex classmates with whom major hypotheses.
they most liked to work or play. They also named the 3 same-sex class-
mates with whom they least liked to work or play. A child's friends
were defined as peers who reciprocated the child's nomination of liking Does Friendship Moderate the Relation of Behavioral
(range 0 - 3 ) ; these scores were then standardized within sex and grade. Risk to Victimization?
Rejection scores were determined by calculating the percentage of same-
sex classmates who nominated the child as 1 of 3 least preferred class- In this section, we evaluate the hypothesis that the relation
mates and then standardizing by sex and grade. of behavioral risk to victimization is moderated by both (a)
number of friends and (b) qualities of friends.
Number of friends. Consistent with prior research (Bukow-
Results
ski et al., 1995), number of friends was correlated with victim-
Factor Analysis of Adjustment Measures ization (r = -.42, p < .01). To determine whether availability
of friends moderates the relation of behavioral risk to victimiza-
To reduce the number of adjustment measures, a principal- tion, three multiple regression analyses were performed—one
components factor analysis with varimax rotation was per- for each behavioral risk variable (internalizing problems, exter-
formed on 10 of the PNI measures, not including victimization nalizing problems, and physical strength); the dependent vari-
and physical strength. Victimization was omitted because it is able was always victimization. In each analysis, the behavioral
the primary dependent variable of the study and because, con- risk variable and the number of friends were entered on the first
ceptually, it is not a measure of behavior that the child emits step; the product term was entered on the second step. (Number
but rather is an index of what the child experiences at the hands of friends was modestly correlated with each behavioral risk
of others. Physical strength was also omitted from the factor variable, rs = - . 3 4 , -.30, and .33, for internalizing problems,
analysis because it is not a social behavior in the same sense externalizing problems, and physical strength, respectively,
as the other measures are. The factor analysis yielded two factors ps < .01.)'
with eigenvalues over 1.0. Because the variable of immaturity The product term was significant in each analysis, and follow-
had high loadings on both factors, the analysis was redone ex- up analyses were conducted to examine how the relation of each
cluding this variable. The analysis of the nine remaining vari- behavioral risk factor to victimization hinges on number of
ables yielded two relatively pure factors. The first was labeled friends. Follow-up analyses (for these and other significant inter-
Externalizing Problems and accounted for 55% of the variance actions reported in this article) were carried out according to
(eigenvalue = 4.95). Aggression, argumentativeness, dishon- the procedure recommended by Aiken and West (1991). In
esty, pushy peer entry style, and disruptiveness loaded positively this procedure, the relation between a predictor variable and a
on this factor, whereas prosocial behavior loaded negatively. The criterion variable is estimated in the form of an unstandardized
second factor was labeled Internalizing Problems and accounted beta coefficient (/3) at each of three levels of a moderator vari-
for an additional 25% of the variance (eigenvalue = 2.21). able ( —1, 0, and +1 SD; i.e., at low, medium, and high levels
Withdrawal, anxiety/depression, and hovering peer entry style
had high loadings on this factor. For factor loadings of all vari-
ables, see Finnegan et al. (1996). Children's scores on the two 1
Because all variables were standardized within sex and grade, each
factors were computed with the Statistical Package for the Social variable was effectively centered (i.e., given a mean of zero). Centering
Sciences regression method, which yields uncorrelated factors. of variables is necessary for interpretation of interactions in multiple
Although the adjustment measures were reduced to two fac- regression analyses (Aiken & West, 1991).
VICTIMIZATION BY PEERS 1035

of the moderator). Comparing the /3s across the three levels of Table 2
the moderator variable allows one to see how the relation (i.e., Correlations of Participants' Qualities With
slope) between the predictor and the criterion changes with level Friends' Qualities
of the moderator.
For the analysis involving internalizing problems, the interac- Friends' qualities
tion term was significant, FA = 8.26, p < .01, and follow-up Participants' qualities 1 2 3 4
analyses indicated that, as predicted, the relation of internalizing
problems to victimization diminished as the number of chil- 1. Victimization .26* -.27* .31* -.12
dren's friends increased (/3s for low, medium, and high numbers 2. Physical strength -.27* .35* -.40* .10
3. Internalizing problems .30* -.40* .48* -.30*
of friends = .65, .49, and .33, respectively).2 In the analysis
4. Externalizing problems -.06 .07 -.22* .21*
involving externalizing problems, the product term was again
significant, FA = 16.21, p < .001; as expected, the relation of Note, n = 163.
externalizing problems to victimization declined as the number *p < .01.
of children's friends increased (/0s for low, medium, and high
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

numbers of friends = .64, .42, and .21). Finally, the interaction


This document is copyrighted by the American Psychological Association or one of its allied publishers.

of physical strength and number of friends was significant, FA


analyses (i.e., product terms were evaluated with main effects
= 11.20, p < .01; as anticipated, the negative relation of physical
already in the model). 3
strength to victimization diminished as the number of children's
Friends' victimization moderated the relations of all three
friends increased (/0s for low, medium, and high numbers of
participant behavioral risk factors to victimization in the antici-
friends = -.56, - . 3 6 , and - . 1 6 ) .
pated way; the product term FAs for analyses involving partici-
The follow-up analyses just reported were also performed pants' internalizing problems, externalizing problems, and phys-
one other way. Children were grouped into those with no friend ical strength were, respectively, 5.80, p < .05; 22.36,/? < .001;
(w = 66) versus those with one or more friends (n = 163), and and 5.04, p < .05. As friends' victimization increased from low
the relations of the behavioral risk variables to victimization to medium to high levels, there were corresponding increases
were compared across groups. Results were essentially the same in the degree to which victimization was predictable from parti-
as those reported above. cipants' internalizing problems (/3s = .22, .39, and .56), from
Qualities of friends. Of the 229 participants, 163 had at participants' externalizing problems (/3s = .21, .52, and .84),
least 1 reciprocated friend. To explore whether the behavioral and from participants' physical strength (/9s = —.22, - . 4 0 , and
qualities of children's friends moderate the relations of behav- -.58).
ioral risk to victimization, it was first necessary to estimate the Friends' internalizing problems did not significantly moderate
qualities of each child's friends. This was done by averaging the relation of participants' internalizing problems to victimiza-
the scores received by each child's friends on each of four tion, product term FA = 2.00, ns, but did moderate the relation
behavioral dimensions: victimization, physical strength, inter- of both participants' externalizing problems, FA = 14.16, p <
nalizing problems, and externalizing problems. (Because the .001, and participants' physical strength, FA = 14.38, p < .001,
number of a child's friends could vary from 1 to 3, these aver- to victimization. As friends' internalizing problems increased
aged scores were based on different numbers of friends for across low, medium, and high levels, there were the expected
different participants). corresponding increases in the degree to which victimization
The top row of Table 2 presents the correlations between could be predicted from participants' externalizing problems
participants' victimization and the four behavioral qualities of (/3s = .39, .60, and .80) and from participants' physical strength
their friends. Participants' victimization was related to friends' (/3s = - . 1 3 , ns, - . 4 2 , and - . 7 2 ) .
victimization, friends' physical strength, and friends' internaliz- Friends' physical strength moderated the relations of all three
ing problems but not to friends' externalizing problems. To de- behavioral risk variables to victimization in the expected way;
termine whether friends' qualities moderate the relations of be- the product term FAs for internalizing problems, externalizing
havioral risk to victimization, 12 regression analyses were run, problems, and physical strength were 12.03, p < .001, 20.31,
3 (participant behavioral risk variables) X 4 (qualities of partic- p < .001, and 6.25, p < .05, respectively. As children's friends
ipants' friends). The n for these analyses was 163. The form increased in strength (from low to medium to high levels),
of these analyses was similar to that of the earlier moderator victimization became decreasingly predictable from partici-
pants' internalizing problems (/3s = .60, .30, and .04, ns), from
participants' externalizing problems (/3s = .78, .50, and .21),
Table 1
Correlations Among Four Peer-Report Measures
2
We also computed the SE, and tested the significance, of each /?
Measure 1 2 3 with a t test. All of the /3s reported in this article are significant (p <
.05) or better except those denoted as ns immediately following the 0
1. Victimization
-.44* value.
2. Physical strength 3
3. Internalizing problems .65* -.56* Because the regression analyses reported in this section were com-
4. Externalizing problems .58* -.06 .00 puted with a subset of 163 participants, it was first necessary to recenter
the data for these children. This was done by restandardizing all the
Note. N = 229. variables within sex and grade for this subsample. The data given in
*p < .01. Table 2 are also based on these restandardized scores.
1036 HODGES, MALONE, AND PERRY

and from participants' physical strength (/3s = - . 5 8 , -.40, and each of the study variables (participants' victimization, partici-
-.21). pants' internalizing, participants' externalizing, participants'
In the theory presented in the introduction to this article, physical strength, number of friends, friends' victimization,
friends' externalizing problems were not presented as a potential friends' internalizing, friends' externalizing, friends' physical
moderator of associations between behavioral risk and victim- strength, and rejection) were examined in separate multiple re-
ization. However, one might argue that the more children's gression analyses in which sex and grade were entered as simul-
friends are aggressive and exhibit other externalizing behaviors, taneous predictors.
the less participant behavioral risk variables should be linked Sex differences (with grade controlled) were evident for four
with victimization. This is because having aggressive friends measures: Compared with girls, boys had higher scores on parti-
might protect a child from being victimized; that is, aggressive cipants' externalizing (p < .05), friends' victimization (p <
friends may be especially likely to defend, and retaliate on behalf .01), friends' internalizing (p < .05), and friends' externalizing
of, their companions. Consequently, we also examined whether (p < .01). Grade differences (with sex controlled) were evident
friends' externalizing problems moderate relations of participant for three measures: As children grew older, they had more
behavioral risk variables to victimization. Friends' externalizing friends (p < .05), but, surprisingly, they were less likely to be
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

problems did not moderate the relation of either participants' named as physically strong (p < .05) or to have friends named
This document is copyrighted by the American Psychological Association or one of its allied publishers.

internalizing problems, FA = .05, ns, or participants' physical as physically strong (p < .01). These various effects of sex
strength, FA = .78, ns, to victimization. However, friends' ex- and grade indicate the usefulness of having controlled sex and
ternalizing problems did moderate the relation of participants' grade (through standardizing) in the main analyses.
externalizing problems to victimization, FA = 10.78,/? < .001. Standardizing by sex and grade does not preclude tests of
As expected, participants' externalizing problems became in- interactions of these variables with other predictor variables.
creasingly less linked with victimization across low, medium, Although our theoretical formulation does not suggest that the
and high levels of friends' externalizing (/3s = .77, .54, and .31, processes under test (i.e., interactive contributions of behavioral
respectively).4 and social risk to victimization) depend on sex or grade, as a
check, all of the previous multiple regression analyses that in-
volved victimization as the dependent variable were rerun to
Does Peer Rejection Moderate the Relation of include product terms involving sex and grade. In each analysis,
Behavioral Risk to Victimization? the main effects of the behavioral risk variable and the social
risk variable were entered on the first step. On the second step,
In this data set, peer rejection bore a sizable zero-order rela- six two-way interactions (all pairwise combinations of the be-
tion to victimization (r = .80). In addition, rejection was corre- havioral risk variable, the social risk variable, sex, and grade)
lated with each behavioral risk variable (rs for internalizing were entered. On the final step, the three possible three-way
problems, externalizing problems, and physical strength = .48, interactions were entered.
.70, and —.32, respectively, ps < .01). However, as predicted,
Only one three-way interaction was significant (the interac-
rejection also moderated the relation of each behavioral risk
tion of participants' externalizing problems, friends' internaliz-
variable to victimization; again, moderation was tested by evalu-
ing problems, and grade, p < .05). We therefore conclude that,
ating the product term with rejection and the behavioral risk
in general, the interactive effects of behavioral and social risk
factor already in the model. Product term FAs for the analyses
reported in earlier sections apply equally well to both boys and
involving internalizing problems, externalizing problems, and
girls and to children of different grades.5
physical strength were, respectively, 24.32, p < .001; 6.56,
p < .01; and 28.72, p < .001. Follow-up analyses confirmed
that as the level of participants' rejection by peers moved from Discussion
low to medium to high, victimization became increasingly pre-
dictable from participants' internalizing problems (/3s = .13, Results provide considerable support for the theory that social
.27, and .41), from participants' externalizing problems (/3s = risk moderates the relation of behavioral risk to victimization
- . 0 8 , ns; .01, ns; and .10, ns), and from participants' physical
strength (/3s = - . 0 1 , ns; - . 2 5 , and - . 4 9 ) .
4
The qualities of children's friends were only weakly related to the
number of friends children had. For the 163 children who had at least
Effects of Sex and Grade 1 friend, the correlations between number of friends (from 1 to 3) and
friends' victimization, friends' internalizing problems, friends' exter-
As noted, variables were standardized within sex and grade nalizing problems, and friends' physical strength were, respectively,
groupings before being analyzed. This is a common procedure - . 0 8 , ns; -.14, ns; .16, p < .05; and .21, p < .01. Because the latter
with peer nomination data and is done primarily to prevent two correlations were significant, the moderator analyses reported here
and above were rerun with number of friends included as a control
differences in group size from unduly influencing the data. The
variable; results were essentially the same.
procedure has other advantages, including centering the data 5
A number of two-way interactions involving either sex or grade
and reducing the number of predictor variables entered into the were also significant in the supplementary analyses but were not relevant
analyses (although still controlling for main effects of sex and to the study. For example, sex interacted with participants' internalizing
grade). However, a disadvantage is that standardization obscures problems to predict victimization (p < .01), owing to the fact that
possible sex and grade differences in the measures. To explore internalizing was somewhat more strongly related to victimization for
such possible differences, the raw (unstandardized) scores of girls (r = .73) than for boys (r = .56).
VICTIMIZATION BY PEERS 1037

by peers. Clearly, children who possess behavior problems that both may reflect a common "social outcast" risk factor. At
put them at risk for victimization are more likely to be chroni- present, there is no theory to suggest that either of these two
cally abused if they are also at social risk for victimization, that related variables should be the superior predictor of aggression
is, if they lack friends who can protect them or if they are widely toward behaviorally at-risk children; hence, we did not perform
devalued by peers. The roles of these two social risk factors— analyses in which the effects of each risk variable were evalu-
lack of supportive friends and peer rejection—are discussed in ated with the effects of the other controlled. Our point is simply
turn. that social risk, however measured, serves to invite aggression
As described by Hartup (1992) and by others, childhood toward susceptible children.
friendships serve several important functions. Friendships are Results of this study support the principle that individual-
contexts for learning social skills, are information sources for level vulnerabilities often depend on social context factors for
self-knowledge and self-esteem, furnish emotional and cognitive expression (Rutter & Garmezy, 1983). This principle has been
resources for support and coping, and provide practice for later supported in research with families. For example, Kagan (1994)
relationships. The present results accord with the notions of reported that, among boys, inhibited temperament leads to anxi-
Bukowski et al. (1995) and of Rizzo (1989) that friendships ety disorder primarily when the mother is overprotective. The
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

serve an additional important function—that of protection present data argue that social context factors residing in the peer
This document is copyrighted by the American Psychological Association or one of its allied publishers.

against bullies and aggressors. The protective role of friendship group, such as the number and nature of one's friends, also
was evident in the present data in two ways. First, the sheer can serve to actualize vulnerabilities (or, alternatively, to buffer
number of a child's friends moderated the link of behavioral against their expression).
risk to victimization. That internalizing problems, externalizing Although most chronically victimized children are character-
problems, and physical weakness all were considerably more ized by internalizing problems, such as anxiety, depression, and
predictive of victimization for children with fewer rather than low self-esteem (Perry et al., 1992), victimized children differ
with many friends is evidence that friendship buffers or protects widely in the degree to which they also display externalizing
behaviorally at-risk children from victimization. Having friends problems, such as aggression, disruptiveness, and other provoca-
may protect children in several ways. First, aggressive children tive behaviors (Olweus, 1978; Perry et al., 1988, 1992). Do the
may fear retaliation or ostracism from a victim's friends. Sec- processes studied in this report apply equally to nonaggressive
ond, children who interact with friends will be alone less often victims and aggressive (or "provocative") victims? We believe
and thus will be less available or less salient as targets. Third, the general principle of behavioral risk interacting with social
children with friends may profit from friends' advice on how to risk to promote victimization applies equally well to both kinds
handle conflicts and threats of victimization. Of course, having of victims, but clearly the nature of the behavioral risk involved
friends did not completely eliminate the risk of victimization will differ for the two kinds of victims, with internalizing prob-
for children with behavior problems because all three behavioral lems predominating for nonaggressive victims and a combina-
risk indexes still bore some relation to victimization even for tion of internalizing and externalizing problems more likely for
children with relatively many friends. aggressive victims. We also suspect that the nature of the social
A second source of support for the hypothesis that friendship risk involved will differ for the two sorts of victims. Malone
serves a protective function comes from the analyses showing and Perry (1995) recently reported that nonaggressive victims
that, for children who did have friends, the friends' behavioral have almost as many friends as normal children do, but their
qualities moderated the relation of behavioral" risk to victimiza- friends tend to be weak, to have internalizing problems, and to
tion. As predicted, when children's friends possessed qualities be victims, too; in contrast, aggressive victims were found to
that rendered them unlikely to serve a protective function very have almost no friends at all. Thus, for nonaggressive vic-
effectively (e.g., the friends were victimized or weak), the rela- tims, social risk may take the form of having less-than-helpful
tion of behavioral risk to victimization was greater than when friends, but, for aggressive victims, it may take the form of
the friends were more capable of providing protection and de- friendlessness.6
fense. It is also informative that when children's friends were Certain strengths and limitations of the current study deserve
characterized by externalizing behaviors, the children's own mention. A strength is the large sample size that allowed tests
externalizing problems were less predictive of victimization than of the generality of the processes under study across sex and
when their friends lacked externalizing problems. Friends who age. However, the use of a homogeneous, largely White middle-
are prone to externalizing behaviors may retaliate on behalf of class sample limits generalizations to other groups.
their friends and thereby serve a deterrent or protective function. A rather large number of statistical tests were performed on
In sum, both the number and the nature of children's friends the present data, raising the possibility that some significant
govern the relation of behavioral risk to victimization by peers findings occurred by chance. However, given that significant
in meaningful ways. findings were predicted and painted a consistent picture, we do
Rejection from the peer group constitutes another social risk
factor that probably serves to encourage attacks upon children 6
Malone and Perry's (1995) finding that aggressive victims have few
who display behavioral vulnerabilities or behave in ways that
friends may appear inconsistent with the conclusion of other researchers
irritate peers. The relation of each behavioral risk factor to (e.g., Cairns et al., 1988) that aggressive children do not lack friends.
victimization was greater for peer-rejected children than for The findings are reconcilable. Malone and Perry found that although
better accepted children. Having few friends and being rejected aggressive-victimized children had very few friends, aggressive-nonvic-
by peers are, of course, related conditions (r between number timized children had many friends (even more than nonaggressive-non-
of friends and peer rejection in the present study = —.38), and victimized children, though not significantly so).
1038 HODGES, MALONE, AND PERRY

not believe that the overall contribution was compromised by References


the number of statistical tests conducted.
In this study, victimization was assessed in terms of receipt Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and
of direct physical and verbal abuse; other sorts of victimization, interpreting interactions. Newbury Park, CA: Sage.
such as indirect or relational victimization (Crick & Grotpeter, Bukowski, W. M., Sippola, L. K., & Boivin, M. (1995, March). Friend-
1995), may operate according to different processes. Relational ship protects "at risk" children from victimization by peers. In J. M.
aggression refers to acts intended to harm a child's relations Price (Chair), The role offriendship in children's developmental risk
and resilience: A developmental psychopathology perspective. Sym-
with peers (e.g., encouraging one's friends to exclude a disliked
posium conducted at the biennial meeting of the Society for Research
child from a social function or clique); girls are more likely in Child Development, Indianapolis, IN.
than boys both to enact and to receive this form of aggression.
Cairns, R. B., Cairns, B. D., Neckerman, H. J., Gest, S. D., & Gariepy,
It is important to bear in mind, however, that the present as well J. (1988). Social networks and aggressive behavior: Peer support or
as prior studies (e.g., Perry et al., 1988) have shown that girls peer rejection? Developmental Psychology, 24, 815-823.
are just about as likely as boys to experience the kind of direct, Coie, J. D., & Dodge, K. A. (1988). Multiple sources of data on social
overt forms of abuse investigated here. Hence, the sort of victim- behavior and social status in the school: A cross-age comparison.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ization studied here is equally relevant to both girls and boys. Child Development, 59, 815-829.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Two design limitations should be noted. First, the measures Coie, J. D., Dodge, K. A., & Kupersmidt, J. B. (1990). Peer group
of behavioral vulnerability and victimization came from the behavior and social status. In S. R. Asher & J. D. Coie (Eds.), Peer
rejection in childhood (pp. 17-59). New "fork: Cambridge University
same set of respondents—children's peers—and therefore rela-
Press.
tions between these variables may be inflated (e.g., due to per-
Crick, N. R., & Grotpeter, J. K. (1995). Relational aggression, gender,
ceptual biases). Of course, tests of the main hypotheses did not
and social-psychological adjustment. Child Development, 66, 710-
rest on the absolute size of relations between behavioral risk 722.
and victimization but rather depended on the relative sizes of Dodge, K. A. (1986). A social information processing model of social
these relations across different levels of the social risk moderator competence in children. In M. Perlmutter (Ed.), Minnesota Sympo-
variables. Nonetheless, there could be different levels of percep- sium on Child Psychology (pp. 77-126). Hillsdale, NJ: Erlbaum.
tual bias for different levels of risk. Clearly, greater confidence Epstein, A. M. (1990). Stability of victimization in elementary school
could be placed in our results if either behavioral risk variables children. Unpublished master's thesis, Florida Atlantic University,
or victimization had been measured independently of peer Boca Raton.
reports. Finnegan, R. A., Hodges, E. V. E., & Perry, D. G. (1996). Preoccupied
and avoidant coping during middle childhood. Child Development,
A second design limitation is that all the measures were col-
67, 1318-1328.
lected concurrently, opening the door to alternative interpreta-
Hartup, W. W. (1992). Peer relations in early and middle childhood. In
tions with regard to direction of effects. For example, variables
V. B. Van Hasselt & M. Hersen (Eds.), Handbook of social develop-
we have conceptualized as causing victimization (e.g., internal- ment: A lifespan perspective. New \brk: Plenum.
izing problems, friendlessness) may actually be outcomes of a Hartup, W. W. (1993). Adolescents and their friends. In B. Laursen
history of abuse. It is also possible, even likely, that behavioral (Ed.), Close friendships in adolescence (pp. 3-22). San Francisco:
risk factors help create certain of the social risk conditions Jossey-Bass.
that encourage victimization. For example, because depressed Hodges, E. V. E., Malone, M. J., Jr., & Perry, D. G. (1995, March).
children tend to befriend other depressed children (see Table 2, Behavioral and social antecedents and consequences of victimization
as well as Hogue & Steinberg, 1995), they may be helping by peers. In N. R. Crick (Chair), Recent trends in the study of peer
to create the very social environments that cause them to be victimization: Who is at risk and what are the consequences? Sympo-
victimized. Longitudinal work would clearly be illuminating. It sium conducted at the biennial meeting of the Society for Research
might be especially informative to measure risk factors and in Child Development, Indianapolis, IN.
victimization in both the fall and the spring of a school year. Hogue, A., & Steinberg, L. (1995). Homophily of internalized distress
This would allow one to see, for example, whether children in adolescent peer groups. Developmental Psychology, 31, 897-906.
Hymel, S. (1986). Interpretations of peer behavior: Affective bias in
who begin the year with behavior problems become victimized
childhood and adolescence. Child Development, 57, 431-445.
primarily to the extent that they also lack a supportive friendship
Kagan. J. (1994). Galen's prophecy: Temperament in human nature.
network.
New 'Vfork: Basic Books.
Ultimately, questions concerning the causal primacy of behav- Khatri, P., Kupersmidt, J., & Patterson, C. J. (1994, April). Aggression
ioral risk, social risk, and victimization may reduce to a and peer victimization as predictors of self-report of behavioral and
"chicken-and-the-egg" problem. It may be best to regard co- emotional adjustment. Poster presented at the biennial meeting of the
occurrences of victimization with various risk factors as systems Conference on Human Development, Pittsburgh, PA.
of mutually reinforcing components. Such a conceptualization Kochenderfer, B. J., & Ladd, G. W. (1996). Peer victimization: Cause
should not imply helplessness concerning intervention, however, or consequence of school maladjustment? Child Development, 67,
1305-1317.
because disruption of one component may lead to desirable
Kupersmidt, J. B., & Khatri, P. (1995, March). Peer victimization and
changes in others, as systems theorists emphasize (e.g., Mi-
aggression as predictors of self-reported behavioral and emotional
nuchin, 1985). Results of the present study suggest that inter- problems across adolescence. In N. R. Crick (Chair), Recent trends in
ventions aimed to give victimized children the benefits of friend- the study of peer victimization. Symposium conducted at the biennial
ship (e.g., arranging for a victimized child to begin the next meeting of the Society for Research in Child Development, Indianapo-
school year with a friend in the same class; see Ladd & Price, lis, IN.
1987) might be one way of breaking the cycle of peer abuse. Ladd, G. W., & Price, J. M. (1987). Predicting children's social and
VICTIMIZATION BY PEERS 1039

school adjustment following the transition from preschool to kinder- Conflict in child and adolescent development (pp. 301-329). New
garten. Child Development, 58, 1168-1189. %rk: Cambridge University Press.
Malone, M. J., & Perry, D. G. (1995, March). Features of aggressive Perry, D. G., Williard, J. C , & Perry, L. C. (1990). Peers' perceptions
and victimized children's friendships and affiUative preferences. Poster of the consequences that victimized children provide aggressors. Child
presented at the biennial meeting of the Society for Research in Child Development, 61, 1310-1325.
Pierce, S. (1990). The behavioral attributes of victimized children. Un-
Development, Indianapolis, IN.
published master's thesis, Florida Atlantic University, Boca Raton.
Minuchin, P. (1985). Families and individual development: Provocations
Pope, A. W., Bierman, K. L., & Mumma, G. H. (1991). Aggression,
from the field of family therapy. Child Development, 56, 289-302.
hyperactivity, and inattention-immaturity: Behavior dimensions asso-
Olweus, D. (1978). Aggression in the schools: Bullies and whipping ciated with peer rejection in elementary school boys. Developmental
boys. Washington, DC: Hemisphere. Psychology, 27, 663-671.
Olweus, D. (1992). Victimization by peers: Antecedents and long-term Rizzo, T. A. (1989). Friendship development among children in school.
outcomes. In K. H. Rubin & J. B. Asendorpf (Eds.), Social with- Norwood, NJ: Ablex.
drawal, inhibition, and shyness in childhood (pp. 315-341). Hills- Rutter, M., & Garmezy, N. (1983). Developmental psychopathology. In
dale, NJ: Erlbaum. E. M. Hetherington (Ed.), Handbook of child psychology: Vol. 4.
Socialization, personality, and social development (pp. 775-911).
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Parkhurst, J. T, & Asher, S. R. (1987, March). The social concerns of


This document is copyrighted by the American Psychological Association or one of its allied publishers.

aggressive-rejected children. Paper presented at the biennial meeting New York: Wiley.
of the Society for Research in Child Development, Baltimore, MD. Schwartz, D., Dodge, K. A., & Coie, J. D. (1993). The emergence of
Patterson, G. R., Littman, R. A., & Bricker, W. (1967). Assertive behav- chronic peer victimization in boys' play groups. Child Development,
ior in children: A step toward a theory of aggression. Monographs of 64, 1755-1772.
Wiggins, J. S., & Winder, C. L. (1961). The Peer Nomination Inventory:
the Society for Research in Child Development, 35 (5, Serial No.
An empirically derived sociometric measure of adjustment in preado-
113).
lescent boys. Psychological Reports, 9, (Al-dll.
Perry, D. G., Kusel, S. J., & Perry, L. C. (1988). Victims of peer aggres-
sion. Developmental Psychology, 24, 807—814. Received April 30, 1996
Perry, D. G., Perry, L. C , & Kennedy, E. (1992). Conflict and the devel- Revision received April 15, 1997
opment of antisocial behavior. In C. U. Shantz & W. W. Hartup (Eds.), Accepted May 5, 1997 •

Develojmiental Psyctvilogy Sentembor 1997

October 1997
Dovclopnenta Psychology
1h <> 0

"
U-
'
•-~™-«=«- — ™ 6,400 6,660
BL-HoiUhly •!.
'• • • '
'/ n,'

:...... iirc i a

(202) 3 36-5579
5,496 5,280

5,496 5,280
t HE, Washington, I
175 17b

Psychological Association, 750 First Street, HE, Washington, DC 20002-4242


175 176

Carolyn Zahn-Waxler, Ph.D., 4305- Dresden Street, Kensington, MD 20895 5,671 5,456

,-,0-^U^.SPO- 729 1,206


Susan Knapp, tooclcan Psychological Association,
750 Firat Street, NE, Ifaahin^ton, DC 20002-4242
-as
6,400 • „ • •

M N M
95.9 96.8

Washington, DC 20002-4242

• Wuar rrMukig MvnMonffllH< Brm

Hoklnj 1 PKMUC. JJor.nl TOM iwnoi« Of Bon*. i t a ^ j M . of ^ I n s l r u c t i o n s to P u b l l i h e r s

CompWUflk, M a r . .

II n» puUtctUan tite M o m W i u •uriortuUsn u * gtnani a ™qw«« pubfcaWon. Oilt Si«»m«« oi Omimht. M»n»B«monl, B « I
riinJiTkm mi ifI ho pufahihJLirtmull b* Dftnltd w >nv h i u In Odofcorof • thfl DUblctElpn 15 not DUblvWd du Oc[ofc* Vt V

View publication stats

You might also like