reactions to bullying and peer victimization: narratives, physiological arousal, and personality

26
Journal of Research in Personality 40 (2006) 803–828 www.elsevier.com/locate/jrp 0092-6566/$ - see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.jrp.2005.09.003 Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality Julie M. Bollmer, Monica J. Harris ¤ , Richard Milich Department of Psychology, University of Kentucky, USA Available online 25 October 2005 Abstract A mediational model of bullying and victimization is proposed and tested. Ninety-nine 10- to 13- year-old children provided two oral narratives of their victimization experiences, as perpetrator and victim, with their physiological arousal being measured while they told the narratives. The children and one of their parents also completed a variety of questionnaires, including a Big 5 measure of per- sonality and measures of bullying and victimization tendencies. Mediational analyses indicated that children who score low on Conscientiousness and high on Neuroticism are more likely to experience negative aVect during peer conXict, such as feeling angrier, blaming the bully more, and forgiving less, and that these reactions are related to higher levels of victimization. For bullies, relations among Agreeableness, Conscientiousness, and bullying appear to be mediated by lesser feelings of guilt and gains in physiological arousal while telling a bullying narrative. Advantages of a mediational model of peer victimization processes and implications for interventions are discussed. © 2005 Elsevier Inc. All rights reserved. Keywords: Peer victimization; Bullying; Personality; Big 5; Physiological arousal; Narrative analysis; Mediation First authorship is shared equally by the Wrst two authors. Julie Bollmer is now at Westat Corporation. This research was supported by a grant from the University of Kentucky Research Foundation. We are deeply grateful to the children and families who participated in our study. We thank Jenny Butler, Jennifer Gardner, Stephanie Haddix, Kyle Linneman, Melissa Maras, Jennifer Milburn, Thomas McCann, and Paul Rosen for their help in data collection and coding. We are also grateful to Charles Carlson and Suzanne Segerstrom for consultations on the physiological measurement and Don Lynam and Rick Hoyle for their statistical help. * Corresponding author. Fax: +1 859 323 1979. E-mail address: [email protected] (M.J. Harris).

Upload: julie-m-bollmer

Post on 09-Sep-2016

214 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

Journal of Research in Personality 40 (2006) 803–828

www.elsevier.com/locate/jrp

Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality �

Julie M. Bollmer, Monica J. Harris ¤, Richard Milich

Department of Psychology, University of Kentucky, USA

Available online 25 October 2005

Abstract

A mediational model of bullying and victimization is proposed and tested. Ninety-nine 10- to 13-year-old children provided two oral narratives of their victimization experiences, as perpetrator andvictim, with their physiological arousal being measured while they told the narratives. The childrenand one of their parents also completed a variety of questionnaires, including a Big 5 measure of per-sonality and measures of bullying and victimization tendencies. Mediational analyses indicated thatchildren who score low on Conscientiousness and high on Neuroticism are more likely to experiencenegative aVect during peer conXict, such as feeling angrier, blaming the bully more, and forgiving less,and that these reactions are related to higher levels of victimization. For bullies, relations amongAgreeableness, Conscientiousness, and bullying appear to be mediated by lesser feelings of guilt andgains in physiological arousal while telling a bullying narrative. Advantages of a mediational modelof peer victimization processes and implications for interventions are discussed.© 2005 Elsevier Inc. All rights reserved.

Keywords: Peer victimization; Bullying; Personality; Big 5; Physiological arousal; Narrative analysis; Mediation

� First authorship is shared equally by the Wrst two authors. Julie Bollmer is now at Westat Corporation. Thisresearch was supported by a grant from the University of Kentucky Research Foundation. We are deeply gratefulto the children and families who participated in our study. We thank Jenny Butler, Jennifer Gardner, StephanieHaddix, Kyle Linneman, Melissa Maras, Jennifer Milburn, Thomas McCann, and Paul Rosen for their help indata collection and coding. We are also grateful to Charles Carlson and Suzanne Segerstrom for consultations onthe physiological measurement and Don Lynam and Rick Hoyle for their statistical help.

* Corresponding author. Fax: +1 859 323 1979.E-mail address: [email protected] (M.J. Harris).

0092-6566/$ - see front matter © 2005 Elsevier Inc. All rights reserved.doi:10.1016/j.jrp.2005.09.003

Page 2: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

804 J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828

1. Introduction

Adults tend to glamorize childhood as a time of carefree innocence and joy. But child-hood is not always carefree, and nearly every child experiences at least occasionallysome form of victimization by peers. In the last decade, researchers have noted the detri-mental eVects of peer bullying on the mental health of children (Glover, Gough,Johnson, & Cartwright, 2000; Nansel et al., 2001; Olweus, 1995, 2003). Considerableresearch has appeared identifying the predictors and consequences of peer victimization(Hodges & Perry, 1999; Juvonen & Graham, 2001; Marsh, Parada, Craven, & Finger,2004; Olweus, 2003; Perry, Hodges, & Egan, 2001). The present study adds to this bur-geoning literature in three primary ways: First, we identify the stable personality traitsthat place children at risk for peer victimization and bullying, anchoring it within the Big5 literature, the dominant model of adult personality. Second, we examine the aVective,cognitive, and physiological experiences that occur during the retelling of a bullying andvictimization event. Third, and most important, we propose and test a theoretical modelof peer victimization that examines how the relations between personality and bullyingand victimization are mediated by the aVective, cognitive, and physiological processesoccurring within the child.

1.1. Theoretical model of victimization

As a theoretical framework for proposing and testing mediational hypotheses of bully-ing and victimization, we propose the following model (see Fig. 1). The model describestwo separate mediation pathways, one for bullies and one for victims. In each case, person-ality variables are posited to aVect bullying and victimization through their eVects on aVec-tive and cognitive responses that occur in the moment of a peer-conXict situation. In otherwords, stable individual diVerences such as personality will determine more transient andnarrower behaviors within a speciWc context, and it is these speciWc behaviors and howthey are construed by the child that help shape a child’s status as a bully or victim. Ourmodel thus endorses the seminal concept proposed by Caspi and Bem (1990) that the rela-tion between personality and behavior is not static but rather dynamic; that is, personalityaVects how an individual may interpret and respond to a speciWc event, which in turn caninXuence the probability with which that event may reoccur in the future. To oVer a speciWcexample from the victimization literature, a child high in Neuroticism may exhibit high lev-els of distress when being victimized, which in turn may reinforce bullies to pick on thischild at a high rate in the future (Patterson, Littman, & Bricker, 1967).

Our model and predictions are consistent with existing theories and models of aggres-sion and bullying (e.g., Dodge & Pettit, 2003; Hubbard, Dodge, Cillessen, Coie, &Schwartz, 2001; Lemerise & Arsenio, 2000; Neuman & Baron, 2003). The major diVerencebetween existing models in the literature and ours lies in the choice and speciWcation of thedistal individual diVerence variables and proximal mediating variables. For example,Dodge and Pettit’s (2003) model speaks of biological predispositions that inXuence con-duct problems, with ‘biological predispositions’ encompassing a broad range of potentialvariables such as polygenetic factors, prenatal toxicities, nervous system reactivity, behav-ioral facilitation and inhibition systems, and temperament. Our model, on the other hand,is more speciWc and focuses solely on the Big 5 personality factors, which the Dodge andPettit (2003) model does not explicitly include.

Page 3: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828 805

Another diVerence between our and existing models is our decision to include bothbullying and victimization pathways within a single model. The advantage of so doing isthat it makes explicit the social interactional components of peer victimization. Thus, inour model, we draw a causal connection between the victim’s reactions to a bullying inci-dent and subsequent increases in aggressive behaviors on the part of the bully. This path-way argues that visible signs of distress on the part of the victim serve as an importantreinforcement and source of pleasure to the bully (Arsenio & Lover, 1997; Perry,Williard, & Perry, 1990), thus increasing the likelihood that the bully will engage infuture bullying behavior, both with this particular victim as well as with other children.We do not draw an analogous pathway between bullies’ reactions and victims’ outcomesbecause we believe that how bullies react is less important in predicting a particularchild’s victimization. In other words, we believe that a victim’s response to being bulliedis a better predictor of subsequent bullying than is a bully’s behavior during bullying inpredicting later victimization.

Fig. 1. Theoretical model of bullying and victimization.

BULLIES VICTIMS

IndividualDifferenceVariables

Immediate Responses toBullying Incident

Long-term Outcomes

Personality –Neuroticism (+) –Agreeableness (-) –Conscientiousness (-) –Extraversion (+)

Personality –Neuroticism (+) –Agreeableness (-)–Conscientiousness (-) –Extraversion (-)

Affective and CognitiveMediators –Physiological arousal –Emotion dysregulation –Visibledistress –Negative affect

Affective and CognitiveMediators –Physiological arousal –Positive affect –Enjoyment –Reduced guilt –Hostile attributional bias

Increased Bullying

Increased Victimization

Page 4: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

806 J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828

1.1.1. Personality predictors of victimization and bullyingA primary goal of the current study is to apply a structural model of personality to

understanding bullying and victimization. As noted earlier, considerable research hasbeen devoted to identifying individual diVerence predictors, including social informationprocessing biases, of bullying and victimization (see Sanders & Phye, 2004, for an excel-lent recent review); for example, an early eVort documented that victimized childrenwere more likely to be physically weak, introverted, selWsh, aggressive, and attentionseeking (Lowenstein, 1978). Less attention has been paid, however, to the role of struc-tural models of personality such as the Big 5; our literature search identiWed only twostudies that have looked at the relation between the Big 5 personality factors and speciWcmeasures of bullying/victimization (Jensen-Campbell et al., 2002; Tani, Greenman,Schneider, & Fregoso, 2003).

Compelling reasons exist to consider the role of personality, and the Big 5 traits in par-ticular, in bullying and peer victimization. Concentrating Wrst on predictions for victims,previous research with adults has found that people who reported having been teased moreoften as a child scored higher on Neuroticism (Georgesen, Harris, Milich, & Young, 1999);in a sample of 232 Italian children, Tani et al. (2003) found a similar relation between neu-roticism and victimization. This Wnding is consistent with theoretical conceptions thatchildhood victims display high levels of distress and negative aVect (Perry et al., 2001), andwe thus predict that neuroticism will be positively associated with victimization in the cur-rent sample.

Both Tani et al. (2003) and Jensen-Campbell et al. (2002) found that Agreeableness wasnegatively associated with victimization. Consistent with these Wndings is work by Ehrler,Evans, and McGhee (1999) showing that Agreeableness was negatively related to “socialproblems,” a broad category that included interpersonal rejection and being teased. Basedon these Wndings and the reasoning that Agreeableness would be the Big 5 dimension mostclosely related to peer relations, we therefore predict that children scoring higher on Agree-ableness would be less likely to be victimized.

Tani et al. (2003) also found that victimization was negatively related to Conscientious-ness. The basis for this relation is less intuitively obvious, although it it is possible that chil-dren who are more conscientious, and therefore more determined and strong-willed, arebetter able to fend oV potential attacks by perpetrators. Alternatively, low conscientiouschildren may be more likely to act in antisocial ways that elicit retaliation and subsequentvictimization from peers. Based on this previous work, we oVer, more tentatively, the pre-diction that victims will score lower on Conscientiousness.

With respect to bullying, the developmental literature indicates that there is no sin-gle, homogeneous “bully” personality type; instead, there appear to be fairly discretesubgroups of bullies. One subgroup is the category called “bully-victims,” that is, chil-dren whose own aggressive behavior elicits subsequent retaliation and bullying by oth-ers (Schwartz, 2000; Schwartz, McFadyen-Ketchum, Dodge, Pettit, & Bates, 1999).These children appear to be the most impaired with respect to long-term outcomes andinternalizing psychopathology (Swearer, Song, Cary, Eagle, & Mickelson, 2001). Thus,the existence of the bully-victim subgroup would lead to the hypothesis that bulliesshould score higher on Neuroticism, a hypothesis that has found support in two studiesthat looked at correlations between bullying and scores on the junior version of theEysenck Personality Questionnaire (Connolly & O’Moore, 2003; Mynard & Joseph,1997).

Page 5: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828 807

A second subgroup consists of children who Wt more closely the classic stereotype of abully, that is, the dominant, imposing child who aggresses against others without fear ofretaliation. With this subgroup, bullying appears to share some overlap with conduct dis-order, which is a noted precursor of antisocial personality (McMahon & Wells, 1998).Other research has shown that individuals with antisocial personality are low on Agree-ableness and low on Conscientiousness (Miller & Lynam, 2001). It is reasonable to assumethat the personality factors associated with antisocial personality would generalize to bul-lying, especially “pure” bullying, and thus we predict that children who score high on bul-lying will also score low on Agreeableness and Conscientiousness. Support for thisprediction is also found in two other studies: Georgesen et al. (1999), who found in a sam-ple of adults that individuals who reported a history of teasing others in childhood scoredlower on Agreeableness and Conscientiousness, and Ehrler et al. (1999), who found thatAgreeableness and Conscientiousness was negatively related to teacher reports of students’conduct problems, a category that included losing one’s temper and picking Wghts withpeers. Tani et al. (2003) similarly found bullies to score lower on Agreeableness.

With respect to the remaining two variables in the Big 5 model, Extraversion and Open-ness to Experience, fewer relations between these variables and bullying and victimizationhave been documented in the literature. A few isolated Wndings suggest that victims may beless extraverted (Mynard & Joseph, 1997) and bullies may be more extraverted (Tani et al.,2003) or (for boys) more popular (Slee & Rigby, 1993). However, because these relationshave not held up consistently in the studies reviewed above or in research from our labora-tory, we include and analyze Extraversion and Openness to Experience only in an explor-atory fashion.

1.1.2. AVective and cognitive responses of bullies and victims during conXictAccording to our model, individual diVerence factors such as personality aVect peer vic-

timization outcomes through their mediating eVects on the aVective and cognitive reac-tions children have in the moment of a peer victimization experience. Past research that hasattempted to identify the emotional reactions of bullies and victims has utilized primarilyretrospective, paper-and-pencil measures. For example, Borg (1998), using a forced-choiceinventory, found that a signiWcant minority of bullies (21%) reported feeling satisWed abouttheir bullying, although a higher percentage (50%) reported feeling sorry. Age, however,was a signiWcant moderator of this Wnding such that older bullies were more likely to feelsatisWed and young bullies to feel sorry (Borg, 1998). In a multicultural study, Menesiniet al. (2003) found that when bullies, victims, and bystanders were asked to imagine how abully would feel in a hypothetical bullying situation, bullies reported higher levels of disen-gagement and egocentrism, and were more likely to indicate a lack of empathy, than didthe other two groups. Relatedly, Smith, Bowers, Binney, and Cowie (1993) had bullies, vic-tims, and control children view videotaped episodes of bullying and respond to questionsabout how the individuals in the videotape felt. Bullies were much more likely to say thatbullies felt happy about the incident and much less likely to say that the victims wereunhappy.

The studies reviewed above thus suggest that bullies should report more satisfaction,happiness, and less empathy in a bullying situation. However, one limitation of the previ-ous research is that it involved either retrospective self-reports or reactions to a Wctionalbullying event, thus raising the possibility that the previous Wndings are contaminated tounknown degrees by social desirability biases. Given the methodological diYculties of

Page 6: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

808 J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828

assessing children’s thoughts and feelings about victimization during actual conXict situa-tions, we have adopted a narrative methodology as an attempt to provide a more authenticpicture of the emotions associated with peer victimization experiences. In this methodol-ogy, we ask children to provide stories of actual victimization events from their own lives.Narratives of life events usually convey a richness of emotion and detail that is not cap-tured by traditional Likert-type questionnaires. Narrative methodology can also be lessbiasing in that participants are free to construe their experiences in a way that has the mostmeaning for them, rather than trying to accommodate the sometimes arbitrary concerns ofthe researcher.

Narrative analysis has been used successfully in the topic areas of interpersonal conXictand teasing, at least for adults (Baumeister, Stillwell, & Wotman, 1990; Bollmer, Harris,Milich, & Georgesen, 2003; Georgesen et al., 1999; Kowalski, 2000). In one such example,Georgesen et al. (1999) collected narratives of college students’ memorable teasing experi-ences. A wide range of incidents were reported, ranging from humorous teasing amongfriends to long-remembered and traumatic incidents of hostile taunting in childhood. Partic-ipants’ personality and teasing histories were also measured, and analyses focused on howthese variables were related to the students’ narratives. For the teaser narratives, studentswho reported teasing other children more often during their childhood were more likely toaccept blame for the incident in the teasing narrative. Participants who were rated as show-ing more empathy for the victim were also more likely to express remorse over their actions.With respect to the victim narratives, students who reported being a more frequent victim ofchildhood teasing were less likely to report forgiving the teaser. Interestingly, students with ahistory of more frequently teasing other children were more likely to forgive their teaser,perhaps because they were aware they had often been guilty of the same oVense. Victim for-giveness was also positively related to victim self-blame for the event.

These and other studies (e.g., Kowalski, 2000) indicate that adults can generatedetailed narratives of interpersonal conXict that are valid and relate to personality andother variables in meaningful ways. This methodology has not yet been used in studies ofchildhood peer victimization, however, and such an application is one of the major con-tributions of the current study. With respect to the current study, we oVer the followinghypotheses regarding the narratives told by our sample. Children who score higher onour bullying index should tell narratives that are more positive in nature; that is, theyshould enjoy the incident more and feel less remorse about it. Children who score higheron our victimization index should tell narratives that are more negative in tone; that is,they should be angrier at and less forgiving of the perpetrator and feel more distressedby the incident.

1.1.3. Contributions of a physiological approach to understanding peer victimizationAn additional major contribution of the current study is its measurement of physiologi-

cal arousal, as indexed by skin conductance levels, while children are telling theirnarratives. A theoretical basis exists for predicting diVerences between chronic victims andnon-victims in physiological arousal during conXict experiences. For example, severalresearchers have argued that one of the factors that predispose children to repeated victim-ization is that they experience a debilitating emotional arousal that precludes them fromcoping well during a victimization event (Perry et al., 2001). There is also evidence to sug-gest that aggressors are particularly drawn to victims who respond to provocation withemotional displays of distress and anger (Perry et al., 1990; Schwartz, 2000).

Page 7: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828 809

Thus, the literature suggests strongly that chronic victims should experience greateremotional arousal during victimization incidents. Although this hypothesis has not yetbeen tested directly empirically, indirect support is found in a study by Hubbard et al.(2001). In this study, children’s levels of reactive and proactive aggression were measuredand related to changes in skin conductance levels (SCL) while playing a game in whichthey lost to a confederate who cheated, a presumably anger-provoking incident. Analy-ses showed that reactive aggression, but not proactive aggression, was related positivelyto increases in SCL during the game. Because chronic victims are more likely to scorehigher on reactive aggression (Dodge, Lochman, Harnish, Bates, & Pettit, 1997), theseresults provide an encouraging basis for our hypothesis that children scoring high onvictimization in our sample should display greater gains in SCL while describing a vic-timization incident.

An argument can be made that increased emotional arousal will also be evident inbullies (see Scarpa & Raine, 1997, for a review of the relation between violent behavior,poor emotion regulation ability, and psychophysiological indicators). For example,Eisenberg et al. (1995) found in a longitudinal study that aggression in elementaryschool children could be predicted by displays of negative emotionality (reacting toprovocation with anger) and lack of emotion regulation in preschool; Shields and Cic-chetti (2001) similarly found in a sample of 8- to 12-year-old children that bullies scoredhigher on emotion dysregulation than a normal control group. An alternativeperspective is that children who engage in repeated bullying of peers do so becausethey Wnd it rewarding and enjoyable; as the Perry et al. (1990) study describedearlier notes, some bullies react positively upon seeing the distress and anger of theirvictims.

In sum, we arrive at a similar prediction for both bullies and victims: They should showgreater increases in physiological arousal when talking about their respective bullying andvictimization experiences compared to children who score lower on the bullying and victim-ization measures. However, we propose that this increase in arousal means something diVer-ent for bullies vs. victims. Because skin conductance levels indicate only the degree ofphysiological arousal and not its underlying nature (Dawson, Schell, & Filion, 2000), we willlook at the emotions expressed in the narratives to approach the question of what the arousalmeans. For victims, we predict the physiological arousal will be negative in nature and stem-ming from the fear, anger, and humiliation that accompanies such attacks. For bullies, on theother hand, we believe the arousal will be more positive in nature, reXecting an excitement or“rush” that develops in such children upon witnessing another child’s distress. Recent workby Owens and his colleagues supports the basis for this prediction; in qualitative analyses ofteenage girls’ use of relational aggression, they have found that a key motive for engaging inaggression was to create excitement in their lives (Owens, Shute, & Slee, 2000a, 2000b). Arse-nio’s work on the “happy victimizer” is also consistent with this interpretation; he and hiscolleagues have found that preschool children who exhibit happy facial expressions whileaggressing against other children were more likely to initiate aggression and experience peerrejection (Arsenio, Cooperman, & Lover, 2000; Arsenio & Lover, 1997).

We chose skin conductance levels as our index of physiological arousal for both theoreti-cal and methodological reasons. First, SCLs have been shown to be reliably related to expres-sions of emotion (Dawson et al., 2000), especially negative emotions (Hughes, Uhlmann, &Pennebaker, 1994). Second, SCL measurement is more robust and produces less noise duringconversation than does heart rate, which is inXuenced to a greater extent by the changes in

Page 8: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

810 J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828

respiration and chest volume that accompany speech, and which may reXect an orientingresponse more than emotional arousal (Hubbard et al., 2002).

1.1.4. Mediational hypothesesThe preceding sections describe the simple associations predicted by our model between

children’s personality and their aVective and cognitive reactions during victimization and theirsubsequent victimization and bullying. Our model goes beyond positing simple associations,however, to making mediational hypotheses: we predict that a child’s personality will deter-mine in part the responses he or she has to a victimization encounter, and that it is these imme-diate reactions that are the primary determinant of a child’s bullying or victimization status.

SpeciWcally, with respect to bullying, we predict that children who score low on Agree-ableness and Conscientiousness will be more likely to respond positively when they aggressagainst another child, feeling more enjoyment and less guilt and becoming more physiolog-ically aroused, and that these positive aVective and cognitive reactions will lead to higherlevels of bullying. With respect to victimization, we predict that children who score higheron Neuroticism and lower on Conscientiousness will react with greater anger and distresswhile being victimized, as well as showing greater physiological arousal, and these reac-tions will lead to higher overall levels of victimization.

2. Method

2.1. Participants

Participants were 99 children (50 boys, 49 girls) between the ages of 10 and 13(MD11.46, SDD0.97). Participants were recruited through a newspaper advertisement fora study of peer relationships. Families were paid $30 for their participation. The ethniccomposition of the sample reXected the demographics of the Lexington, Kentucky area,with 86% of the children being White, 12% Black, and 2% Other.

2.2. Procedure

2.2.1. Physiological recordingLeads for the physiological equipment were attached to two Wngers of the child’s non-

dominant hand. Skin conductance levels (SCL) data were recorded with a J + J I-330 comput-erized physiograph that integrated data at a rate of 50 samples per second and computed anaverage score for each experimental phase. Although SCL data were recorded continuously,we were interested primarily in four distinct phases: resting baseline, control narrative, bullynarrative, and victim narrative. The resting baseline consisted of the 3rd–7th min of a 10-minperiod during which the children were instructed to sit quietly while watching a videotape ofa saltwater aquarium. The three narratives are described later.

2.2.2. Children’s self-report dataChildren were asked to complete two self-report measures:1

1 Children were also asked to complete a measure of friendship quality, and parents were asked to complete theCBCL, for a study looking at the role of friendships in moderating peer victimization; these data are reportedelsewhere (Bollmer, Milich, Harris, & Maras, 2005).

Page 9: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828 811

(1) Perception of peer support scale (PPSS; Ladd & Kochenderfer-Ladd, 2002;Kochenderfer & Ladd, 1997). The PPSS is a 12-item self-report scale that asks chil-dren to indicate on a three-point response scale (“never,” “sometimes,” and “a lot”)the extent to which they encounter a range of victimization experiences. The measurebegins with four Wller items to familiarize the participant with the scale format; theseitems ask if there were ever times, for example, when the respondent had ice creamfor dessert or rode the bus to school. The four Wller items are followed by the ques-tion stem “Does anyone in your class ever:”, which is then followed by 22 items,including 12 victimization items that tap into both physical and relational victimiza-tion (e.g., “pick on you at school,” “call you names,” “say bad things about you toother kids at school,” “tease you at school,” and “beat you up at school.”) The 12 vic-timization items are interspersed with 10 positively worded Wller items (e.g., “say theylike your clothes” and “talk about school work with you”) so as to deXate the nega-tive aVect that might be aroused by responding to the items. The Wller items are notanalyzed. In the current sample, the PPSS had a coeYcient � of .81.

(2) Reactive-proactive aggression questionnaire (RPAQ; Raine et al., in press). TheRPAQ is a recently created scale that yields scores on reactive and proactive aggres-sion subscales, as well as a total aggression score. The measure consists of 23 items,measured on three-point scales (never, sometimes, often), that ask children torespond how often they engage in a certain behavior. For the current study, we wereinterested in the proactive aggression subscale only, which consisted of items such as“How often have you had Wghts with others to show who was on top?” “How oftenhave you used force to obtain money or things from others?” and “How often haveyou threatened and bullied someone?” The coeYcient � reported by Raine et al. (inpress) for the proactive subscale was .84, and the � for our sample was .70. Conver-gent validity of the proactive subscale has been documented through signiWcant cor-relations with the aggression, delinquency, and hostility-aggression subscales of theCBCL, and criterion validity was documented through signiWcant correlations withmaternal ratings of psychopathy, use of violent strong-arm tactics, initiating Wghts,and self-reported delinquency (Raine et al., in press).

2.2.3. Narrative tasksFollowing completion of the measures, the narrative tasks were introduced. Children

were told that we wanted them to make a personal diary about themselves, but insteadof writing, we would videotape them talking out loud. We Wrst had all children spendtwo minutes providing a control narrative by talking about their past school year. Wegave children a reminder sheet listing topics to talk about deliberately chosen to elicitfactual information that would be largely devoid of emotion; example topics includedwhat their teacher’s name was; what the classroom looked like; what they typically atefor lunch; what classes they had; and the kind of homework they did. We had no sub-stantive interest in the content of the school narrative; the primary purpose was simplyto get children accustomed to talking about themselves and using the reminder sheet, aswell as to provide a control for the physiological changes associated with talking outloud.

After the school narrative was Wnished, participants were asked to provide two narra-tives, one about an incident in which they were a victim of bullying or hostile teasing andone about an incident in which they were involved in teasing or bullying another child.

Page 10: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

812 J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828

Order of narrative was counterbalanced across participants. The instructions provided toparticipants for this task included a deWnition of victimization adapted from Olweus(1978); for example, the instructions for the victim narrative were:

All kids at some time or another experience problems with other kids. These prob-lems can include being bullied, picked on, or teased in a hostile manner. When wetalk about bullying or hostile teasing, it is diYcult for the person being bullied todefend him or herself. We also call it bullying when a student is teased repeatedly in amean and hurtful way. But we do not call it bullying when the teasing is done in afriendly and playful way. In other words, we are talking about times when kids dosomething really mean to another kid. We would like you to describe a time in yourlife when you experienced a problem like this with another kid or group of kids doingsomething mean to you, like bullying, making fun of, picking on, or teasing you in amean manner.

At this point the experimenter handed children a reminder sheet of points to cover whileproviding the narrative. We instructed them to talk for at least 2 min and delivered stan-dardized prompts if they stopped talking before the two minutes were up. Pilot testing ofthis procedure conWrmed that it was eVective in producing narratives that were complete interms of detail as well as being closely matched in terms of duration.

Instructions for the perpetrator narrative asked participants to describe “a time in yourlife when you did something mean to another kid or group of kids, like bullying, pickingon, or teasing them in a mean mannerƒ” If a participant claimed not to have a relevantperpetrator narrative, the experimenter delivered a standardized prompt stressing thateverybody has times in their lives when they have said or done something to hurt some-body else. These instructions and prompts were eVective, as only 1 of the 99 childrenclaimed he or she could not generate a perpetrator narrative and only two children wereunable to generate a victim narrative.

2.2.4. Structured interviewFollowing each narrative, participants completed a structured interview. Questions

were administered orally, with the experimenter presenting visual guides that providedthe response scale and appropriately phrased anchors. For the victim narrative, weasked how the participants felt during the event; how they thought the bully felt; andhow mad they feel at the bully now. Questions for the perpetrator narrative were simi-lar in content although rephrased to reXect the diVerent perspective. At the conclusionof both the narratives and structured interviews, the children were asked four questionson Wve-point scales about their frequency of victimization experiences as both victimand perpetrator: two questions about how often they tease other children and getteased, and two questions about how often they physically hurt other children or getphysically hurt.

2.2.5. Parent dataWhile the children were undergoing the above procedures, the parent was asked to com-

plete several questionnaires. Parent reports were obtained so as to avoid the interpreta-tional problems caused by relying solely on child reports. Considerable evidence existsattesting to the validity of parent reports of children’s internal states and abilities in gen-eral (Kamphaus & Frick, 1996) and parent reports of children’s personality and behavior

Page 11: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828 813

and peer diYculties in particular (Angold & Costello, 1996; Barbaranelli, Caprara,Rabasca, & Pastorelli, 2003; Bierman & McCauley, 1987). Moreover, research by Laddand Kochenderfer-Ladd (2002) supports the superiority of multi-informant indices of peervictimization over single source measures. The measures completed by parents in the cur-rent study were:

(1) Parent report of victimization and bullying. Several items asked parents to rate theirchild’s victimization experiences. The Wrst item asked “Does your child tease, pick on,or bully other children more or less often compared to other children your child’sage?” The second item asked, “Does your child get teased, picked on, or bullied byother children more or less often compared to other children your child’s age?” Par-ents rated both items using Wve-point scales with anchors of “much less than others”and “much more than others.” An additional Wve items, all on Wve-point Likertscales, asked parents more speciWcally (a) the extent to which their child hadexpressed distress over being teased or picked on by peers; (b) how often their childhad been the victim of physical attacks; (c) how often their child had been involved inteasing or picking on other children; (d) how often their child had been involved inphysical attacks on his or her peers; and (e) how often they had been contacted byothers complaining about their child’s interpersonal behavior toward peers.

(2) Little Five personality questionnaire (LFPQ; Lynam et al., 2005). The LFPQ is a 62-itempersonality questionnaire completed by parents regarding their child and is designed tomeasure the Big Five personality dimensions of Neuroticism, Agreeableness, Conscien-tiousness, Extraversion, and Openness to Experience in children and adolescents. Foreach item, respondents indicate on a Wve-point scale how characteristic the item is oftheir child. In our sample, coeYcient �s were .83 for Neuroticism, .88 for Agreeableness,.78 for Conscientiousness, .74 for Extraversion, and .78 for Openness to Experience,thus indicating that the LFPQ is reliable for children of our age range.

(3) Perception of peer support scale—Parent version (PPSS-P). We adapted the PPSS(Ladd & Kochenderfer-Ladd, 2002) for completion by parents. The basic item contentremained identical to the child version; items were merely rephrased from the perspec-tive of the parent responding about the child. For example, the item “Does anyone inyour class ever pick on you at school?” was rephrased “Does anyone in your child’sclass ever pick on your child at school?” and the item “Does anyone in your class eversay mean things to you?” was rephrased to “Does anyone in your child’s class ever saymean things to your child?” We also changed the three-point response scale used in thechild version to a Wve-point response scale (ranging from 1Dnever happens to my childto 5Dhappens very often to my child) in order to increase the sensitivity of the mea-sure. The coeYcient � for the current sample for the PPSS-P was .92.

2.2.6. Coding of narratives and structured interviewsFive judges, blind to the victim/bully scores of the children, observed the videotaped nar-

ratives and rated them on a number of dimensions. Dimensions were chosen because theywere of theoretical relevance and/or had been shown to be useful in previous studies in ourlaboratory. Judges were extensively trained prior to coding the tapes, through a process ofsupplying them with deWnitions and examples of each of the dimensions rated. For example,judges were asked to rate how much empathy the participant expressed toward the victimwhile telling the bullying narrative. Empathy was further deWned as “to what extent did the

Page 12: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

814 J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828

subject talk about how the victim must have felt during the incident?” Judges then practicedmaking ratings as a group and then individually on subset of the videotapes to ensure theywere agreeing suYciently in their ratings prior to rating the entire sample. Ratings were madeon Wve-point scales, with anchors of 1D“none” or “not at all” with respect to the dimensionbeing rated and 5D“complete” or “extreme” levels of the dimension.

Interrater reliabilities were computed using procedures given by Rosenthal and Rosnow(1991). This method involves computing the mean interrater correlation for each rating, andthen computing an eVective reliability coeYcient that corrects for the number of raters usingthe Spearman-Brown prophecy formula. The dimensions rated and eVective reliabilities areas follows: for victim narratives, the judges rated how much anger the children expressed(rD .74), how much they took responsibility for the incident and blamed themselves (rD .74),how much they blamed the perpetrator (rD .80), how much the children forgave the bully(rD .78), and how distressed the children appeared to be feeling at the time of the videotapingabout the incident (rD .81). For the bully narratives, coders rated how much guilt the childrenexpressed (rD .86), how much they seemed to enjoy the bullying (rD .80), how much they heldthe victim responsible for the incident (rD .84), how much empathy the children expressedtoward the victim (rD .84), how much they attempted to rationalize their bullying behavior(rD .65), and how vivid their narratives were (rD .80). Means of the Wve judges for each vari-able were then computed and used in subsequent analyses.

2.2.7. Creation of victimization and bullying scoresWe created composite variables of bullying and victimization to reduce the number of

variables and increase the reliability of measurement. We created the indices by standardiz-ing the items of the corresponding measures and computing a mean Z-score across itemsfor each child. The bullying composite measure consisted of (a) parents’ reports of howoften their child bullied others; (b) how often their child teased others; (c) how often theirchild physically attacks others; (d) how often they received complaints about their child’sinterpersonal behavior; (e) the child’s self-report of how often they teased others; (f) howoften they hurt others physically; and (g) the proactive anger subscale of the RPAQ. ThecoeYcient � for the bullying composite variable was .80.

The victimization composite measure consisted of (a) parents’ reports of how oftentheir child was teased by others; (b) the degree to which their child was distressed byteasing; (c) how often their child was the victim of attacks; (d) the adult version of thePPSS; (e) the child’s self-reports of how often they were the victim of teasing; (f) howoften they were physically hurt by peers; and (g) the child version of the PPSS. The vic-timization composite variable had a coeYcient � of .91. Our indices are therefore broad,reliable measures that reXect both verbal and physical bullying/victimization and bothchild and parent report.2

2 To assess the convergence between parent and child reports, mean Z-scores for the parent and child measuresof victimization and bullying were computed separately and then intercorrelated. We found that children’s andparents’ reports were signiWcantly correlated for both bullying, r (97)D .44, p < .05, and victimization experiences,r (97) D .45, p < .05. Thus, our data are consistent with other research showing good agreement between parentsand children (Ladd & Kochenderfer-Ladd, 2002). Moreover, Ladd and Kochenderfer-Ladd (2002) argue that across-informant composite helps reduce the impact of shared-method variance and provides a more accurate pre-dictor of relational adjustment than any single informant, providing further justiWcation for our decision to com-bine parent and child data in our composites.

Page 13: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828 815

The bullying and victimization composites were moderately intercorrelated,r (97) D .43, p < .05. As a check to ensure that we were justiWed in conceptualizing andanalyzing these composites separately, we entered all of the bullying and victimizationvariables comprising the composite variables into a factor analysis with varimaxrotation. Inspection of the scree plot indicated that a two-factor solution was optimal,with eigenvalues of 4.22 and 1.85, respectively. Inspection of the factor loadingsindicated that, with the exception of a single variable, the victimization items loaded onthe Wrst factor and the bullying items loaded on the second factor. The only item thatdid not load as predicted was the parent report of how often their child was thevictim of physical attacks, which loaded more strongly on the second factor. Becauseof its theoretical relevance to victimization, however, and because reliabilityanalysis indicated that including it increased the coeYcient �, we decided to retain it inthe victimization composite. In sum, the factor analysis supports the distinctionbetween bullying and victimization and our analyses of the composites as separatevariables.

3. Results

3.1. Overview of analyses

Our intensive laboratory-based protocol resulted in a broad, multimethod data setcontaining several distinct sets of variables. The framework adopted for our analyseswas to follow the outlines of the model shown in Fig. 1. First, we look at the relationsbetween bullying and victimization and the Wve personality dimensions. Next, we focuson the relations between bullying, victimization, and the aVective and cognitivevariables coded from the narratives. Third, we look at eVects of bullying and victimiza-tion on the physiological arousal of our participants while telling their narratives.Finally, we oVer a series of mediational analyses testing our hypotheses that the cogni-tive and aVective variables taken from the narrative and physiological measurementswould mediate the relation between personality and bullying and victimizationtendencies.

3.2. Personality predictors of bullying and victimization

Our Wrst set of analyses involved identifying personality variables that predict children’sexperiences with peer victimization as either the target or perpetrator.3 Table 1 shows thecorrelations between the Big 5 measures and the bullying and victimization composites. Aspredicted, children who scored higher on the bullying composite were described by theirparents as being less Agreeable and less Conscientious. The predictions that bullies wouldalso score higher on neuroticism and extraversion were not supported. In sum, the picture

3 We also conducted initial analyses examining gender and race (Black vs. White/Other) diVerences on the nar-rative variables, bullying and victimization composites, and physiological arousal. These analyses revealed onlyone signiWcant diVerence for gender (boys scored higher on the bullying composite than did girls, t (97)D 3.48,p < .05, r D .33), and one signiWcant diVerence for race (Black children reported feeling less good about themselveswhen telling a bullying narrative than did White children, t (96)D 2.61, pD .05, r D .26).

Page 14: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

816 J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828

of a bully that emerges from this analysis is a child who breaks rules, has less self-control,and behaves unpleasantly toward peers.4

With respect to personality correlates of victimization, as seen in Table 1, as predicted,children scoring higher on the victimization composite scored signiWcantly lower on Con-scientiousness. As predicted, victimization was also positively associated with Neuroticism.Although this correlation is not statistically signiWcant, given that we had predicted thisrelation a priori and the increased Type II error rate due to the relative low power of thestudy, we feel that this correlation warrants some attention. The correlations betweenAgreeableness, Extraversion, and victimization were not signiWcant.

3.3. Bullying and victimization as expressed in children’s narratives

In this set of analyses, our interest is in how children who are bullies and victimsdescribe speciWc bullying and victimization events. Correlations were computed betweenthe victimization and bullying composite variables and the intervening variables derivedfrom the victim and bully narratives, which include both the ratings made by blind judgesfrom the videotapes and the children’s self-reports obtained in the structured interviewsfollowing the narratives (see Table 2).

With respect to the victim narratives, we hypothesized that victimization scores wouldbe related to more negative aVect. Strong support was obtained for this hypothesis. Whenasked to talk about a time in which they were teased or bullied by another child, childrenwho scored higher on our index of peer victimization were rated as sounding signiWcantlyangrier at the bully, blaming the bully more for the incident, forgiving the bully less, andappearing to have been more distressed by the incident. They also told narratives that wererated by our judges as being more vivid. These impressions of our judges were additionallyconWrmed by the children’s self-reports in the structured interview. Children who scoredhigher on victimization reported feeling worse about the incident, being signiWcantlyangrier at the bully, and believing that the bully had felt good during the incident.

4 Although in the results and discussion we periodically use the terms “bullies” and “victims,” it is important tokeep in mind that our analyses adopted a dimensional approach, and we do not identify discrete subgroups ofbullies and victims. Rather, we use these terms as convenient shorthand labels to represent children scoring rela-tively high on these dimensions.

Table 1Relations between bullying and victimization composites and individual diVerence measures

¤ p < .10.¤¤ p < .05.

Individual diVerence measure Composite scores

Victimization Bullying

“Little Five” personality questionnaireAgreeableness ¡.14 ¡.36¤¤

Extraversion ¡.02 .10Neuroticism .18 ¤ .13Conscientiousness ¡.31¤¤ ¡.20¤¤

Openness to experience .08 ¡.03

Page 15: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828 817

With respect to the bullying narratives, we hypothesized that bullying scores would berelated to more positive aVect. Partial support was found for this hypothesis, as childrenscoring higher on the bullying index were rated by our judges as expressing less guilt aboutthe incident, and they reported feeling better during the bullying incident. Somewhat sur-prisingly, bullying was not signiWcantly related to rated enjoyment of the incident, ratedempathy, or narrative vividness. It is possible that social desirability concerns (childrenknow they are not supposed to express enjoyment in hurting others) attenuated the rela-tions for these variables.

In sum, even though they received identical instructions for the narratives, children withhistories of bullying or being victimized tell very diVerent stories about their experiences,and these diVerences fall into meaningful and predicted patterns. Victims tell stories of vic-timization that are full of negative aVect, involving both visible distress and anger directedat their perpetrators. Bullies, on the other hand, tell stories of bullying that minimize thenegative aVect involved or even put a positive spin on the incident; they felt better and lessguilty for their victims compared to children who scored lower on the bullying index. Theobtained pattern of results is noteworthy also in that it provides impressive evidence ofdiscriminant validity: Scores on the victimization composite variable are related tovariables coded from the victim narrative but not the bully narrative, and scores on the

Table 2Correlations between victim/bully composite scores and narrative variables

¤ p < .05.

Dependent variable Composite scores

Victimization Bullying

Victim narrative variablesRated by judgesVictim anger .34¤ .08Victim takes responsibility ¡.14 .00Victim blames bully .30¤ .08Victim forgiveness of bully ¡.31¤ ¡.09Victim distress .34¤ .06Vividness of narrative .25¤ .06

Children’s self-reportsHow good victim felt ¡.20¤ .04How good bully felt .26¤ .18How mad victim was .21¤ ¡.06

Bully narrative variablesRated by judgesBully guilt ¡.15 ¡.20¤

Bully enjoyment of incident .12 .11Victim was responsible .17 .14Bully empathy for victim ¡.15 ¡.11Bully rationalization for incident .10 .09Vividness of narrative .14 .07

Children’s self-reportsHow good bully felt .04 .28¤

How good victim felt .08 .05How sorry bully was ¡.07 ¡.16

Page 16: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

818 J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828

bullying composite variable are related to variables coded from the bully narrative but notthe victim narrative.

3.4. Physiological arousal while telling bullying and victimization narratives

3.4.1. Changes in arousal across narrativesThe preceding analyses focused on the thoughts and emotions of children as they

described incidents of bullying and victimization. Our next research question focused on thephysiological reactions of children while they are telling their stories. The Wrst step was todocument that the children became emotionally aroused when discussing bullying and vic-timization experiences and that this arousal was not merely an artifact due to the cognitiveand physiological demands of talking out loud. We thus conducted 2 (sex)£3 (experimentalphase: resting baseline, control narrative, and bully/victim narrative) repeated measuresANOVAs with a priori contrasts, computed separately for the bully and victim narratives.With respect to the bully narratives, the phase factor was highly signiWcant, F(2,192)D45.55,p < .05. A priori contrasts showed that children exhibited signiWcantly greater skin conduc-tance levels when they were telling a bully narrative (MD10.74) compared to both the restingbaseline period (MD9.27), contrast F(1,96)D71.08, p < .05, rD .65, and the control narrativeperiod (MD10.31), contrast F (1,96)D9.08, p < .05, rD .28.

Similar results were obtained for the victim narrative, with the omnibus phase factorbeing signiWcant at F (2, 190)D 45.93, p < .05. Contrasts revealed that children telling victimnarratives had signiWcantly higher SCL (MD10.82) than when they were in the restingbaseline period (MD9.40), contrast F (1,95)D 74.87, p < .05, rD .66, or when telling thecontrol narrative (MD10.43), contrast F (1,95)D 8.38, p < .05, rD .28. Children’s gender didnot signiWcantly aVect SCL in either analysis, nor did gender interact with the narrativephase factor, all Fs < 1.0.

These analyses thus provide reassuring evidence that our narrative instructions wereeVective in getting children to talk and think about emotionally tinged events. Talkingabout a time in which one was either victimized or engaged in victimization of anothersigniWcantly increased the amount of physiological arousal as indexed by SCL. More-over, we know that this arousal was not caused merely by the task demands of mentallycreating a narrative and telling it out loud, as the contrasts comparing the bully and vic-tim narratives to the control narrative, which involved similar cognitive demands, weresigniWcant.

3.4.2. Relations among arousal, bullying, victimization, and personalityTo look at the relations among physiological arousal, bullying, victimization, and per-

sonality, we next created residualized gain scores by regressing SCLs while telling either thebully or victim narrative on SCL while telling the control narrative. The residualized gainscores for the two narratives were then correlated with the victimization and bullying com-posites and personality and are shown in Table 3. It is worth noting that SCLs for the con-trol narrative were highly correlated with SCLs for both the bully and victim narratives,r (98)D .91, p < .05 and r (97)D .92, p < .05, respectively. Thus, in the analyses using theresidualized gain scores, our tests are stringent indeed; we are attempting to explain the smallamount of leftover variance in SCLs for the bully and victim narratives after partialling outthe over 80% of the variance that is attributable to individual diVerences in SCLs whiletalking.

Page 17: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828 819

As seen in Table 3, contrary to prediction, physiological arousal while telling the victimnarratives was not signiWcantly related to children’s status on the victimization composite.However, arousal while telling the bully narrative was signiWcantly positively related toscores on the bullying composite. Thus, our prediction was conWrmed that children whoconsistently engage in bullying others would experience more physiological arousal whiletalking or thinking about their bullying.

We had also predicted that change in physiological arousal should be related to the per-sonality variables theorized to be related to bullying and victimization. As shown in Table3, increased arousal while telling the bully narrative was negatively related to both Agree-ableness and Conscientiousness. Increased arousal while telling the victimization narrativewas also negatively related to Agreeableness and Conscientiousness. In addition, change inarousal for the victim narrative was negatively related to Extraversion. Our hypothesis thatarousal would be associated with Neuroticism for both narratives was not supported.

Looking at the individual diVerence correlates of changes in arousal thus helps to inter-pret the nature of these changes. With respect to the bullying narrative, the children whoshow the largest gains in arousal score higher on the traits that were shown earlier to bemost strongly related to the bullying composite, and so their arousal may reXect therewarding properties and their greater enjoyment of these incidents. This interpretation issupported by a signiWcant correlation between SCL residualized gain scores for the bully-ing narrative and ratings of how much the child enjoyed the bullying incident, r (92)D .28,p < .05. With respect to the victimization narrative, although change in arousal was not sig-niWcantly related to scores on the victimization composite, it was signiWcantly negativelyrelated to Agreeableness and Conscientiousness, personality traits that were themselvesrelated to victimization.

3.5. Joint eVects of bullying and victimization

Collecting measures of both victimization and bullying tendencies from the same sam-ple allows the question of overlap between victim and bully status to be raised. We corre-lated children’s scores on the bullying and victimization composites and found that theywere indeed signiWcantly related, r (97)D .43, p < .05. Our results are thus consistent with a

Table 3Personality predictors of changes in physiological arousal while telling bullying and victimization narratives

Note. Change in physiological arousal was indexed by the residualized gain scores obtained by regressing skinconductance levels while telling the bully or victim narratives on the control narrative SCL.¤ p < .05.

Individual diVerence measure Change in arousal

Victim narrative Bully narrative

Bullying composite .12 .22¤

Victimization composite .02 .11“Little Five” personality questionnaire

Agreeableness ¡.23¤ ¡.30¤

Extraversion ¡.22¤ ¡.17Neuroticism ¡.02 ¡.02Conscientiousness ¡.20¤ ¡.23¤

Openness to experience ¡.14 ¡.06

Page 18: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

820 J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828

review by Schwartz, Proctor, and Chien (2001), who report correlations ranging between.20 and .50 across studies that directly addressed the relation between aggression/bullyingand victimization.

Because our sample was not large enough to analyze distinct subgroups of bullies andvictims, we addressed the issue of bullying/victimization overlap through a series of regres-sion analyses in which the narrative and physiological arousal variables were regressed onthe bullying and victimization composite variables, entered in the Wrst step, and thebullying£victimization interaction term, entered in the second step. These analyses wereconducted using the standard Aiken and West (1991) procedure, whereby continuous pre-dictors were centered prior to analysis, and interactions were interpreted using predictedvalues 1SD below and above the mean.

These regressions revealed no signiWcant interactions between bullying and victimiza-tion for any of the bullying narrative and physiological variables, all ts < 1.7, ps > .09.SigniWcant bullying and victimization interactions, however, were found for three of theten victimization variables: (a) how much forgiveness the victims appeared to expresstoward the bully, t (91)D 2.15, p < .05, rD .24; (b) the children’s self-report of how goodthey felt during the incident, t (93)D 2.05, p < .05, rD .23; and (c) the children’s self-reportof how mad they felt at the bully, t (93)D 3.32, p < .05, rD¡.36. The nature of the interac-tion was identical for all three variables: no diVerences as a function of victimization forthe high bullying children, but signiWcant eVects as a function of victimization for thelow bullying children. In other words, children scoring high on both the victim and bul-lying composites behaved the same as the high bully low victim children, consistent withthe notion of a subgroup of aggressive victims who react similarly to the “pure” bullies.However, for children low on bullying, those scoring high on victimization reacted morenegatively (i.e., they forgave less, felt angrier, and felt worse about themselves) than didlow-victim children. In sum, these analyses suggest that the Wndings we report betweenvictimization and the narrative variables in Table 2 appear to be accounted for more by“pure” victimization in the traditional sense, rather than a subgroup of aggressivevictims.

3.6. Mediation analyses: Do narrative and physiological variables mediate the personality-bullying/victimization link?

The preceding analyses have shown that personality is related to bullying and victimiza-tion in meaningful ways. We have also shown that children’s narratives of peer victimiza-tion events, and their physiological arousal while telling these events, are related tobullying and victimization. In our Wnal set of analyses, we address the question of primaryimportance: To what extent do the narrative and physiological arousal variables mediatepersonality–peer victimization relations?

We adopted the following strategy in conducting our mediational analyses. First, to reducethe number of potential analyses conducted, we did not test mediational analyses for all ofthe Big 5 factors, only those for which we had either a priori predictions or which had beenshown in our initial correlational analyses to be signiWcantly associated with bullying and vic-timization. Second, we decided to analyze only those mediators that came from a diVerentsource than either the personality or outcome variables, namely, the six variables coded byobjective judges and the residualized gain scores reXecting change in physiological arousal.Although this removed several potential mediators from consideration, it reduced the total

Page 19: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828 821

number of analyses involved and had the beneWt of ensuring that any signiWcant mediatingeVects obtained could not be attributed to shared method variance.

The statistical procedure we used to test mediation was the product of coeYcientsapproach recommended by MacKinnon and colleagues in a recent article that reviewedand compared diVerent methods (including the more common Baron and Kennyapproach) of testing mediation (MacKinnon, Lockwood, HoVman, West, & Sheets, 2002).In the product of coeYcients approach, a series of regression analyses are run whereby themediating variable is regressed on the independent variable, and the dependent variable isregressed on both the independent and mediating variable. Then, the regression coeYcientsrepresenting the IV-mediator and mediator-DV links are converted into Z scores, and theproduct of the two Z-scores (ZAZB) is computed and compared against the distribution ofthe product of two normal random variables (MacKinnon et al., 2002), with the criticalvalue of the distribution for p < .05 being 2.18.

Looking Wrst at the mediators of victimization, none of the narrative or physiologicalvariables appeared to mediate the Agreeableness–victimization relation, all ZAZB 6 1.85.However, several signiWcant mediators of the Conscientiousness–victimization and Neu-roticism–victimization relations were obtained. With respect to Conscientiousness, vari-ables that emerged as signiWcant mediators included the degree of anger displayed in thenarrative (ZAZBD4.71), how distressed the children appeared to objective raters while tell-ing the narrative (ZAZBD 6.59), how much they blamed the bully (ZAZBD 4.61), and theirdegree of forgiveness of the bully (ZAZBD6.03), all ps < .05.

Consistent with the idea of neuroticism as reXecting negative emotionality, signiWcantmediators of the Neuroticism–victimization relation included degree of anger expressed atthe bully (ZAZBD 5.55), blaming the bully (ZAZBD2.26), and ratings of distress while tell-ing their narratives (ZAZBD3.65), all ps < .05.

Less evidence for mediation was found in the analyses of the bullying composite vari-able, where signiWcant results were obtained for only two of the mediating paths. Chil-dren’s feelings of guilt appear to mediate the relation between Agreeableness and bullying,such that highly agreeable children feel more guilty when they bully other children andtherefore engage in bullying to a lesser degree, ZAZBD2.18, p < .05. Increases in physiolog-ical arousal while telling the bullying narrative also appears to mediate the relationbetween Conscientiousness and bullying, such that children who score low on conscien-tiousness showed greater gains in skin conductance levels while telling the bullying narra-tive, which in turn was related to higher levels of bullying, ZAZBD2.80, p < .05.

In sum, our mediational analyses support the hypothesized model that for victimization,and bullying to a lesser degree, the inXuence of personality on subsequent victimizationand bullying is accounted for by children’s aVective responses to the event. In the case ofvictimization, the important mediators appear to be related to negative emotionality ingeneral, including distress, lack of forgiveness, anger at the bully, and blaming the bully. Inthe case of bullying, the crucial mediators appear to be a lack of guilt and an increasedphysiological arousal that appears to reXect a positive reaction to the event.

4. Discussion

Through either hard experience in our own childhoods or popular culture, most of uscan easily summon a mental picture of childhood bullies and their victims. Research onpeer victimization in recent years has focused on developing a more scientiWc description

Page 20: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

822 J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828

of this picture. The current study proposed and tested a mediational model of bullying andvictimization that extends existing research on the correlates of peer victimization throughproviding a fuller understanding of the underlying mechanisms involved in peer victimiza-tion.

4.1. Understanding the “Who” in peer victimization: Personality predictors and narratives of bullying and victimization

4.1.1. Who are the bullies?Our study is one of the Wrst to approach the question of individual diVerences in

bullying and victimization using a structural model of personality. Using a recentlydeveloped measure of the “Big 5” traits in childhood, we discovered that bullies scoredlower on Agreeableness and Conscientiousness. Individuals with this proWle tend to berebellious, hostile, lacking in sympathy and warmth, antisocial, impulsive, self-cen-tered, manipulative, and confrontational. Whereas previous research has tended tofocus on aggressive, conduct-disordered behaviors displayed by bullies, thecurrent study is one of the Wrst to anchor bullying in the broader, dispositionaltraits that may predispose children to antisocial behavior and help to explain theirbehavior.

Examination of the narratives the bullies told provides insight into what might beaccounting for the personality–bullying relations. The lack of remorse and apparentglee expressed by some of our participants while telling a bullying narrative was chillingto observe. This was seen in the objective coding of the narratives as well as the struc-tured interview following the narrative, both of which revealed that bullies felt lessguilty and sorry about their behavior and reported feeling better during the incident.Such an interpretation was supported by our mediational analyses, which found thatthe relation between Conscientiousness and bullying was mediated by lower levels ofguilt and greater physiological arousal while telling the narrative. The followingexcerpt from a bullying narrative given by one of our participants captures these char-acteristics well:

I kind of did an accident on someone that I really hate. It was at gym class duringaround the last day of the yearƒTravis had kicked a ball toward us (the kid I hatethe most). We were playing a game of scoop ball. He asked, ‘Hey, give us the ballback.’ And he turned around for a couple of seconds to answer what his friends wassaying. I picked up the ball and chucked it at him. Fortunately and unfortunately, itconked him on the head. He ran to the gym teacher crying. I felt kind of happy untilTravis told the gym teacher.

Past research on peer aggression has similarly implicated bullies’ positive aVect whileaggressing as being particularly problematic. For example, Arsenio’s work on the“happy victimizer” has shown that preschool children who display facial expressions ofhappiness while aggressing against a peer were more likely to be the initiators of aggres-sion, were less accepted by peers, and were judged as less socially skilled by their teachers(Arsenio et al., 2000). Our Wndings extend this work to a middle-school population, aswell as documenting that the positive aVect experienced by some bullying children playsan important mediating role in accounting for the relation between personality and sub-sequent bullying.

Page 21: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828 823

4.1.2. Who are the victims?Victimization in our study was signiWcantly related to low Conscientiousness and,

although it did not reach statistical signiWcance, high Neuroticism. Although not statisti-cally signiWcant, the magnitude and direction of the Neuroticism association was predictedby our model and thus worthy of further investigation. Obviously, ascertaining the direc-tion of causality is impossible with our correlational data. Certainly it is plausible thatchildren who exhibit lots of angry emotionality would elicit bullying by peers, but it isequally plausible that being chronically victimized would lead a child to express angryemotionality.

The association between victimization and Conscientiousness is less intuitive. Whyshould low Conscientious children be more likely to be victimized? One possibility is thatthese children reXect the subgroup of aggressive victims in our sample. Another possibility,raised earlier, is that conscientious children, who possess traits of determination and beingstrong-willed, are better able to resist bullying from peers. It is also possible that the nega-tive aVect associated with Neuroticism creates a particular vulnerability to victimizationwhen it is combined with the lack of restraint associated with low Conscientiousness. Chil-dren scoring high on Neuroticism and low on Conscientiousness may be especially proneto being unable to regulate their behavior in a conXict situation, thereby reacting in a waythat may exacerbate the situation. Unfortunately, our sample was too small to enable iden-tiWcation of a discrete group of high-N and low-C children; it would be valuable to exploresuch subtype analyses in future research.

When providing their narratives, victims were also judged to be angrier at the perpetra-tor, blaming him or her more, and being less forgiving. Our mediational analyses furtherconWrmed the pivotal role these variables play in the victimization process. These Wndingssuggest that ruminating over a victimization event and holding a grudge in particular,although understandable, may be especially counterproductive. Rather, in our sample,children who were high on Conscientiousness, and therefore are more reXective and delib-erate in their interactions, seemed better able to recognize the beneWts of forgiving the bullyor at least of letting go of the negative experience and were in fact victimized less often.This Wnding is consistent with recent work on the importance of forgiveness in victims’recovery from traumatic experiences at the hands of others (Witvliet, Ludwig, &Vander-Laan, 2001). Although our data are correlational in nature, they raise the intrigu-ing question of whether an intervention designed to help victims let go of the pain andresentment they feel regarding the incident could actually result in fewer subsequent vic-timization experiences.

Our Wndings also reinforce the idea in the literature that how a victim responds visiblyto the provocation may be critical in determining whether the attack persists or reoccurs.When parents and teachers advise children to “just ignore it,” their well-meant but oftenimpossible-to-follow advice captures this idea that a child who shows distress upon beingteased or physically attacked is actually inviting further abuse. A narrative provided byone of our participants illustrates this point vividly:

Okay, it was at schoolƒmy friends and I playing a game of freeze tag. I stink at, Istink at running so when I was ‘It,’ I couldn’t really catch anyoneƒ.Well, I was tryingto run away from whoever was It or posing to be It. I ran right near one of my goodfriends and, puV, he was It. Getting me frozen. People kept on tagging me even whenI told them not to. Then I, uh, it happened. I started crying, whining, trying to get

Page 22: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

824 J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828

them to stop freezing me. I even pleaded at some point. I was feeling really angry andsad at that moment because no one would leave me alone. I told Mom and Dadabout it but they said just to avoid it. I think it happened because they all know thatit is more fun to pick on me.

4.2. Physiological bases of bullying and victimization

Some of our most intriguing results concerned the role of physiological arousal in chil-dren while telling a bullying narrative. Our analyses revealed that gains in arousal wereassociated with children’s scores on the bullying composite variable; moreover, physiologi-cal arousal mediated the associations among Conscientiousness, Agreeableness, and bully-ing. Although skin conductance levels cannot be unambiguously interpreted by themselves,some of our other results provide important clues for understanding what the increase inphysiological arousal for the bullies means. The gain in arousal, rather than reXecting anx-iety or guilt over the incident, appears instead to reXect a meaner set of emotions: The chil-dren who show this arousal are less Agreeable and Conscientious in terms of personalityand they appear to naive coders to have enjoyed their bullying of another child.

This interpretation points to some troubling implications for intervention eVortsdesigned to reduce the frequency and severity of peer victimization, especially in light ofthe personality traits shown to be related to bullying in our sample. Children who are notmotivated to treat others well or to follow rules are not likely to respond well to appealsfor better behavior. Indeed, our results may help to explain why interventions for conduct-disordered children are often not eVective (Kazdin, 2001). Such interventions, which tendto rely on training adaptive cognitive and social skills in the children or training parents onbetter behavior management practices, may fail because they do not recognize that chil-dren who engage in bullying receive considerable internal positive reinforcement from theiracts. If it feels good to bullies to engage in these acts, intervention eVorts need to bedirected toward making it feel bad to act cruelly toward another child.

4.3. Strengths, limitations, and directions for future research

The current study possesses several important strengths. We used a multi-pronged strat-egy capitalizing on a variety of methodological approaches: self- and parent-report, behav-ioral observations, participant narratives, structured interviews, and physiological data.One of the most important advantages of our design is that it is unlikely our obtained rela-tions are an artifact of common method variance, and indeed, we deliberately tested onlythose mediational pathways that would be uncontaminated by shared method variance.Parents’ judgments of their children’s victimization histories and personality were relatedto stories that the children told about speciWc incidents (many of which the parents werecompletely unaware), which were related to objective judgments made by coders who wereblind to the children’s victimization histories, which were related to the children’s physio-logical arousal while telling the stories. The convergence of our results across such widelyvarying methodologies is impressive indeed and increases greatly our conWdence in ourWndings.

The comprehensiveness of our data collection procedures, however, necessarily resultedin a limitation in our procedures, which is that the sample was not large enough to permitstructural equations analysis that might have shed additional light on underlying peer

Page 23: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828 825

victimization processes. Furthermore, the cross-sectional nature of our design precludesdrawing Wrm causal inferences regarding the temporal ordering of our variables.

Another consequence of our sample size and recruiting procedures is that we wereunable to select children at the extremes of bullying and victimization. As a result, we werenot able to analyze diVerences among subtypes of victims or bullies. The correlation of .43we obtained between our bullying and victimization composites suggests that at least someof our victims could be considered “provocative victims,” that is, children whose ownaggressive behavior prompts retaliation from others. We do not view this limitation as seri-ous, as the regression analyses we conducted that included the interaction of bullying andvictimization oVered reassuring evidence that our results were not solely attributable to thesubgroup of provocative victims. Further, our data collection focused on the two majorroles of bully and victim; there are other roles children may assume (e.g., bystander,defender) that should be addressed in future studies.

A related limitation is that our recruitment method (advertising in the newspaper) did notallow us to collect sociometric or peer nomination data, generally considered the optimalmethod of measuring bully/victim status but which requires data collection in classrooms.Although the literature in general suggests adequate agreement among self- and other-reports of victimization (Ladd & Kochenderfer-Ladd, 2002; Pellegrini, 2001), future researchmight wish to include peer nomination measures of bullying and victimization. In particular,it is possible that the parent- and self-reports of bullying obtained in the current study,although suYciently reliable, underestimated the degree of bullying engaged in by our partic-ipants. We do not view this limitation as being especially problematic, however, as the eVectof underestimating bullying should only make it harder to detect signiWcant relations.

4.4. Concluding thoughts

Bullying is a part of most children’s lives, but unlike other childhood rites of passage,bullying possesses few redeeming features. No child should have to live in fear of being bul-lied, and for society’s sake if not their own, no child should revel in the abuse they wreakon other children. By studying the factors that are associated with peer victimization froma variety of perspectives and methodologies, we hope to improve our understanding ofwhat makes children hurt other children. Finally, we think it is important to stress that wedo not wish to be seen as engaging in a form of ‘blaming the victim’ or suggesting that vic-tims are somehow responsible for bringing about their mistreatment. Our goal was tounderstand better the vulnerability factors that increase a child’s risk for victimization orengaging in bullying behaviors. While we believe that victims’ reactions can in some casesbe maladaptive, in the sense of reinforcing bullies and increasing the likelihood of futurevictimization, this does not absolve bullies of the blame or responsibility for their cruelbehavior. Identifying aspects of victims’ behavior that increases their subsequent victimiza-tion is oVered instead in the hopes that this knowledge can be used in designing interven-tions for peer victimization, so that more children’s childhoods can be as carefree as ournostalgic memories would have them be.

References

Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA:Sage Publications.

Page 24: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

826 J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828

Angold, A., & Costello, E. J. (1996). The relative diagnostic utility of child and parent reports of oppositional deW-ant behaviors. International Journal of Methods in Psychiatric Research, 6, 253–259.

Arsenio, W. F., Cooperman, S., & Lover, A. (2000). AVective predictors of preschoolers’ aggression and peeracceptance: Direct and indirect eVects. Developmental Psychology, 36, 438–448.

Arsenio, W. F., & Lover, A. (1997). Emotions, conXicts, and aggression during preschoolers’ freeplay. British Jour-nal of Developmental Psychology, 15, 531–542.

Barbaranelli, C., Caprara, G. V., Rabasca, A., & Pastorelli, C. (2003). A questionnaire for measing the Big Five inlate childhood. Personality and Individual DiVerences, 34, 645–664.

Baumeister, R. F., Stillwell, A., & Wotman, S. R. (1990). Victim and perpetrator accounts of interpersonal conXict:Autobiographical narratives about anger. Journal of Personality and Social Psychology, 59, 994–1005.

Bierman, K. L., & McCauley, E. (1987). Children’s descriptions of their peer interactions: Useful information forclinical child assessment. Journal of Clinical Child Psychology, 16, 9–18.

Bollmer, J. M., Harris, M. J., Milich, R., & Georgesen, J. C. (2003). Taking oVense: EVects of personality and teas-ing history on behavioral and emotional reactions to teasing. Journal of Personality, 71, 557–603.

Bollmer, J. M., Milich, R., Harris, M. J., & Maras, M. (2005). A friend in need: Friendship quality, internalizing/externalizing behavior, and peer victimization. Journal of Interpersonal Violence, 20, 701–712.

Borg, M. G. (1998). The emotional reaction of school bullies and their victims. Educational Psychology, 18, 433–444.

Caspi, A., & Bem, D. J. (1990). Personality continuity and change across the life course. In L. A. Pervin (Ed.),Handbook of personality: Theory and research (pp. 549–575). New York: Guilford Press.

Connolly, I., & O’Moore, M. (2003). Personality and family relations of children who bully. Personality and Indi-vidual DiVerences, 35, 559–567.

Dawson, M. E., Schell, A. M., & Filion, D. L. (2000). The electrodermal system. In J. T. Cacioppo, L. G. Tassinary,& G. G. Berntson (Eds.), Handbook of psychophysiology (2nd ed., pp. 200–223). Cambridge: Cambridge Uni-versity Press.

Dodge, K. A., Lochman, J. E., Harnish, J., Bates, J., & Pettit, G. (1997). Reactive and proactive aggression inschool children and psychiatrically impaired chronically assaultive youth. Journal of Abnormal Psychology,106, 37–51.

Dodge, K. A., & Pettit, G. S. (2003). A biopsychosocial model of the development of chronic conduct problems inadolescence. Developmental Psychology, 39, 349–371.

Ehrler, D. J., Evans, J. G., & McGhee, R. L. (1999). Extending Big-Five Theory into childhood: A preliminaryinvestigation into the relationship between Big-Five personality traits and behavior problems in children. Psy-chology in the Schools, 36, 451–458.

Eisenberg, N., Fabes, R. A., Murphy, B., Maszk, P., Smith, M., & Karbon, M. (1995). The role of emotionality andregulation in children’s social functioning: A longitudinal study. Child Development, 65, 109–128.

Georgesen, J. C., Harris, M. J., Milich, R., & Young, J. (1999). “Just teasingƒ”: Personality eVects on perceptionsand life narratives of childhood teasing. Personality and Social Psychology Bulletin, 25, 1254–1267.

Glover, D., Gough, G., Johnson, M., & Cartwright, N. (2000). Bullying in 25 secondary schools: Incidence, impact,and intervention. Educational Research, 42, 141–156.

Hodges, E. V. E., & Perry, D. G. (1999). Personal and interpersonal antecedents and consequences of victimizationby peers. Journal of Personality and Social Psychology, 76, 677–685.

Hubbard, J. A., Dodge, K. A., Cillessen, A. H. N., Coie, J. D., & Schwartz, D. (2001). The dyadic nature of socialinformation processing in boys’ reactive and proactive aggression. Journal of Personality and Social Psychol-ogy, 80, 268–280.

Hubbard, J. A., Smithmyer, C. M., Ramsden, S. R., Parker, E. H., Flanagan, K. D., Dearing, K. F., et al. (2002).Observational, physiological, and self-report measures of children’s anger: Relations to reactive versus proac-tive aggression. Child Development, 73, 1101–1118.

Hughes, C. F., Uhlmann, C., & Pennebaker, J. W. (1994). The body’s response to processing emotional trauma:Linking verbal text with autonomic activity. Journal of Personality, 62, 565–585.

Jensen-Campbell, L. A., Adams, R., Perry, D. G., Workman, K. A., Furdella, J. Q., & Egan, S. K. (2002). Agree-ableness, extraversion, and peer relations in early adolescence: Winning friends and deXecting aggression.Journal of Research in Personality, 36, 224–251.

Juvonen, J., & Graham, S. (2001). Peer harassment in school: The plight of the vulnerable and victimized. NewYork: The Guilford Press.

Kamphaus, R. W., & Frick, P. J. (1996). Clinical assessment of child and adolescent personality and behavior. Bos-ton: Allyn and Bacon.

Page 25: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828 827

Kazdin, A. E. (2001). Treatment of conduct disorders. In J. Hill & B. Maughan (Eds.), Conduct disorders in child-hood and adolescence (pp. 408–448). New York: Cambridge University Press.

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

Kowalski, R. M. (2000). ”I was only kidding!”: Victims’ and perpetrators’ perceptions of teasing. Personality andSocial Psychology Bulletin, 26, 231–241.

Ladd, G. W., & Kochenderfer-Ladd, B. (2002). Identifying victims of peer aggression from early to middle child-hood: Analysis of cross-informant data for concordance, estimation of relational adjustment, prevalence ofvictimization, and characteristics of identiWed victims. Psychological Assessment, 14, 74–96.

Lemerise, E. A., & Arsenio, W. F. (2000). An integrated model of emotion processes and cognition in social infor-mation processing. Child Development, 71, 107–118.

Lowenstein, L. F. (1978). The bullied and nonbullied child: A contrast between the popular and unpopular child.Home and School, 12, 3–4.

Lynam, D. R., Caspi, A., MoYtt, T. E., Raine, A., Loeber, R., & Stouthamer-Loerber, M. (2005). Adolescentpsychopathy and the Big Five: Results from two samples. Journal of Abnormal Child Psychology, 33, 431–433.

MacKinnon, D. P., Lockwood, C., HoVman, J. M., West, S. G., & Sheets, V. (2002). A comparison of methods totest mediation and other intervening variable eVects. Psychological Methods, 7, 83–104.

Marsh, H. W., Parada, R. H., Craven, R. G., & Finger, L. (2004). In the looking glass: A reciprocal eVects modelelucidating the complex nature of bullying, psychological determinants, and the central role of self-concept. InC. E. Sanders & G. D. Phye (Eds.), Bullying: Implications for the classroom (pp. 63–106). New York: ElsevierAcademic Press.

McMahon, R. J., & Wells, K. C. (1998). Conduct problems. In E. J. Mash & R. A. Barkley (Eds.), Treatment ofchildhood disorders (2nd ed., pp. 111–207). New York: Guilford.

Menesini, E., Sanchez, V., Fonzi, A., Ortega, R., Costabile, A., & LoFeudo, G. (2003). Moral emotions and bully-ing: A cross-national comparison of diVerences between bullies, victims, and outsiders. Aggressive Behavior,29, 515–530.

Miller, J. D., & Lynam, D. (2001). Structural models of personality and their relation to antisocial behavior: Ameta-analytic review. Criminology, 39, 765–792.

Mynard, H., & Joseph, S. (1997). Bully/victim problems and their association with Eysenck’s personality dimen-sions in 8 to 13-year-olds. British Journal of Educational Psychology, 67, 51–54.

Nansel, T. R., Overpeck, M., Pilla, R. A., Ruan, W. J., Simons-Morton, B., & Scheidt, P. (2001). Bullying behaviorsamong US youth: Prevalence and association with psychosocial adjustment. Journal of the American MedicalAssociation, 285, 2094–2100.

Neuman, J. H., & Baron, R. A. (2003). Social antecedents of bullying: A social interactionist perspective. In S. Ein-arsen, H. Hoel, D. Zapf, & C. L. Cooper (Eds.), Bullying and emotional abuse in the workplace (pp. 185–202).London: Taylor & Francis.

Olweus, D. (1978). Aggression in the school: Bullies and whipping boys. Oxford, England: Hemisphere.Olweus, D. (1995). Bullying or peer abuse at school: Facts and intervention. Current Directions in Psychological

Science, 4, 196–200.Olweus, D. (2003). Bully/victim problems in school: Basic facts and an eVective intervention pro-gramme. In S.

Einarsen, H. Hoel, D. Zapf, & C. L. Cooper (Eds.), Bullying and emotional abuse in the workplace: Internationalperspectives in research and practice (pp. 62–78). London: Taylor & Francis.

Owens, L., Shute, R., & Slee, P. (2000a). ”Guess what I just heard!”: Indirect aggression among teenage girls inAustralia. Aggressive Behavior, 26, 67–83.

Owens, L., Shute, R., & Slee, P. (2000b). “I’m in and you’re outƒ”: Explanations for teenage girls’ indirect aggres-sion. Psychology, Evolution, and Gender, 2, 19–46.

Patterson, G. R., Littman, R. A., & Bricker, W. (1967). Assertive behavior in children: A step toward a theory ofaggression. Monographs of the Society for Research in Child Development, 32, 1–43.

Pellegrini, A. D. (2001). Sampling instances of victimization in middle school. In J. Juvonen & S. H. Graham(Eds.), Peer harassment: The plight of the vulnerable and victimized (pp. 125–144). New York: GuilfordPress.

Perry, D. G., Hodges, E. V. E., & Egan, S. K. (2001). Determinants of chronic victimization by peers: A review anda new model of family inXuence. In J. Juvonen & S. H. Graham (Eds.), Peer harassment: The plight of the vul-nerable and victimized (pp. 73–104). New York: Guilford Press.

Perry, D. G., Williard, J. C., & Perry, L. C. (1990). Peers’ perceptions of the consequences that victimized childrenprovide aggressors. Child Development, 61, 1310–1325.

Page 26: Reactions to bullying and peer victimization: Narratives, physiological arousal, and personality

828 J.M. Bollmer et al. / Journal of Research in Personality 40 (2006) 803–828

Raine, A., Dodge, K., Loeber, R., Kopp, L., Lynam, D., Reynolds, C., Stouthamer-Loeber, M. (in press). TheReactive-Proactive Aggression (RPQ) Questionnaire: DiVerential correlates of reactive and proactive aggres-sion in adolescent boys. Aggressive Behavior.

Rosenthal, R., & Rosnow, R. L. (1991). Essentials of behavioral research: Methods and data analysis (2nd ed.). NewYork: McGraw-Hill.

Sanders, C. E., & Phye, G. D. (2004). Bullying: Implications for the classroom. New York: Elsevier Academic Press.Scarpa, A., & Raine, A. (1997). Psychophysiology of anger and violent behavior. Psychiatric Clinics of North

America, 20, 375–394.Schwartz, D. (2000). Subtypes of victims and aggressors in children’s peer groups. Journal of Abnormal Child Psy-

chology, 28, 181–192.Schwartz, D., McFadyen-Ketchum, S., Dodge, K. A., Pettit, G. S., & Bates, J. E. (1999). Early behavior problems

as a predictor of later peer group victimization: Moderators and mediators in the pathways of social risk.Journal of Abnormal Child Psychology, 27, 191–201.

Schwartz, D., Proctor, L., & Chien, D. H. (2001). The aggressive victim of bullying: Emotional and behavioral dys-regulation as a pathway to victimization by peers. In J. Juvonen & S. H. Graham (Eds.), Peer harassment: Theplight of the vulnerable and victimized (pp. 147–174). New York: Guilford.

Shields, A., & Cicchetti, D. (2001). Parental maltreatment and emotion dysregulation as risk factors for bullyingand victimization in middle childhood. Journal of Clinical Child Psychology, 30, 349–363.

Slee, P. T., & Rigby, K. (1993). Australian school children’s self appraisal of interpersonal relations: The bullyingexperience. Child Psychiatry and Human Development, 23, 273–282.

Smith, P. K., Bowers, L., Binney, V., & Cowie, H. (1993). Relationships of children involved in bully/victim prob-lems at school. In S. Duck (Ed.), Learning about relationships (pp. 184–204). Thousand Oaks, CA: Sage Publi-cations.

Swearer, S. M., Song, S. Y., Cary, P. T., Eagle, J. W., & Mickelson, W. T. (2001). Psychosocial correlates in bullyingand victimization: The relationship between depression, anxiety, and bully/victim status. In R. A. GeVner, M.T. Loring, & C. Young (Eds.), Bullying behavior: Current issues, research, and interventions (pp. 95–121). Bing-hamton, NY: The Haworth Press.

Tani, F., Greenman, P. S., Schneider, B. H., & Fregoso, M. (2003). Bullying and the Big Five: A study of childhoodpersonality and participant roles in bullying incidents. School Psychology International, 24, 131–146.

Witvliet, C., Ludwig, T. E., & Vander-Laan, K. L. (2001). Granting forgiveness or harboring grudges: Implicationsfor emotion, physiology, and health. Psychological Science, 12, 117–123.