You are on page 1of 12

J Abnorm Child Psychol (2016) 44:63–74

DOI 10.1007/s10802-015-9974-1

Narcissism, Bullying, and Social Dominance in Youth:
A Longitudinal Analysis
Albert Reijntjes & Marjolijn Vermande &
Sander Thomaes & Frits Goossens & Tjeert Olthof &
Liesbeth Aleva & Matty Van der Meulen

Published online: 3 February 2015
# The Author(s) 2015. This article is published with open access at

Abstract A few previous studies have shown that narcissistic indirect bullying. Hence, high narcissism is a risk factor for
traits in youth are positively associated with bullying. However, bullying in boys, but not in girls. However, narcissism is not
research examining the developmental relationship between always accompanied by high bullying, given that many boys on
narcissism and bullying is lacking. Moreover, it is unclear the high bullying trajectories were not high in narcissism.
whether narcissists constitute a homogeneous group and Results show that among narcissistic youth only those who
whether the bullying of narcissistic youth results in establishing engage in high levels of bullying are high in social dominance.
social dominance over peers. The present work addresses these
gaps. Children (N=393; Mage =10.3; 51 % girls) were followed Keywords Narcissism . Bullying . Social dominance . Joint
during the last 3 years of primary school. Person-centered anal- trajectory analysis . Gender differences
yses were used to examine whether groups with distinct devel-
opmental trajectories for narcissism and two bullying forms
(direct and indirect) can be identified, and how these trajectories
are related. Multiple groups emerged for all constructs exam- During the past decades, researchers have increasingly ac-
ined. For girls, higher narcissism was neither related to more knowledged that bullying is a strategic attempt to acquire a
intense bullying, nor to higher social dominance. In contrast, central, powerful and dominant position in the peer group
highly narcissistic boys were more likely than their peers to (e.g., Olthof et al. 2011; Salmivalli and Peets 2009). For in-
show elevated direct bullying, and in particular elevated stance, Farrington (1993) observed that when asked Bwhy do
you bully?^, the most frequently reported answers are Bto feel
Electronic supplementary material The online version of this article powerful^ and Bto look cool^. Moreover, in early adolescence
(doi:10.1007/s10802-015-9974-1) contains supplementary material, bullies score significantly higher on status, power, and pres-
which is available to authorized users. tige goals than do their peers (Sijtsema et al. 2009).
A. Reijntjes (*) : M. Vermande The motivation of bullies to gain power, dominance, and
Department of Pedagogical and Educational Sciences, prestige over others suggests that elevated narcissism might be
Utrecht University, PO Box 80150, Utrecht, The Netherlands a contributing factor. Narcissism is a dispositional trait that
involves a sense of entitlement of privileged status over
S. Thomaes others, the belief that one is unique and more important than
University of Southampton, Southampton, UK others, and an excessive need for approval and admiration
from others to feed the grandiose - but ultimately vulnerable
F. Goossens : T. Olthof
- self (Miller et al. 2007; Morf and Rhodewalt 2001). Similar
Free University of Amsterdam, Amsterdam, The Netherlands
to adults, youth with narcissistic traits often display rather
L. Aleva aversive interpersonal behavior, such as arrogance, lack of
Department of Developmental Psychological, empathy, exploitativeness and aggression (Morf and
Utrecht University, Utrecht, The Netherlands
Rhodewalt 2001; Thomaes and Brummelman 2015).
M. Van der Meulen According to the self-regulatory model of Morf and
Groningen University, Groningen, The Netherlands Rhodewalt (2001), narcissistic individuals use several

Salmivalli (2001) posited that the exploitativeness Screening Device (APSD). inclined to engage in direct forms of aggression than girls. relational aggression least applicable to children with the most extreme scores. Direct bullying pertains to behaviors in Second. 2010. it has been argued group validly (Von Eye and Bogat 2006).. bullying have used instruments that do not tap all facets of the sively in an attempt to restore their self-esteem. and indirect bullying. albeit hensive and psychometrically sound measure indexes narcissism weakly. . previous work has almost exclusively employed a which the victim is overtly harassed (e. bullying was higher and more stable among those scoring higher Fourth. Salmivalli 2001). Specifically. Hawley 1999). no study has exam- prospective study. between narcissistic features and bullying in youth. assesses narcissistic may lead to aggression being employed instrumentally to foster behavior that tends to co-occur with psychopathic traits. whether narcissistic youth differ in the extent to which they enact it should be noted that several studies have shown that differ. may be more effective than direct aggression in terms of acquir- Moreover. having access to desirable.g.. employed by Stellwagen and Kerig and lack of empathy that figure prominently in narcissists (2013) as well as Fanti and Kimonis (2012).. but not their grandiose self-views. in the case of distinct linked to both gender and narcissism. Social dominance is indexed by resource search gaps remain. This widely used. we construe social dom- bully. tive peer relations is an essential developmental task (Hartup Pepler et al.g. In a similar vein. a few studies have examined the link that are at the core of the narcissism construct. (2010) only ose self-perceptions and sense of entitlement characteristic of used the BExploitativeness^ subscale of the Narcissistic youth with psychopathic traits) was concurrently positively linked Personality Questionnaire for Children-Revised (NPQC-R). with bullying. For this reason. An important reason to group in terms of how the predictors operate on the outcomes make this distinction is that both forms may be differentially (Laursen and Hoff 2006).e. the link between narcissism and bullying.g. physically. there is a paucity of longitudinal research examining material resources (Hawley 1999). when the potential heterogeneity of narcissism ing and preserving a dominant position in the peer group and bullying in this respect is taken into account. Fanti and Henrich (2015) found that narcissistic ined whether they are successful in this regard. At his point. leagues (Ang et al. the Antisocial Process bullying. high and me. person-centered analyses were employed (Nagin 2005). longitudinal work among as a general construct. struct as Bpsychopathy-linked^ narcissism (for differences be- in an inpatient sample of youth aged 10–15. As in resource children with low general self-esteem are in particular likely to control theory (RCT. volved^ children. when faced with criticism.g. as both genders are about equally likely to display indirect forms ipants). Similarly. 2008). Importantly. verbally). to the extent that narcissistic children high in bul- on narcissism at baseline (Fanti and Kimonis 2012). and narcissism was and form of the association between the constructs examined. and they are often that for narcissistic youth the use of indirect. Moreover. 2008) was employed. but scarce social and colleagues. both forms of bullying. feelings of entitlement) During the past decade. unfold over time. In so doing. A significant drawback of this while indirect bullying refers to behaviors that do not directly approach is that participants are treated as one homogeneous confront the victim (e. which is an aspect of rela- Notwithstanding the merits of these studies. the Childhood Narcissism Scale (CNS. except for the study of Fanti and control. Ang and col. i. First.. of aggression (Card et al. treated as a continuous variable. For instance. researchers using the APSD typically refer to the measured con- er. tal period because during preadolescence the formation of posi- dium) that differ in their functioning and development (e. inance as competitive superiority. and concerns about social status figure prominently Henrich (2015) distinguished between bullies and Bunin.e. the stability of the core constructs over time is largely examining the relations between narcissism and bullying as they unknown. important re.g. In a recent lying pursue social dominance and power. they often react aggres. Consequently. 2008. Reijntjes et al. see Thomaes and (2013) found that psychopathy-linked narcissism (i. gossiping). variable-centered approach. With regard to narcissism construct. when a summary sta. 2010) showed that narcissistic exploitativeness Thomaes et al. and is well validated in Dutch samples of Greek-Cypriot adolescents aged 12–14 has demonstrated that youth. the dynamic. boys are more subgroups of bullies or narcissists (e. findings support Salmivalli’s (2001) hypothesis. We studied children in this developmen- ent trajectories of bullying behavior exist (e. Brummelman 2015). Although Fanti and 1996). the present study. In with scores for (ringleader) bullying. where- tistic such as a correlation does not equally apply to all partic. Moreover. Moreover. they did not distinguish between potentially To capture potential subgroup differences in the strength different bullying trajectory subgroups. Stellwagen and Kerig tween this construct and narcissism. the grandi.64 J Abnorm Child Psychol (2016) 44:63–74 techniques to maintain or enhance their inflated self-esteem.. Ang et al. longitudinal relationship be..g. we distinguished between direct tween narcissism and bullying remains to be investigated. tionship asymmetry. compre- in Asian youth was concurrently positively associated. interventions (Golmaryami and Barry 2010). Taken togeth. Third. also The present three-wave study that followed children from late because Fanti and coworkers only assessed narcissism once (at childhood into early adolescence addressed these limitations by baseline). mean-level parameters may not describe any sub. studies examining the link between narcissism and For instance. (Fossati et al. For instance. We therefore wanted to examine can be fine-tuned for specific groups of children. narcissistic cognitions and feelings (e.. 2013b)..

Procedure We expected to find at least two different developmental trajectories for both narcissism and the two types of bullying..3. and standardized interviews were following the highest narcissism trajectory would be more laptop administered. Other groups represented were pupils with at longitudinal course of two constructs of interest (Nagin and least one parent originating from Turkey. SD= graders. given that the narcissistic traits in youth in the gen. with complete data in their scores on relevant variables cissism and aggression may differ between boys and girls at T1 (p’s>0. and not themselves. Moreover. Participation rates within classroom dren confined to a stable classroom in which they mainly were very similar across years. Pauletti et al. or another European country. Children’s self-reported narcissism scores were obtained dur- including a high and a low trajectory. non-participating schools. of the study. and how these tra. Thomaes could withdraw from the study at any time. restricting wide range of social backgrounds. the sample included pupils from a grademates (which is the case in the Netherlands). we al. Morocco. Children not par- on their resource control scores. 2012). The peer nominations were collected during an inter- Similarly. At T2. segregation is salient. only a relatively small group of participants. Children could discontinue their participation at positively skewed (Thomaes and Brummelman 2015). so expected that relatively few children would be consistently research assistants were extensively trained. (e. Tremblay 2001). for chil- with the same peers.5). nience. They could either provide passive consent for their child’s participation by not communicating fur. We consider it likely that this will also be the case for bullying. although somewhat grounds. we expected that social dominance dren used a list containing the names of their class- would be highest for children displaying both high bullying mates. any time. jority (83 %) of the children was Caucasian (native jectories are related. Research on peer sociometric status has shown that. We also examined the out.g. We also expected that ing a classroom session run by trained research assistants. and high narcissism. We also obtained IRB ap- whether groups with distinct developmental trajectories of proval and permission from the schools. There was no school transition during this period. Although SES was not interact with their classmates and not much with other formally assessed. likely than their peers to simultaneously follow a higher bul.10). the sample contained 336 partic- nance. ric nominations (i. we hypothesized that children protocols were employed. we also examined potential gender differences. ipants (85. All children et al. written research high in narcissism. When providing peer nominations for bullying. data were available for 371 participants (94.e. The number of nominations was unlimited. after identifying joint trajectories of bul. the original sample). Mixed-sex nominations were Participants were 393 children (51 % girls) from 12 used and nominations were conducted within classrooms rath- elementary schools throughout the Netherlands. Finally. chil- lying trajectory. dynamic co-occurrence over time. nominations from other-sex classmates improved the predic- and almost all children remained in the same classroom tive validity of the sociometric measure. To minimize interviewer effects. as well as research showing that the link between nar. Attrition was mainly due to participants mov- lying and narcissism. Blike most^ and Blike least^) in sixth all participants were in fourth grade (Mage =10. The large ma- narcissism and bullying can be identified. relative to the limited nominations approach. predominantly Teachers rated children’s resource control at their own conve- boys. The er than within grade. Specifically. joint trajectory analyses were used to examine provided their own assent. Parents and children Narcissism The Childhood Narcissism Scale (CNS. we compared the joint trajectory groups ing to other. At the start of the study in 2006 (T1). thereby comprehensively investigating their Surinam. 2008) is a 10-item self-report measure that indexes trait . Parents received a the voting population to the classroom peers did not affect the letter in which they were informed about the purpose predictive validity of the measure.J Abnorm Child Psychol (2016) 44:63–74 65 Specifically. This person-centered approach relates the Dutch). Participants Children could only nominate children from their own classroom. Poulin and Dishion (2008) observed that including 0.5 %). the unlim- ited nominations procedure yields a more reliable and Method valid assessment (Terry 2000). At T3. or refuse by returning a preprinted objection form (4 %). in their study examining peer sociomet- mentary school. For the reasons outlined ticipating at T2 and/or T3 did not differ from children above. Measures ther with the researchers (96 %). view by a research assistant in a quiet room at the school eral population are normally distributed. Moreover.4 % of comes of bullying and narcissism in terms of social domi. would engage in consistently high levels of bullying.. Although in early adolescence gender children were followed during their last 3 years at ele. but no child did.

Subsequently. on a scale ranging from 0 (never or almost never) ly true). desirable material and follow. see Goossens analysis (Thomaes et al. Scores were then stan- estimated and no residual correlations between items were dardized within classes to take differences between allowed. damaging belongings of others..g. with self-esteem contingency. the scores were standardized within each class.11 (Muthén and Muthén 2010). proportion of individuals following each of the trajecto- ly strong peers. the total number of nominations by the number of nom- factor model was tested in MPlus using confirmatory factor inators minus 1 (the participant himself. forms of bullying. in which its three core features were explained: of individuals who follow distinct developmental trajec- intent to harm. Next. calling names) and indi. tional.90 at all three Thomaes and Brummelman 2015). NFI = 0.75 for both genders). ceived popularity. b) has shown vidual items) and for those who did not complete one or two that bullying as indexed by the BRNP is associated in complete assessments were handled through full information theoretically meaningful ways with peer-nominated per.^ Items are social resources (Hawley 1999). agentic but not communal so- cial goals. tories. physical. Several measures of model fit indicated that a nominating groups into account. attesting to assessment points). maximum likelihood (FIML). the person-centered analyses proceeded in three steps. high internal consistency (Cronbach’s alpha > 0. 2006. Using a large sample (N=1020). was also calculated by summing the scores for the two Specifically. The scale showed only weakly with self-esteem (see Thomaes et al. 2011. The of ways and differs from a quarrel or fight between two equal. but what it wants?^ The ratings were averaged. ting. children first received an elaborate description of growth curves for that variable. and indirect relational) and nominations were obtained rowly defined dimensions or facets such as adaptive versus by asking BDo you know which classmates carry out that maladaptive narcissism. 2008.95. In the case of three assessment points. without distinguishing between more nar. Research to 4 (very often). Analyses were performed for the two Standardized factor loadings ranged from 0. Moreover.. and self. 2009). Reijntjes et al. latent class growth analyses models (LCGA’s.g. Sample items are: BTo what extent is this shows that childhood narcissism has largely similar correlates child usually at the center of attention in a group of and outcomes as adult narcissism. and negatively with empathy (Thomaes et al.94. Children were told that bullying can occur in a number i. repetition over time. 2008). thereby yielding clusters the concept. psychopathic and Machiavellian personality traits. score. and BIf so. First. CFI = 0. could you give us 2008) has shown that CNS scores have both adaptive (agentic their names?^ Continuous scores for both direct and indirect interpersonal orientation) and maladaptive correlates bullying were computed within classrooms by dividing (exploitativeness). retest stability of the instrument are good (see Thomaes et al. 2011). scores on the CNS are positively associated account. Kärnä et al. verbal. Plan of Analyses aggression.e. All factor loadings were freely et al. 2013a. these tra- ence between perpetrator and victim (Salmivalli and Peets jectories are identified on the basis of two parameters. Muthén 2000) were employed. intercepts (starting values) and linear slopes. Previous work who did not complete an entire measure (as opposed to indi- (Olthof et al. . A total bullying score single-factor model provided a good fit to the data. spreading rumors. CNS scores children^?.47 to 0.^ and BI love showing all the things I can do. sum scores were used. In the present study. The CNS has a one.64 and bullying forms separately and for the total bullying all were significant. 2011) was used. used to estimate separate models for the developmental trajec- Olthof et al. peer-rated likeability. Missing data for participants stirring others up to exclude someone). Resource Control This construct indexes social dominance Sample items are BI am a great example for other kids to and involves having access to scarce. (>0.05.66 J Abnorm Child Psychol (2016) 44:63–74 narcissism in youth aged 8 and older. and BTo what extent does this child usually get are positively associated with self-appraised superiority. For instance. Previous research (Thomaes et al. RMSEA = 0. lying.. To take differences between teachers into construct validity. 2008). Using MPlus version 6. 2008). possession-related.. on six items. The internal consistency and the test. The BRNP is a comprehensive tories of narcissism. Latent class growth analysis uses an outcome variable perceived social competence. ries is estimated. and a patent power differ. direct rela- general construct. particular form of bullying?^. a single. Muthén and rect forms of bullying (e. and total bul- measure that indexes both direct forms of bullying (e. Cronbach’s alpha was adequate at all assessment points We first present descriptive data and correlational analyses. the group-based trajectory approach (Nagin 2005) was Bullying The bullying role nomination procedure (BRNP. hit. the two forms of bullying.e. measured at multiple time points to define a latent class To avoid potential interpretation differences of the term model in which the latent classes correspond to different bullying. five specific types of bullying were de- factor structure and was developed to measure narcissism as a scribed (i. Teachers rated participants rated on a scale ranging from 0 (not at all true) to 3 (complete. gossiping.

current software does not targeted variables separately. of narcissism and direct (indirect) bullying. straints between the parameters of interest (Klugkist et al. Across gender. boys consistently scored higher 2005).g. The results of the Bayesian approach are expressed in than girls. we used a though the differences were small (Cohen d’s<0. Joint probabilities pertain to were assigned to the high (medium. synonymous with an observation of high (low) direct bully- ing trajectory). simultaneously belonged to the high (medium.g. A model was consid- series of models was fitted. group for variable A (e. jectory group. allow for controlling for (the probability of) latent class jectories for (a) narcissism and (b) each of the two bullying membership. and total bullying a probability A is equal to probability B).70. with less discrepancy ship of indirect bullying when examining the effects of direct indicating better model fit.65. the two the probability that the specific model at hand receives most forms of bullying were substantially correlated at all three support from the data among a set of models (e. ….. and vice versa. the joint prob. founded decision regarding the optimal number of groups. To evaluate scored higher on narcissism than did girls (p’s<0.g.g. This signifies that. bullying (and vice versa). al- differences between observed probabilities. Specifically. total bullying) relate to social dominance (resource mated probability of belonging to a specific trajectory control) scores. Gender differ- terms of posterior model probabilities (PMP’s). examining chance that children assigned to a particular trajectory how narcissism and each of the two bullying forms separately group actually belong to that group is at least 0.. versus Model 2: Tables in the Electronic supplementary material). whereas controlling for a After determining the best fitting trajectory models for the continuous covariate is possible. However. including the Bayesian Information Criterion sism. we also examined whether waves. contribute to social dominance proved problematic. Given that both bullying forms were associated with bullying and narcissism. in the second step the joint tra. it should be noted that an additional joint outputs of a joint model are joint probabilities and two sets of trajectory analysis showed that almost all participants who conditional probabilities. and vice versa. Conditional probabilities pertain to the esti. Given our approach. representing ences were largest for direct bullying. 2005). However. and entropy.. Second. der differences were observed for resource control (p’s>0.. we compared these joint LRT). except for indirect bullying at T3. …. Key At this point.05).25). high direct bullying) given membership of a specific trajectory group for variable B (e. . as well as the total bullying score were estimated. No gen- Bayesian model selection approach with (in)equality con. These probabili- Results ties are denoted by π j|kand πk| j and are calculated as follows: π jjk ¼ ππk . For both forms of bullying.. we examined whether the odds of correct clas. one should control we compared the model estimated proportion of the for the effect of indirect (direct) bullying. children who follow both the high that an observation of high (low) indirect bullying is almost narcissism trajectory as well as the high physical bully. K and πk j j ¼ ππ j . Finally. conditional probabilities do not imply a time Table 1 displays the means and standard deviations for all order relationship but reflect the probability of simultaneously variables at the three assessment points. boys following two trajectories during the same period.20). see correlation probability A is larger than probability B. p’s < 0. narcissism and total bullying score. To make a well. population following a particular trajectory group (πj) which yields distinct trajectory groups (latent classes) for both with the corresponding proportion of the sample types of bullying. 0. the are related to social dominance scores.e.J Abnorm Child Psychol (2016) 44:63–74 67 For narcissism. Specifically. k ¼ 1. jk jk Preliminary Analyses with π j ¼ ∑ j π jk . When j and k index the trajectory groups ing. j ¼ 1. we decided to examine how the joint abilities are denoted by π jk and are provided as part of trajectories of narcissism and both bullying behaviors com- the output. this requires controlling for class member- assigned to that trajectory (Pj). when examining social dominance as a function sification (OCCj) were at least 5 for all groups. Across waves. the Lo-Mendell-Rubin likelihood ratio test (LMR. Finally.001.70.95 (Klugkist et al. starting with a one-trajectory mod. low) direct bullying tra- the proportion of children estimated to belong simulta. Model 1: assessment points (r’s > 0. groups on their mean resource control scores across the three Following Nagin (2005). We also aimed to investigate how the joint trajectories for all groups the average posterior probability (AvePPj) of narcissism and the two different bullying forms separately exceeded 0. J : Importantly. on average. This finding indicates and variable B (e. we examined how social dominance scores dif- several statistical indicators were used (as recommended by fered as a function of the joint effects of bullying and narcis- Nagin 2005). neously to certain trajectory groups of both variable A low) indirect bullying trajectory group. forms. the two bullying forms. ered to outperform another model when its PMP was at least el and moving to a five-trajectory model. bined (i. after identifying the joint trajectories of (BIC). strongly interwoven. high narcissism). w i t h πk ¼ ∑ j π jk .

the LMR-statistic was significant. p > 0.162 0. bullying. entropy = 0. respectively). p>0.68 J Abnorm Child Psychol (2016) 44:63–74 Table 1 Means and standard deviations of narcissism.44 ences between Pj and πj less than 2 %.62 4.54 0.8 % and 55.49. 23.4 %) engaged in stable.b 0.20).3. the highest scores.41 displayed stable medium narcissism scores (intercept (I)= Wave3 0. the fit indices for the three- Wave2 8.7 to 2290. and under- Wave2 0.11 2 0.83.33 b 0. and the BIC Wave 3 2.39 1.a 0.47 8. S=0. indirect bullying. OCC j’s>5. Specifically.99 0. entropy decreased from 0.162 4.08 0.78). Direct bullying As depicted in Figure 1 (see Electronic supplementary Wave1 0.77 value decreased from 2318. children in the largest group (n= source control.882 . Different subscripts in the same column (letters) de.52 trajectory groups (56. The scores of children in the third largest group resource control was low (r’s<0.101 0.07 0. S=−0.09.401 4.07 0.80 1. 37.63 group model were good (AvePP j’s>0.13). p>0.74.10).0 to 2347.05 1. p<0.07 0.3. average T2.48 1.96 1. when mov- Narcissism ing to a four-group model. 29.21.01 and the BIC value in- they showed the highest posterior probability.02. differ- Wave3 8.08.61 .001. p<0.99 1.10 scores (I=5.44 material).08.00 1. As depicted in Figure 1.10 0. p<0. S=−0.28 0. 0.53 0.01 the data best.71.61 7.21 0.052 0. Children in the Wave2 .25 0. levels of direct bullying (I = −0.001. entropy decreased to 0.21 0.48 Resource control Direct Bullying A four-group model was selected as fitting Wave 1 1.statistic was not significant.20. At all three time points.001. Only a −0. and this change only applied to boys.82 0. However.18).35 7.60 . time (I=0. p<0.73. when moving to a four- direct bullying. and they did not change over T1 and T3 (see Electronic supplementary material). When moving from a three-group to a four- Wave 2 2.27.16 group model.01.b 0.50 0. and Narcissism The statistical indicators provided most support resource control for a three-group model.81 creased to 6200. p>0. group model.74. p<0. Importantly. the linkage between narcissism and p>0. creased to 2391.27 0. entropy increased from 0.36.37 a 0. Indirect Bullying A three-group model fitted the data best.41 b 0.07 8. and the BIC value note significant differences over time (p’s<0. when moving from a Boys Girls Total two-group to a three-group model. the LMR-statistic was significant.75 2. the LMR-statistic was significant. stability for all constructs over time. However. ed (r’s<0.12 2 0.55 1.61.76 to 0.4 %) displayed the highest Wave3 0.6 %).7 and 24.9 %): χ2(2)=8.2 and 69.12 . Boys were overrepresented in the high and medium Wave1 . the Participants were assigned to the trajectory group for which LMR-statistic was not significant. For narcissism and re.40 1.471.83 8. Different subscripts (numbers) in the same row denote significant ing to a five-group model. 9.25 1. and the BIC value decreased Separate trajectory analyses were performed for narcissism. which were stable over time (I=1.00 1. The smallest at all three time points scores for narcissism and each of the group (n=37.3.5 %) showed consistently low Wave1 .5 %) were medium.68 to 0.11 third and smallest group (n=37.05) increased to 2307.11 0.571. entropy = 0.37 4.07 2. S=−0. 43.911 4.192 0. the LMR-statistic was not significant. entropy decreased from 0.16 0.30).29 a 0. S = 0. p<0.61 7.1 %): χ2(3)=57.70). Total Bullying p>0. Children in the sec- Indirect bullying ond largest group (n=171.4 %.60. The fit indices for the four-group model were good (AvePP j’s>0. p<0.25 b 0. the mean score did not change over time.10 scores that were stable over time (I=15. although significant at (n=92. Wave3 . when mov- Note.172 0. as well as total bul.8 % of the sample) Wave2 0.20). underrepresented in the average and low direct bullying tra- lying (r’s ranging from 0. Boys were overrepresented in both trast.03. Similarly. albeit significantly in several instances. jectory groups (41. and the BIC value in- Wave1 9.70.12 . and resource control and both bullying types.68 1.01). p>0.39 to 0.20). the LMR.07 0.462 4.11 2 0. differences between Repeated measures ANOVA analyses revealed substantial Pj and πj less than 2 %. OCC j’s>5.01. the largest group (n=184.78 to gender differences.7. S=0.91 4.47 1. 46. Children in the second decrease in direct and total bullying was observed from T1 to largest group (n=115.052 0.8 to 6189.31 b 0.49 represented in the lowest group (40. and the total bullying score.091 0.b 0.111 4.56. p’s<0. LCGA Analyses When moving from a two-group to a three-group model. p>0. from 2364. In con. and the BIC MeanD SD Mean SD Mean SD value decreased from 6230. 148.05 0.a 0.5 %) consisted of those children displaying two bullying forms and total bullying were only weakly relat. However. The fit indices for the three-group model . p<0.20).b 0. slope (S)=0.3.09 0. across time substantial linkages were found between the high and medium trajectory groups (89.6 % of the sample) displayed stable low scores (I= Scores for bullying were also quite stable over time.20). 9.

differences between whereas for boys following the low narcissism trajectory these Pj and πj less than 2 %. Given these observations. S=0.0 %). χ2(3)=49. were good (AvePP j’s>0. likely to be high on narcissism. −0. However.001. That is. 17.01. and the probability to follow the high p>0. p < 0.05.4 to 2307. scored high on both constructs simultaneously.02. which did not change over time girls separately.03.10). p>0.75. Participants in outlined above (see Plan of Analyses).. p<0.3 %) displayed the lowest scores. Conditional Probabilities of Direct Bullying Given Total Bullying A four-group model was selected. p<0.95). probabilities across (I=1. Boys on the high narcissism trajectory were much el. A relatively small subgroup of the boys (n=8.10). boys who are not high in narcissism.1 %) engaged in the highest membership for narcissism and direct bullying.13. boys assigned to the other three bullying groups. OCC j’s>5.7 and 62. Specifically.4 and 34.95). Children in the largest probabilities were equally high (0. Results showed that these high levels of direct bullying are also quite common among joint probabilities differed substantially across both genders.0 %) consisted of participants groups. and children in this group um on both narcissism and direct bullying (n=46.95).001. the two sets of conditional probabilities were more these analyses separately by gender. contrary to trajectory groups (86.18 and 0.08. p<0. panel of Table 2 shows the joint probabilities of trajectory ond largest group (n=67.7 %): χ2(2)=13. computed and compared using the Bayesian approach tween Pj and πj less than 3 %. p<0. This finding indicates sism and high indirect bullying). Conditional Probabilities of Narcissism Given Direct Primary Analyses Bullying The bottom panel of Table 2 presents the likelihood of following one of the three narcissism trajectories condition- The joint analyses were initially performed for boys and girls al on membership of a specific direct bullying trajectory combined.75. group (n=146. which slightly increased over time (I = −0.8 %) scored average on the highest bullying trajectory (PMP’s > 0.g. 37.2 %) displayed stable low scores (I=−0. while most girls scored low on both constructs (n = 62.77.20. showed stable medium scores (I=0. OCC j’s>5. The largest subgroup of boys scored medi- almost as large (n=64. p>0. The third group was all cells sum to 1. the BIC value increased to 2315.88. results for girls indicate no systematic developmental the average and low trajectory groups (40. 24 %). S=−0.001.26). when going to a five-group mod. 4 %). The (PMP’s > 0.03.95). S=0. for all four bullying groups the prob- girls distinct joint and conditional probabilities were ability to belong to the high narcissism group was lowest (and .07.10). Taken together. 66.23. Narcissism The middle panel of Table 2 presents the likeli- ing from a three-group to a four-group model. 24. In fact.25. PMP > Specifically. for boys and levels of indirect bullying.76). differences be. 11. findings were markedly different.001. Joint Trajectories of Narcissism and Direct Bullying The top p<0.5 %): overlap between narcissism and direct bullying. respectively).19.001.74). 27. see Figure 1).26 to 0. Boys were overrepresented in the high and medium bullying trajectory was lowest.20). p<0.20. for girls the probability to follow the low bullying displaying stable high scores (I = 1.6). hood of following one of the four direct bullying trajectories statistic was significant.J Abnorm Child Psychol (2016) 44:63–74 69 were adequate (AvePP j’s>0. highly narcissistic boys were substantially more p<0. such that for boys and symmetrical.39). respectively. across the three narcissism trajectory smallest group (n= 43.0 and 67. Children in the third largest girls in all three narcissism trajectory groups were far more group (n=94. Moreover. and the BIC value decreased from conditional on membership of a specific narcissism trajectory 2320. and their scores did not change over time (I= For girls.81. S = 0. and the LMR-statistic more likely to follow the high than the low bullying trajectory was not significant. Boys on the highest bullying trajectory were about entire sample the joint probabilities πjk of belonging to two equally likely to follow the three narcissism trajectories (prob- different trajectory groups simultaneously (e.43. When mov.08. S=0. lying.0 %) showed stable scores that were medium likely to follow the low than the high bullying trajectory in magnitude (I=0. PMP’s > 0. combining both genders trajectories were much less likely to be high on narcissism when presenting the joint and conditional probabilities would (probabilities ranging from 0 to 0. trajectory was highest. and underrepresented in boys. in all instances we first determined for the group. p<0. but underrepresented in but no girl. entropy=0.5 %). high narcis. p>0. The fit indices for the four-group model (probabilities 0.15). entropy=0.001. the low trajectory group (42. Those assigned to likely than their peers medium or low in narcissism to follow the second largest group (n=109. S=0. Hence. Boys were overrepresented in the high and medium 31 %). Children in the sec. S=−0. total bullying. Whereas a substantial and that girls were underrepresented in both the high narcis. p>0. the largest group (n= 260.001. minority of the high bullying boys was also high in narcissism sism trajectory group and the two high bullying trajectory (probability 0. see Figure 1).38 and 0. trajectory groups (62.72. the LMR. have concealed important differences between boys and girls boys who do not score high on direct bullying are quite un- in terms of the dynamic overlap between narcissism and bul. 16. group. In this part of the Table. p>0.01.5. We therefore considered it more appropriate to perform For girls. p>0. abilities ranging from 0.

63 (n=5) 0.37(n=58) 0.75(n=50)/ 0. Ng =73) 0.18(n=12)/0.23(n=24)/0.51(n=55)/ 0.19(n=37)/ 0. Ng =63) – / 0.35 (n=11) / 0.06(n=12)/ – 0.34 (n=37) 0.04(n=8) 0. (Nb =71. Girls high or Joint Trajectories of Narcissism and Indirect Bullying Most medium on bullying were most often assigned to the medium boys were low in indirect bullying. Ng =108) 0.18(n=13) 0.52 (n=24) / 0.50 (n=55)/ 0.06(n=4) / 0.31(n=62) Direct bullying High Medium Average Low Probabilities of direct bullying conditioned on narcissism b High narcissism (Nb =21.38 (n=15)/ 0.26(n=50)/ 0.53(n=9) Medium narcissism (Nb =105.18(n=35) 0.38(n=8)/ – 0.54 (n=84) a Cells sum to 1. and simultaneously medi- narcissism group.03(n=1) 0. Ng =8) 0.09(n=17)/0.13 (n=8) 0.53 (n=24) / 0.07 (n=1) 0.04(n=3) 0.51(n=37) Low narcissism (Nb =67.02(n=4) / 0.59 (n=13) 0.60 (n=12)/ – 0.45 (n=50)/ 0. Ng =108) 0.02(n=3) 0.02(n=3) 0.55 (n=35) Low bullying (Nb =45.30(n=6) / 0.38 (n=17) / 0.39 (n=12) / 0.33(n=22)/0.48 (n=22) / 0.37 (n=3) Medium indirect bullying (Nb =43. Ng =17) 0.47(n=8) 0.11(n=22)/0. Ng =15) 0.05 (n=10) 0.04(n=8)/ – 0.43(n=9)/ – – / 0.19(n=12)/ 0.02(n=13) 0.28(n=55)/ 0.09(n=14) 0.03(n=5) 0.22 (n=16) / 0.12(n=24)/0.29 (n=10) 0.01(n=2) 0.08(n=15)/ 0. Ng =5) 0. (Nb =46.10(n=2) / 0.79 (n=58) Low narcissism (Nb =64.60 (n=3) 0.05(n=2)/ 0. while girls in the average and low bullying um or low in narcissism (28 and 26 %.41(n=9) Average bull. Ng =108) 0. largest subgroup of girls scored low on both constructs Table 3 Joint and conditional probabilities of narcissism and indirect bullying trajectories Narcissism Indirect bullying High Medium Low Joint probability of trajectory group membership a High 0. Ng =34) 0.005 (n=1) 0.32(n=35) 0. b Rows sum to 1.53 (n=62) a Cells sum to 1.27(n=20) 0.19(n=4)/0.12 (n=23) 0.65 (n=46) / 0.40(n=2) Medium bull.23(n=24)/0.32 (n=20) 0.12(n=24)/0.24(n=16)/0.70 J Abnorm Child Psychol (2016) 44:63–74 Table 2 Joint and conditional probabilities of narcissism and direct bullying trajectories Narcissism Direct bullying High Medium Average Low Joint probability of trajectory group membership a High 0. see Table 3).42 (n=84) Indirect bullying High Medium Low Probabilities of indirect bullying conditioned on narcissism b High narcissism (Nb =20.10(n=20) 0.07(n=5) 0.57(n=62) Narcissism High Medium Low Probabilities of narcissism conditioned on direct bullying b High bullying (Nb =31.10(n=11)/0.14(n=15) / 0.03(n=3) 0.05(n=9) Medium 0.02(n=4)/0.93 (n=14) Medium narcissism (Nb =107.05 (n=6)/ 0.24(n=46)/0.44(n=46)/0.25(n=17)/0.09 (n=9) 0.03(n=6) / 0.14 (n=10) 0.06(n=12)/ 0. Ng =22) 0. b Rows sum to 1.31 (n=12)/ 0.05(n=9) 0.31 (n=12)/ – 0.09 (n=4) / 0.02(n=2) 0.29 (n=58) Low 0. The group were most likely to be simultaneously low in narcissism. Figures in bold pertain to girls even zero for the two highest bullying groups).05(n=9)/ – – / 0. Figures in bold pertain to girls .08(n=9) 0. Ng =156) 0.07 (n=14) Medium 0.09(n=4) / 0.68 (n=23) Low indirect bullying (Nb =111.06(n=12)/0.01(n=2) / 0.26 (n=8) / – 0.06(n=11)/0. Ng =73) 0.86 (n=37)/ 0.21 (n=23) 0.19(n=37) Low 0.08(n=16)/0.13 (n=9) / – 0.35(n=37)/ 0.76 (n=84) Narcissism High Medium Low Probabilities of narcissism conditioned on indirect bullying b High indirect bullying (Nb =39.

In the bottom panel of the Table. PMP > 0.81) Medium narcissism (Nb =101. Results for girls were markedly different. girls rarely followed the high bullying source control over the three assessment points for each of the trajectory (all probabilities < 0.11(n=21)/0.57(1. across gender and level of narcissism.32) 0.10).01(n=2) 0. and these children scored significantly higher than probability 0.56 (0. whereas the girls in the two other bullying groups were most Conditional Probabilities of Indirect Bullying Given often low in narcissism. sism to follow the high indirect bullying trajectory (0.60 (0.40 (0.26(n=51) Total bullying High Medium Average Low Mean score on resource control over time b High narcissism (Nb =21.62) Low narcissism (Nb =71. group.60) was whereas the largest subgroup of girls scored low on both var- substantially higher than was the probability of highly narcis.09(n=17)/0. bullying trajectory group were significantly higher than were cating that high narcissism is one of multiple factors contrib.62)/1. indi.14 and 0.05)/−0.06(n=12)/0.24 (N/A) −0.64) /−0.75 (0.60) than their peers medium or low Single and Combined Effects of Narcissism and Total Bullying in narcissism to follow the high bullying trajectory (probabil.05(n=9) Medium 0.08(n=15) 0.95). those in the low bullying trajectory group (p<0. PMP’s > 0.005(n=1) 0. Conditional Probabilities of Narcissism Given Indirect p<0. b Figures in bold pertain to girls.55) −0.42 (0.50)/ – 0.02(n=4)/0.08(n=16) 0.31.001).47 (0. In turn.02(n=4) 0.53) a Cells sum to 1. but no girl.73) /−0.66)/ 0.40 (0. were most likely to belong to the medium narcissism group. Narcissism trajectory group. among variable.18(n=35) Low 0.51) −0. A small subgroup of the sistic boys to follow the high direct bullying trajectory (prob.J Abnorm Child Psychol (2016) 44:63–74 71 (42 %). dium on both narcissism and total bullying (n=37.47 (0.61)/ 0.67) −0. Taken Girls in all indirect bullying trajectory groups were unlikely together. iables simultaneously (n=51.38 (0. Post-hoc multiple group comparisons using the highest indirect bullying trajectory were about equally Tukey’s d showed that the scores of children in the highest likely to follow each of the three narcissism trajectories. and gender were the between-subjects factors. figures between parentheses are SD’s .93 (0.001). 365) = 26. scores for children in the medium bullying trajectory group uting to high indirect bullying. Compared to the high bullying (p<0. whereas no ability 0. 3 %) scored high on both constructs.75) −0.09(n=18)/0. 19 %).40. Twelve boys. PMP’s > 0. 26 %). No other significant main or interaction effects Bullying The bottom panel of Table 3 shows that boys on emerged.75).66 (0. Boys medium Table 4 displays the number of children assigned to each of the or low in narcissism were substantially more likely to follow joint trajectories of narcissism and total bullying. joint trajectory groups are shown. Ng =101) 0. Ng =74) 0. Instead.48 (0.95. these latter scores significantly exceeded boys. more intense to simultaneously belong to the high narcissism group (all bullying is associated with higher resource control.05 for both groups. Score on Social Dominance Outcome Scores The top panel of ities 0. it was far less common for boys medium or low on those for children in the average bullying trajectory group bullying to be high in narcissism (probability 0. versus (p<0. the probability of boys high on narcis.03(n=7)/ – 0.25 (0. Across narcis.95).001).95). boys (n=7.13(n=25)/0. total bullying trajectory girls narcissism and indirect bullying were not related.01(n=2)/ 0.73) / 0.86 (0.31 (0. see Table 3).24(1.001.04(n=8)/ 0. probabilities < 0.68)/ 1. girl did so. Narcissism Boys on the high narcissism trajectory were much more likely (probability 0. in all three narcis. Results revealed a main effect for bullying group: F (2. Table 4 Joint probabilities of narcissism and total bullying trajectories (upper panel) and mean resource control scores for the joint trajectory groups (bottom panel) Narcissism Total bullying High Medium Average Low Joint probability of trajectory group membership a High 0.03(n=5) 0. The eight girls high on indirect bullying sism and indirect bullying simultaneously.02) −0.12 (0.13 (0.14(n=28) 0.16(n=32) 0.19.01) −0. Hence.72) / 0.24 (0. scored high on both narcis.10).09(n=17)/0. An ANOVA was performed sism groups girls were most likely to follow the low indirect with mean resource control score serving as the dependent bullying trajectory (all probabilities > 0.19(n=37)/0. the mean scores for re- sism trajectory groups.12(1.69) /−0.77)/ 0. The largest subgroup of boys scored me- > 0. for both the lowest than the highest indirect bullying trajectory (PMP’s genders separately. Ng =15) 0. Noteworthy.13(n=25)/0.37 (0.38.

. Moreover. contrary to previous work in which et al. For instance. showed high bullying also received high resource control sism and bullying trajectory groups. who noted Bthere are two different path- positively skewed) and the mean score for this high narcissism ways to achieving status (visibility and influence) within the trajectory group being more than 1 SD above the mean. boys scored above average in terms of resource control. Relative to girls. and girls were to a victim being rejected or ostracized by a group. served among youth medium or low in bullying. and direct quite unlikely to be assigned to the highest bullying trajectory possession-related bullying may be most effective when per- group. However. we investigated to what extent groups and can therefore not capture potential differences be- children high in narcissism are successful in obtaining social tween these groups in terms of associations between variables. . An important conclusion of the present bullying type is that narcissistic boys perceive indirect bullying work. In a similar vein. high scores for resource control were also ob- number of children was assigned to the high narcissism tra. boys following the high narcissism either direct or indirect bullying. The present multi-informant study examined longitudinal link. 398). Interestingly. For boys. the other through the possession of peer-valued character- vein. is that with respect to nar. analyses or SEM do not distinguish between different trajectory ent forms of bullying. In contrast. Only a relatively small Noteworthy.g. In fact. inance. 2007). nei. For most narcis. in both instances the To the extent that narcissism involves the ongoing need to reverse pattern was not observed. require cooperating with peers to most effectively harass a For girls. boys high in narcissism were substantially more likely While highly narcissistic boys were likely to be high on to display high levels of both direct and indirect bullying than both direct and indirect bullying. highly narcissistic girls were trajectories revealed a clear link between high narcissism and not more likely than their peers to engage in high levels of high direct bullying. albeit less jectory group. Interestingly. given the almost frequently. One possible explanation for this difference as a function of tributes to this outcome. More important for the present pur. one through the explicit use of aggressive behav- were moderately overrepresented in this group. boys with high ings reported by Salmivalli and colleagues (Salmivalli et al. and how elevated bullying con. hiding or damaging a bicycle togeth- high levels of bullying. direct relational bullying often pertains sociation with level of narcissism emerged. as being more effective in obtaining their goals of power. low narcissistic boys high on tween narcissism and more intense bullying. Fanti and Kimonis (2012) found a linkage be..highly narcissistic cuss our findings in more detail. For instance. our analyses strongly suggest that bullying. diose self. for girls no trajectory were more than twice as likely as their peers medium systematic overlap between high narcissism and high resource or low in narcissism to be assigned to the highest direct bullying control was observed. and to feel powerful. This came as no surprise. consistent with expectations. Specifically. Importantly. trajectory with the highest bullying scores were not on the Although for boys being consistently high in narcissism is high narcissism trajectory. and prestige. victim.g. poses. rather ined multiple-group trajectory models fitted the data best. resource control. the majority of the boys assigned to the cissism was stronger for indirect than for direct bullying. Narcissistic boys were also obtain admiration. Miller with bullying. Boys peer group. engage in the kinds of (indirect) bullying they can most easily marked differences between boys and girls were found with perform solitarily. it appears that highly quite successful in their pursuit of social dominance. but only narcissistic boys are quite successful in nourishing their gran- when they engaged in high levels of bullying. throwing around a schoolbag between multiple classmates). among boys high narcis- strongly associated with being consistently high in both types sism is one of more factors that predict intense bullying. er. Hence. the few girls that bullies or subgroups of narcissistic children. In fact. This link with nar. variable-centered approaches such as regression ages between preadolescents’ narcissistic traits and two differ. given the difficulties nar- ther in terms of mean scores. This finding is in line with the work of Vaillancourt normal distribution of narcissism scores (although somewhat and Hymel (2006). boys were more likely to engage in formed collectively (e. Alternatively. or changed only slightly over time. nor with respect to the linkages cissists encounter when cooperating with others (e.72 J Abnorm Child Psychol (2016) 44:63–74 Discussion levels of narcissism were even substantially more likely to be assigned to the highest indirect bullying trajectory group. it may also be that narcissistic boys preferentially no gender differences emerged (Fanti and Kimonis 2012). Examples of these characteristics include be- ed the highest trajectory. Moreover. these relationships were not were their male peers lower in narcissism. most – but not all . In a similar ior. dom- cissism children do not constitute one homogenous group. although they were not high in narcissism. than narcissism as such. relative to direct bullying. for both forms of bullying relatively few children follow. across both forms of bullying no systematic as. dominance in the peer group. Below we dis. the joint analysis of the narcissism and bullying Contrary to narcissistic boys. These observations converge with find- group. istics^ (p. symmetrical. for all variables exam. is the critical factor yielding high Noteworthy. mean scores were stable scores. but they did not bullying were more successful in this regard than their high distinguish between potentially different trajectory groups of narcissistic peers low in bullying. ing athletic or physically attractive. Using a person-centered approach. of bullying (as well as total bullying). with the large majority being male. several forms of direct bullying respect to the link between narcissism and bullying.

2008a). recent research (Kärnä et al. in youth. bullying and social dominance these traits (Thomaes and Brummelman 2015). Third. and vides important new knowledge regarding the developmental not much is known about the factors that cause and maintain linkages between narcissism. indi- their grandiose sense of self and privileged status over others viduals cannot be assigned to one of the distinct (latent) classes with their only average social dominance scores? One possi. center of attention^. Research has Conflict of Interest The authors declare that they have no conflict of interest. chopathy tend to co-occur (Thomaes et al. by pears that the bullying of high narcissistic boys is instrumental increasing athletic competence). confronting narcissistic Commons Attribution License which permits any use. Hence. likely to refrain from this successful behavior. latent classes do not try to satisfy their self-motives of grandiosity and power in necessarily correspond to truly existing different groups in the other. and narcissism and psy. For instance. but are somehow not successful.. 2011). longitudinal designs do not permit causal inferences. Kuefner et al. Although What may account for this marked gender difference ob. re- duces aggression among children high in psychopathy (Van Open Access This article is distributed under the terms of the Creative Baardewijk et al.g. Fourth. are (far) less likely to bully. our person- likely than narcissistic boys to bully. Bcan’t take criticism^) was positively as. recent work has shown that the Finnish anti. whereas girls’ bully. we did not examine children’s motivation gerate their qualities in terms of being exceptionally (but in. Second. distribution. 2012). 2008b). population. and bullies more directly and explicitly with the misery they bring reproduction in any medium. it appears that among boys high narcissism leads to high bul- served in two independent studies? Salmivalli et al. . speak to the specificity of the results for narcissism. The peer group is pivotal in this narcissistic boys show an elevated inclination to engage in regard. Limitations and Directions for Future Research ent dimensions of self-esteem and different participant roles in bullying situations (e. identified for girls. associated with indirect versus direct bullying. To the extent that bullying provides narcissistic boys the is needed to investigate why children prefer certain types of position of dominance and power they aspire. how can they reconcile centered approach also has disadvantages. and the role of narcissism in this regard. Third. but the distribution of true scores may be tige. or general Bclique^ members. our findings do not by dispositional traits (e. they may exag. narcissism). shown that making victims’ distress cues more salient. it may also be that the rewards of bullying maintain or speculated that the bullying of boys is more strongly driven further elevate their narcissism. the presence of discrete bility is that narcissistic girls do long for admiration and pres. with perfect precision. the present study pro- treatment of narcissistic traits in youth is still in its infancy. there are different trajectory groups of children vention is more likely to be effective when it seeks to reduce who differ in level of narcissism. Hence.g. and it remains to be strumentally) kind and trustworthy (see Gebauer et al. narcissism and high bullying is stronger for indirect versus bullying program BKiVa^ effectively weakens the link be. provided the original author(s) and the about might decrease their unwanted behavior. Although the behavior of narcissistic Useful extensions may be to teach narcissistic bullies that youth is thus not necessarily interpersonally effective. source are credited. future research should examine egotism^ (assessed with items such as Balways wants to be the youth from a broader age range and other ethnic groups. the link between high Interestingly. Finally. 2013) has an unsuccessful strategy by focusing in particular on changing shown that many narcissists are at increased risk for peer re- the behavior of uninvolved bystanders. Hence. However. groups is assumed. defender) among Our findings are based on primarily Caucasian pre-adolescents. Alternatively. when their actions are reinforced by peers. which may viduals are typically low in empathy. The KiVa program aims to render bullying among both adults and youth (e. Currently. Finally.. Van Baardewijk et al. versus ing is more contingent on situational and psychological fac. whereas no such linkage was prospectively examine the overlap of two constructs of interest. Future research tion. for instance the two other Bdark triad traits^ of psychopathy tors such as being stimulated to bully by close friends or and Machiavellianism (Paulhus and Williams 2002). who examined cross-sectional linkages between differ. Second. although a joint trajectory approach is well suited to sociated with bullying for boys. whereas highly the rewards of bullying. jection and isolation. the Notwithstanding these limitations. Moreover. Given that narcissistic indi. ringleader bully. For instance. serve to maintain or enhance their grandiose self. adolescents aged 14–15.. they are not bullying over others. Fourth. 2008.. First. to engage in different types of bullying. they may continuous instead of discrete. direct bullying. investigated why elevated narcissism among boys was stronger The present research has possible implications for interven.J Abnorm Child Psychol (2016) 44:63–74 73 1999).g. (1999) lying.g. their bullying may be sustained by low sensitivity to signs of distress in others. even if narcissistic girls are less self-esteem (Fanti and Henrich 2015). boys low and medium on narcissism prestige. for girls higher narcissism is not asso- tween bullying and a dominant position in the peer group ciated with more frequent bullying. given that bullies can only achieve dominance and high levels of bullying. in establishing social dominance in the peer group. more communal domains. it ap- there are also other ways to achieve social dominance (e. inter. These authors found that Bdefensive To examine generalizability.

. being bad to others? Gebauer. Group-based modeling of development over the life 718–725.. Sawalani.. (2015). Relations of bullying in the context of the peer group: a three wave longitudinal proactive and reactive dimensions of aggression to overt and covert analysis. of Crime and Justice. Child Development. & Connolly. Ong. F. 83. E. 6. E. Journal of Personality Assessment. Laursen (Eds. Journal of jectories of bullying and social dominance in youth. Journal of approval-of aggression beliefs. relationships in middle childhood and early adolescence. F. & Bogat.. (2013b). Verplanken..). Meenaksi. Nestler. esteem on bullying and victimization during early adolescence. The ontogenesis of social dominance: a strategy- Sijtsema. defensive egotism as predictors of adolescents’ participation in bul- Hartup. M. Borroni. Bushman. H.. bles. Recent advances in measurement theory and the use of being an (un)popular narcissist.).). Van Baardewijk. developmental significance. C. L. Stellwagen. Olthof. & Vermeiren. T. (2009). Olthof. A. C.. 184–195. Salmivalli. Miller. 322–340).. scales: a comparison between various criteria for assigning roles and (1999). T. A. In D. Communal narcissism. Unraveling the paradoxes of nar. Child Development. F.. W. 17. T. Goossens. 375–393. Journal of Personality and Social Aggression and Violent Behavior. Empirical test of bullies’ status goals: assessing direct goals. A. Feeling good about oneself. (2006). Self-evaluated self-esteem. O. Kaistaniemi. 339–359. (2013). Narcissism and adjustment in preadolescence. In K. Handbook of peer and self-esteem among adolescents in a residential setting. L. Hawley. & Lagerspetz. reported and peer-reported relational aggression with narcissism Rubin. Neglect. Van Baardewijk. & Vermeiren. A. (Eds. Y. Person-oriented and variable-oriented desired and acquired dominance in the peer group. Jiang. J. 908–921. S.. K. and groups (pp. Ang. & Salmivalli. A. 50.. Narcissism. (2000). Aggression and social status: the variable-centered analyses: growth mixture modeling with latent moderating roles of sex.. 377–389. Journal of School research: concepts.. A. & Dekker.. The company they keep: friendships and their lying situations.. Child Development. 882–891. D.). F. 1185–1229. 81. 15. Bullying and victimization: the R. Salmivalli. (2006). R. D... D. Journal of Early Adolescence. (2012).1177/0272431613519498. Craig. W. Bukowski.. (2000). C. and bully-victim Golmaryami. M. Van de Schoot. hostility and aggressive behavior. Muthén. 36. 12. L. P. J. Goossens. analytic review of gender differences. Menon. Journal of Personality. The new participant role Salmivalli. Methodological issues in the use of Fanti. Bullies. D. 854–878. Salmivalli.. (2005)... 79. C. D. Quarterly.. Y. psychiatric inpatients. Costs and benefits of Fossati. L. Eisenberg. R.. A.and peer valued characteristics. 24. (2013a). 18–34. L. C. Van de Schoot. Aleva. T. Understanding and preventing bullying. 381–458. Farrington. 22. F. N. based evolutionary perspective.. 82. in children. The associations of self.. 103. J. K. Goossens.. P. (2009). H. 343–357. & Pilkonis. S. Analyzing developmental trajec. D.. (2005). & B.. (2008). J. P. Lim. S. N. F. M. M. 31... M. (2012). S.. MPlus user’s guide (6th ed. C. C. Nagin. Developmental Review. A.. (2013). (2006). and indications for their validity. (2006). W. 391. ment. & Dishion.. relationships. W. (2001)... 1217–1229.. Fanti.. Guilford. (2008). D. M. Cillessen & W. Stegge. Journal Reijntjes. Child Abuse & Research on Adolescence. and psychopathy.74 J Abnorm Child Psychol (2016) 44:63–74 References Pauletti. 25. & Henrich.. 90. H. 52.. J. & Kerig. 97–132. 19. Andershed. and development. 177–196.. B. Meulen. (2008b). K. and aggression Angeles: Muthén & Muthén. K. The two pathways to Terry. Aggressive Behavior.. E. 721–735. H. 199–209. (2010).. 325–338. victims. D. & Hoijtink. & Maffei. (1996). The dark triad of personality: exploitativeness to bullying behavior: the mediating role of narcissism. A. I. O. J. Developmental tra- role of conduct problems and psychopathic traits. N. K. E. G. Stegge. 49. K. 224–234. Thomaes. L. R. traits inventory: child version.. K. 128–133. & Denissen. J. Aggressive Behavior. Aleva. Cambridge: Harvard University Press. (2010). ogy.. Little. 382– Inquiry. Pospikarta.. Poulin. Research in Personality. and relations 79. & Kimonis. E. 41. T. M. H. Aggressive trajectory classes. Aggressive Behavior. E. 57–67. Scholte. 396–408. A.. K. A. Morf. & Peets.. Vermande. P. A. Los (2008a). aggres- Kärnä. 6. Muthén. Psychological Methods. 311–330.. Laudy. (2001). 52. E. & Brummelman. 19. Kaukiainen. W. F. Child Development. & Maio. Research. San Francisco: Jossey–Bass. Remarks on self-esteem... & Little. Vermande. Developmental and indirect aggression during childhood and adolescence: a meta. From narcissistic Paulhus. sociometric techniques. Voeten. D. & van der Psychiatry. peer sociometric nominations with middle school youth. doi:10. M. & Van der Meulen. Alcoholism: Clinical and Experimental Behavior. 35. & Lim.. 44.. Psychological cissism scale. . A. to maladjustment. results.. R. Madhavi. H. L. J.. 281–299. Psychology. (2007). Journal of Abnormal Child Psychology. M. 556–563. Bullying as strategic behavior: relations with Von Eye. & Muthén. Y. Direct Pepler. Cicchetti (Ed. S. Veenstra. & Muthén. 21–27. course. J.. Campbell. Bukowski Laursen. (2002). trajectories of bullying and associated factors. Card. & Tremblay. & Perry. R. Journal interaction. Child Psychiatry and Human Development. N. Bayesian evaluation of tion with psychopathic narcissism and theory of mind among child inequality and equality constrained hypotheses for contingency ta. 17. (2011). M.. S. K. Journal of Child Psychology and Psychiatry.. E. Vermande. B. 37. J. A. Merrill Palmer Quarterly. C. C. M. C. B. T. Person-centered and variable-centered ap. R. Effects of narcissism and self.. 612–620. Narcissistic Thomaes. & Hoff. & Van der Meulen. Kuefner. D. K. R. T. and prestige.. A. Menon.. Personality and Social Psychology Bulletin. 831–837. Social Development. & Back. New York: of Clinical Child and Adolescent Psychology. jection in the peer system. G.. 1268–1278. Merrill-Palmer Psychology.. M. M. 36. 617–631. Child Development. B. Stegge. measured through self-report? Initial test of the youth psychopathic Psychological Methods. Psychopathic traits. W. International Journal of Law and Olthof. & Hymel. Nagin. A.. Integrating person-centered and Vaillancourt. D. T. Sedidikes. (2010). (1999). J. Developmental psychopathol- personality disorder: relations with distress and functional impair. P.. E. H. Sociometric Assessment. intercorrelations. (2010). J. E.. Tobin. K. Recent advances in the measurement of acceptance and re- proaches to longitudinal data. T. M. C. 67.. (2008). (2011). B. & sion. Stucky. Thomaes. D. B. & Barry. O. C. R. R. 390–420. G. Can psychopathic traits in children be tories of distinct but related behaviors: a group-based method.. Ringleader bullying: associa- Klugkist. Machiavellianism. 170–177. (2012). (1993). D.. T. S. Development and validation of the childhood nar- cissism: a dynamic self-regulatory processing model. Reijntjes. In A.. R. Olthof. victim distress.. P. Olthof. (2001). Bushman.. Comprehensive Psychiatry. B.. Aleva. (2015). A... A. narcissism in nonclinical adolescents.. H. 1–13. E. T.. A large-scale evaluation of the KiVa antibullying program: Grades 4–6. New York: Wiley. Kaljonen. 48. S.. (2008). 32. 39. Goossens. & Williams. & Rhodewalt. peer-evaluated self-esteem.. R. 32.. Lindenberg. C. P.