Narcissism and popularity among peers: A cross-transition longitudinal study

ABSTRACT The dual-pathway model posits that narcissism can both benefit and compromise popularity, depending upon whether narcissists’ assertive or adversarial interpersonal tendencies surface in social interaction. A 5-wave longitudinal study followed Dutch adolescents (N = 322, 53% female, M age = 12.2) who transitioned from primary into secondary school and examined how narcissism, along with self-esteem (measured at the end of primary school), contributes to cross-transition change in peer-rated popularity. Narcissism predicted rank-order increases in popularity among children with modest self-esteem but decreases in popularity among children with high self-esteem. These effects emerged shortly after the transition and were maintained throughout the school year. The results illustrate how self-esteem can act as a marker for the different faces of youth narcissism.

school, many children seek to attain status in their new peer group (Adler & Adler, 1998;Brown, 2011;LaFontana & Cillessen, 2010). As a one-two punch, the secondary school transition is timed in early adolescence, a developmental period during which children become increasingly concerned about the impressions they make on their peers (Eccles et al., 1993;Harter, 2012;LaFontana & Cillessen, 2010).
The formation of popularity hierarchies is an ongoing process, but can be broken down into the stages of "popularity attainment" (becoming popular in a new peer group) and "popularity maintenance" (staying popular; Cillessen, 2011). Popularity attainment requires children's motivation and ability to be socially visible (i.e., to grab and hold the attention of peers) and to have an impact on others. Popularity maintenance, in turn, additionally requires children to be able to anticipate how their actions influence their popularity, and to flexibly adjust to changes in peer group norms (Rubin, Bukowski, & Parker, 2006).

Narcissism and the "dual-pathway model" of popularity
We focus on narcissism as a trait on which youth and adults in the general population vary (Barry, Wallace, & Guelker, 2011;Sedikides, Rudich, Gregg, Kumashiro, & Rusbult, 2004;)a trait characterized by grandiosity, entitlement, and exploitative interpersonal attitudes (Thomaes, Stegge, Bushman, Olthof, & Denissen, 2008). We call an individual with high values on this trait a "narcissist" in the present paper. Much like their adult counterparts (Morf & Rhodewalt, 2001), narcissistic youth have a strong need to get attention and be admired by others, and they often use their social relationships as a platform to garner approval or status Thomaes, Brummelman, & Sedikides, 2018). For example, research has shown that narcissistic youth are prone to present themselves to others as attractive, fashionable, or 'cool' (Ong et al., 2011; for similar evidence in college students, see Back, Schmukle, & Egloff, 2010), they are socially facile and easily disclose personal information to others (Hawk, van den Eijnden, van Lissa, & ter Bogt, 2019;Liu, Ang, & Lwin, 2016), and they seek approval and admiration (more than closeness) in their relationships (Ojanen, Findley, & Fuller, 2012;. It seems plausible that these self-promotional, agentic tendencies will typically help narcissists to gain popularity among peers. Indeed, if attaining popularity requires a drive and ability to be visible and have an impact, then narcissists have what it takes. However, research has also shown that narcissistic youth can behave in ways that potentially compromise popularity. In particular, narcissistic youthor at least subsets of themcan behave abusively and aggressively (Golmaryami & Barry, 2010;Reijntjes et al., 2016;, and are often perceived by peers as competitive and hostile (Grafeman, Barry, Marcus, & Leachman, 2015).
How successful, then, are narcissists at gaining popularity? The "dual-pathway model" (Küfner, Nestler, & Back, 2013;Leckelt, Küfner, Nestler, & Back, 2015) explains how narcissism may influence popularity in the early stages of relationship formation. This model posits that narcissism can influence the degree to which individuals become popular or unpopular, depending on whether their interpersonal style predominantly manifests in narcissistic assertive behaviors (e.g., dominance, agentic tendencies) or narcissistic adversarial behaviors (e.g., self-centeredness, lack of genuine concern for others, competitiveness, envy). Küfner et al. (2013) illustrated the model in research involving college students. In the context of a brief (<15 min) group discussion with unacquainted peers, narcissism was positively related to peer judgments of both "assertiveness" and "aggressiveness." These judgments, in turn, had opposing effects on the impressions that peers formed: Assertiveness was positively related to popularity, whereas aggression was negatively related to popularity. Leckelt et al. (2015) found support for the model by systematically varying social context. In the context of self-presentation, narcissists' assertiveness surfaced in social interactions, which predicted increased popularity with peers. In the context of task-oriented interaction with peers, narcissists' adversarial tendencies surfaced, which predicted decreased popularity. Recently, the contextual evidence for the dual pathway model has been complemented with an individual difference approach. Research showed how narcissism has differential effects on popularity across subsets of narcissists who differ in their tendencies to behave assertively versus aggressively (e.g., Back, Küfner, & Leckelt, 2018;Leckelt et al., 2015). Narcissists' level of self-esteem and popularity across the secondary school transition Some initial attempts have been made to distinguish between different "faces" (or facets) of youth narcissism. For example, some scholars have tried to measure vulnerable manifestations of narcissism in addition to grandiose manifestations (Derry, Bayliss, & Ohan, 2018). Others have distinguished between the facets of narcissistic superiority and exploitativeness (Ang & Raine, 2009). Still, others distinguished between maladaptive and adaptive facets of narcissism, based on a factor analysis of the Narcissistic Personality Inventory for Children (Barry, Frick, Adler, & Grafeman, 2007). Despite the informativeness of these approaches, broad consensus on whether youth narcissism actually is a multifaceted trait (and if so, what youth narcissistic facets can be distinguished) is still lacking  There is, however, a consensus that narcissistic youth differ in their level of selfesteem and concomitant social behavior (Barry, Grafeman, Adler, & Pickard, 2007;Pauletti, Menon, Menon, Tobin, & Perry, 2012;Thomaes et al., 2018). Indeed, narcissism and self-esteem are relatively independent psychological traits, especially so in youth (Barry & Ansel, 2011;Brummelman, Thomaes, & Sedikides, 2016). The self-absorption and feelings of entitlement that characterize narcissism may co-occur with either self-assurance and hubris (in narcissistic youth with high self-esteem), or more measured feelings of worth and occasional self-doubt (in narcissistic youth with modest self-esteem; Pauletti et al., 2012;Thomaes et al., 2018).
Importantly, narcissistic youth with high self-esteem differ from narcissistic youth with more modest self-esteem when it comes to their interpersonal style. Compared to narcissistic youth with modest self-esteem, those with high self-esteem are more likely to be hostile, abusive, and aggressive. For example, in one experimental study in which young adolescent participants played a competitive computer game against a peer and lost, narcissists with high (but not modest) self-esteem were prone to aggress against their opponent when they experienced ego-threat . Similar evidence has been obtained using self-report and peer nomination aggression data (e.g., Golmaryami & Barry, 2010;Lau, Marsee, Kunimatsu, & Fassnacht, 2011; but see Pauletti et al., 2012). A recent study among adults also provided evidence that narcissists with high self-esteem are more aggressive than those with modest self-esteem. Using cluster analyses, this study identified two groups of individuals high on entitlement (a key feature of narcissism): Both groups had equally high levels of extraversiona trait characterized by assertive behavior. One group was characterized by high levels of both self-esteem and antisocial behavior (including aggression), whereas the other group was characterized by low levels of self-esteem and low levels of antisocial behavior (Crowe, LoPilato, Campbell, & Miller, 2016).
How will these differences in interpersonal style play out in terms of how successful narcissists are at attaining popularity in a new peer group? Although narcissistic youth (regardless of their level of self-esteem) are socially facile and skilled at obtaining attentiontraits that may benefit their popularity (Liu et al., 2016;Ojanen et al., 2012;Ong et al., 2011) the competitive, hostile, and aggressive behaviors that narcissistic youth with high self-esteem tend to engage in are likely to backfire and compromise popularity. Accordingly, we propose that narcissism will foster popularity among adolescents with modest levels of self-esteem, but not among those with high self-esteem.
That said, different dynamics might apply for the short-term attainment and longerterm maintenance of popularity. For example, research involving college students has shown that the popularity benefits associated with narcissistic assertiveness decrease over time, while other facets of narcissism (e.g., exploitativeness, entitlement) become increasingly influential and decrease narcissists' popularity (Carlson & Lawless DesJardins, 2015;Leckelt et al., 2015). Thus, even if narcissists with modest self-esteem may readily attain popularity in the short term because they come across as assertive rather than aggressive, in the longer term their self-centeredness and entitlement may be picked up by peers the more they interact with them, and turn them off (Carlson & Lawless DesJardins, 2015;Paulhus, 1998). If so, then narcissists with modest self-esteem may eventually lose at least some of their initial status among peers and gradually decrease in popularity.

Overview
The present longitudinal study examines how narcissism, along with self-esteem, predicts children's popularity attainment and subsequent maintenance. It does so right at a time when children enter a new peer-group, with the transition into secondary school. Five waves of data collection were timed from the end of primary school (T1 = spring at primary school) continuing throughout the first year of secondary school (T5 = spring at secondary school).
We hypothesized that narcissism would interact with self-esteem in predicting both popularity attainment (i.e., rank-order change in popularity from T1 to T2; 2 to 3 weeks after the transition) and popularity maintenance (i.e., rank-order change in popularity throughout the first school year; from T2 to T5), but in different ways. For children with lower self-esteem, we expect narcissism to predict initial increases in status (i.e., in the popularity attainment stage) but subsequent decreases in status (i.e., in the popularity maintenance phase). For children with high self-esteem, we expect narcissism to predict no increases in status (i.e., neither in the status attainment nor in the status maintenance stages).

Participants
Participants were 322 students (53% girls) from schools in a medium-sized town (35,000 inhabitants) in the Netherlands. All participants transitioned from primary to secondary school. At the study start, students were on average 12.2 years old (SD = 0.45). Most students (77%) were of ethnic Dutch origin; others were mainly of mixed cultural or ethnical origin. As is common in the Dutch secondary school system, students transitioned into tracks ranging from pre-vocational to pre-university training.
Schools were selected following a two-step procedure. First, all secondary schools in town were asked to take part in the study; four (out of six) schools agreed. Second, all primary schools feeding at least five students into these secondary schools were asked to take part; 22 (out of 29) schools agreed. Next, parental consent letters were sent home from primary schools, were signed, and returned (consent rate = 75%). This procedure resulted in an initial Grade 6 sample of 485 participants. Of them, a proportion of 66% transitioned into the secondary schools that took part in our study (i.e., the 322 participants in the transition sample). Participants in the transition sample did not differ from students who only participated in Grade 6 in terms of self-esteem, age, gender, and cultural or ethnical diversity (ps > .56). Participants in the transition sample were higher in narcissism and popularity (ps < .02), but effect sizes were small (ds < .29). Students in all academic tracks were represented in the transition sample. Attrition was 5% at T3, 3% at T4, and 3% at T5. Missing values were estimated using Full Information Maximum Likelihood (FIML) in Mplus 6.0 (Enders & Bandalos, 2001).

Procedure
In the Netherlands, children transition into secondary school at Grade 7. The five waves of data collection were timed from the end of primary school (T1 = spring prior to the transition) continuing throughout the first year of secondary school (T2 = September, 2 to 3 weeks after the transition; T3 = December; T4 = March; T5 = June). Questionnaires and peer rating procedures were administered in students' classes.

Measures
Narcissism Narcissism was measured at T1 using the 10-item Childhood Narcissism Scale . This scale measures narcissism as a unidimensional construct, tapping children's grandiose, inflated self-views, feelings of entitlement, and exploitative interpersonal attitudes. Sample items include: "I think it is important to stand out" and "I love showing all the things I can do." Items were rated along a 4-point scale (1 = not at all true, 4 = completely true). Higher scores indicate higher levels of narcissism (Cronbach α = .79).
Self-esteem Self-esteem was measured at T1 using the 5-item Global Self-Worth scale of the Self-Perception Profile for Adolescents (Harter, 1988). This scale measures how satisfied children are with themselves. A sample item includes: "Some kids are happy with themselves". As in the previous research (Brendgen, Vitaro, Turgeon, Poulin, & Wanner, 2004;, items were rated along a 4-point scale (1 = I am not like these kids at all, 4 = I am exactly like these kids). Negative items were recoded, with higher scores indicating higher levels of self-esteem (Cronbach α = .75).

Popularity
Popularity was measured at each time point using standard peer rating procedures (Schwartz, Gorman, Nakamato, & McKay, 2006;Singleton & Asher, 1977). Participants rated for each of their classmates how "popular" they thought these classmates were. Ratings were provided on a 5-point scale (1 = not at all, 5 = very much). Because of the school transition (between T1 and T2), T1 involved a different set of raters for each child than T2 to T5. Received ratings (M received ratings in primary school = 17, SD = 4.9; M received ratings in secondary school = 21, SD = 4.6) were averaged, with higher scores indicating higher levels of popularity.
As in the previous research involving youth (Barry & Ansel, 2011;, and consistent with their conceptualization as separable dimensions of the self , narcissism and self-esteem were unrelated. Boys were more narcissistic than girls (p < .001, d = 0.43); no sex differences were found for self-esteem or popularity (ps > .13).

Primary analyses
We used structural equation modeling and the Latent Moderated Structural equation technique (LMS; Klein & Moosbrugger, 2000) to test our hypotheses in Mplus 7.2  (Muthén & Muthén, 1998-2012. LMS allows for testing interactions with latent variables (i.e., narcissism and self-esteem). To account for dependencies in the data (i.e., students were nested in classes), we analyzed the data using Mplus's robust maximum likelihood estimator, Type = Complex (Asparouhov, 2005). Because standard fit indices are not available for models that use latent variable interactions (due to adjustments made during estimation; Muthén & Muthén, 1998-2012, we looked at the fit for the model including main effects and compared the model with interactions with this model. We estimated a model (Figure 1) that included stability paths of popularity (from T1 to T5), main effects of narcissism and self-esteem (at T1) at each wave of popularity, and the interaction between narcissism and self-esteem (at T1) at each wave of popularity. Predictors were centered to reduce multicollinearity (Cohen, Cohen, West, & Aiken, 2003). Factor loadings can be found in the Appendix.

Prediction of popularity attainment and maintenance
The model including main effects for narcissism and self-esteem showed adequate fit, χ 2 (160) = 330.20, p< .001, χ 2 /df = 2.06, TLI = .94, CFI = .93, RMSEA = .06. The model including the interaction between narcissism and self-esteem demonstrated a better fit compared to the model without interactions. That is, the loglikelihood was significantly closer to zero (−5559.87 vs. −5553.67, χ 2 (5) = 12.39, p = .03) in the model with interactions compared to the model without interactions. As shown in Figure 1 and Table 2, popularity showed considerable rank-order stability over time. Popularity was less stable across the transition (from T1 to T2) than after the transition (from T2 to T5): Not surprisingly, children were most likely to show rank-order change in popularity across the secondary school transition, when they entered a new peer group (i.e., and they were thus evaluated by a new group of raters). Later in the school year, their popularity stabilized to a large degree. The goal of this research was to investigate how narcissism, along with self-esteem, predicts popularity attainment (i.e., popularity shortly after the school transition at T2, controlled for popularity in primary school at T1) and popularity maintenance (i.e., popularity later in the school year at T3, T4, and T5, controlled for popularity at the previous time point). Main effects of narcissism and self-esteem for popularity attainment and maintenance were nonsignificant. For popularity attainment, the Narcissism × Self-esteem interaction was significant. We conducted a region of significance analyses (Preacher, Curran, & Bauer, 2006) to test how self-esteem qualified the link between narcissism and popularity attainment. Region of significance analyses provide the range of moderator values (i.e., values of self-esteem) for which the association between the predictor (i.e., narcissism) and the outcome (i.e., popularity attainment) is significant. As predicted, the region of significance analyses showed that narcissism was significantly positively associated with popularity attainment among children with modest selfesteem; those whose absolute self-esteem levels ranged up to 2.9 on the 1-4 selfesteem response scale (i.e., near the neutral midpoint of the scale at which children report equal amounts of self-satisfaction and self-dissatisfaction). Narcissism was not associated with popularity attainment among children whose absolute self-esteem levels ranged between 2.9 and 3.7 (i.e., intermediate levels of self-esteem). Finally, narcissism was significantly negatively associated with popularity attainment among children with high self-esteem; those whose absolute self-esteem levels were 3.7 or higher on the 1-4 self-esteem response scale. Figure 2 shows the boundaries of the region of significance, along with the simple slopes for children who score one standard deviation below the mean (self-esteem score = 2.7, b = 0.27, t = 2.22, p = .027, β = .20), and one standard deviation above the mean (self-esteem score = 3.8, b = −0.21, t = −2.11, p = .035, β = −.16) on the self-esteem scale.
Regarding popularity maintenance, the Narcissism × Self-esteem interaction did not significantly predict popularity change from T2 onwards. Thus, we found no support for the hypothesis that narcissists with modest self-esteemwho initially attained popularitywould subsequently decrease in popularity over the course of the school year. Rather, regardless of their level of self-esteem, narcissists maintained their initially attained position in the popularity hierarchy throughout the school year.

Concurrent popularity
Although our hypotheses did not concern effects for concurrent popularity (at T1), we considered them for exploratory purposes. Both narcissism and self-esteem were significantly positively associated with concurrent popularity, and the Narcissism × Selfesteem interaction was not significant.

Potential gender differences
Because boys were more narcissistic than girls, we explored potential gender differences in the links between narcissism, self-esteem, and popularity. The model was re-analyzed using a multigroup procedure, which compared a model in which regression parameters were constrained to be equal across gender to a model in which regression parameters were allowed to differ across gender (Kline, 1998). By default, means, intercepts, and variances were allowed to differ across gender. Models were compared using the Satorra-Bentler scaled χ 2 -difference test (Satorra & Bentler, 2001). The links between high self-esteem (+1 SD = 3.8) lower bound region of significance (=2.9) upper bound region of significance (=3.7) Figure 2. Narcissism predicts decreased popularity (T1-T2) for children with high self-esteem and increased popularity for children with modest self-esteem. The black solid line represents the significant simple slope of the self-esteem score one standard deviation above the mean. The black dashed lines represent the significant simple slope of the self-esteem score one standard deviation below the mean. The grey lines represent the boundaries of the region of significance. Associations between narcissism and popularity are significant at self-esteem values above the upper boundary (= 3.7) and below the lower boundary (= 2.9).
narcissism, self-esteem, and children's popularity were similar for boys and girls. Releasing equality constraints across gender did not significantly improve model fit, Δχ 2 (11) = 17.71, p = .09.

Discussion
The present study found longitudinal support for the overall hypothesis that narcissism can both benefit and compromise peer popularity Leckelt et al., 2015). We followed children across the transition from primary into secondary schoolan eventful time when new relationships and popularity hierarchies are formed. Narcissism interacted with self-esteem to predict popularity attainment (i.e., rank-order change in popularity from shortly before to shortly after the transition). Specifically, narcissism predicted increased cross-transition popularity among children with modest self-esteem, decreased cross-transition popularity among children with high self-esteem, a no change in popularity for children with intermediate self-esteem. Later in the school year, children's levels of popularity stabilized and changes in popularity were no longer predicted by the interaction of narcissism and self-esteem. We found no indications that the popularity of those narcissists whose popularity increased across the school transition waned over time when peers got to know them better. Previous evidence that narcissists can make both positive and negative impressions on their peers was obtained in get-acquainted tasks involving college students who first introduced themselves and then engaged in group discussions (Carlson & Lawless DesJardins, 2015;Küfner et al., 2013;Leckelt et al., 2015). Our results extend this evidence in that they are based on a sample of young adolescents who were followed over an extended period of time, in a naturalistic setting, and following a major life transition.
Our results are consistent with the dual-pathway model. Narcissistic youth tend to have a drive and ability to be seen and obtain power or admiration (i.e., assertive behavior in the dual-pathway model; Liu et al., 2016;Ojanen et al., 2012;Ong et al., 2011). And yet, some narcissistic youththose with high self-esteemalso engage in interpersonal behaviors that are likely to turn others offthey tend to be competitive, hostile, and aggressive (Golmaryami & Barry, 2010;Lau et al., 2011;. It is thus not surprising that we found differences in the popularity attainment of narcissists with high versus modest self-esteem. What does it mean for narcissistic youth to hold "modest" self-esteem? Although one might intuitively believe that all narcissists should have high self-esteem, this is not the case . Although some narcissists are self-assured, others occasionally experience feelings of imperfection and self-doubt (along with feelings of entitlement and specialness; Geukes et al., 2017;Thomaes, Stegge, Olthof, Bushman, & Nezlek, 2011;Zeigler-Hill & Besser, 2013). The present research illustrates, for the first time, that this particular constellation of narcissism and modest self-esteem may have certain interpersonal benefits, at least in terms of youth's social standing among peers. Narcissists' moderate feelings of worth may come with an interpersonal style that is perceived as less hostile and aggressive by peers (Crowe et al., 2016). Future research will need to further explore the psychological costs and benefits of this interpersonal style.
Some of our findings were unexpected. Previous research in college students has suggested that narcissists make increasingly negative impressions on their peers and lose status over time (Carlson & Lawless DesJardins, 2015;Leckelt et al., 2015;Paulhus, 1998). Accordingly, we predicted that those narcissistic individuals who were initially successful at attaining popularity would gradually decrease in popularity over the school year. We found no evidence for such a trendpopularity hierarchies were remarkably stable after they were established at the beginning of the school year, and narcissists whose popularity increased shortly after the school transition were able to uphold their popularity throughout the school year. What accounts for this apparent discrepancy with previous research? The developmental timing of our research may provide an explanation. In early adolescence, compared to early adulthood, peers may be less skilled at recognizing the relatively concealed, subtle cues of entitlement and superiority that narcissists with modest self-esteem exhibit, and so they may be less easily turned off by thema possibility that deserves further empirical scrutiny.

Limitations and future research
A central tenet of the dual-pathway model of narcissistic popularity is that narcissists' level of popularity depends on how assertiveness and aggression surface in social interaction. We did not measure those behaviors directly. Instead, we measured children's level of self-esteema marker of individual differences in narcissists' interpersonal style (e.g., Bushman et al., 2009;Crowe et al., 2016;Golmaryami & Barry, 2010;. Future research may seek to identify the social behaviors that explain the heterogeneous effects of narcissism on children's popularity among peers. Intensive methods of data collection (e.g., daily diary methods, repeated observational methods; Pellegrini & Bartini, 2000;Sandstrom & Cillessen, 2003), timed during early stages of peer relationship formation, may be especially suited for such an approach.
The measure of youth narcissism that we used, the Childhood Narcissism Scale, assesses narcissism as a single-dimension construct, and was not developed to disentangle potential facets of youth narcissism. A viable direction for future work is to explore the multifaceted nature of youth narcissism and its interpersonal correlates, including popularity attainment and maintenance. For example, such research could explore whether assertive and antagonistic facets of narcissism can be distinguished in youth  and whether these have opposing consequences for popularity (Leckelt et al., 2015), as has been found in adult populations.
Another interesting direction for future research may be to study how youth narcissism (along with self-esteem) is not just a predictor of, but indeed, a consequence of changes in popularity, such as they may occur across the secondary school transition. Still, little is known about whether or how narcissism may be rooted in impactful peer experiences. An ideal developmental period to explore such effects is in (early) adolescencea time when youth care deeply about how they are viewed by peers.

Conclusion
Narcissism has many faces, but in youth, these faces are only beginning to be unveiled. Our research illustrates how self-esteem is a marker of different faces of youth narcissism, one that is consequential for the popularity of narcissistic youth. Narcissism benefits the popularity attainment of children with modest self-esteem but compromises the popularity attainment of those with high self-esteem.

Disclosure statement
No potential conflict of interest was reported by the authors.
Factor Loadings for the Items of the Childhood Narcissism Scale and the Global Self-Worth Scale All loadings and correlations were significant at p < .001, except item 4 for narcissism, at p = .002. Dashes indicate that loadings were not estimated but fixed at 1.