You are on page 1of 11

Developmental Psychology

The Functions of Aggression in Gaining, Maintaining,


and Losing Popularity During Adolescence: A Multiple-
Cohort Design
Yvonne H. M. van den Berg, William J. Burk, and Antonius H. N. Cillessen
Online First Publication, August 1, 2019. http://dx.doi.org/10.1037/dev0000786

CITATION
van den Berg, Y. H. M., Burk, W. J., & Cillessen, A. H. N. (2019, August 1). The Functions of
Aggression in Gaining, Maintaining, and Losing Popularity During Adolescence: A Multiple-Cohort
Design. Developmental Psychology. Advance online publication.
http://dx.doi.org/10.1037/dev0000786
Developmental Psychology
© 2019 American Psychological Association 2019, Vol. 1, No. 999, 000
ISSN: 0012-1649 http://dx.doi.org/10.1037/dev0000786

The Functions of Aggression in Gaining, Maintaining, and Losing


Popularity During Adolescence: A Multiple-Cohort Design
Yvonne H. M. van den Berg, William J. Burk, and Antonius H. N. Cillessen
Radboud University

This study examined associations between perceived popularity trajectories and changes in proactive and
reactive aggression in middle adolescence. The sample included 1,374 adolescents from 5 cohorts who
were followed from Grade 7 to Grade 9 (49% males, Mage ⫽ 12.67 years, SD ⫽ .67). Popularity
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

trajectories were identified based on initial status and changes in status over time, and linear mixed
This document is copyrighted by the American Psychological Association or one of its allied publishers.

effects models tested group differences in changes of aggression. The results partially confirmed our
expectation that changes in popularity were associated with specific developmental patterns of proactive
and reactive aggression and also varied as a function of status already achieved. Specifically, adolescents
in groups characterized by gains in status increased in proactive aggression, and those in groups
characterized by losses in status decreased in proactive aggression. A more nuanced pattern of group
differences emerged for reactive aggression. Changes in reactive aggression were only found for groups
characterized by changes in unpopular status. Taken together, these findings suggest that proactive
aggression has universal benefits for popularity in middle adolescence, whereas reactive aggression has
differential costs. Implications and directions for future research are discussed.

Keywords: popularity, proactive aggression, reactive aggression, peer status, adolescence

Numerous studies have shown that aggression is related to a perceived popularity and the two functions of aggression (i.e.,
wide range of negative outcomes and maladjustment (Card, proactive and reactive aggression) during adolescents’ first 3 years
Stucky, Sawalani, & Little, 2008). Yet, in contrast to research on in secondary education. Second, to test the idea that prospective
the detrimental nature of aggression, more and more studies argue associations between popularity and the two functions of aggres-
aggression can also be functional and beneficial (Ellis et al., 2012; sion depends on adolescents’ existing status. Specifically, this
Hawley & Bower, 2018). Specifically, they argue that status within study investigated whether adolescents who obtain, maintain, or
a peer group can be achieved and maintained through the strategic lose popular status and those who obtain, maintain, or lose unpop-
use of aggression, with high status individuals using aggression ular status differ in their use of proactive and reactive aggression
deliberately and in a goal-directed manner to demonstrate their over time.
dominance and control over resources. Empirical studies have
found support for this by demonstrating that perceived popularity
is positively associated with more deliberate and goal-directed
Functions of Aggression
aggression (i.e., proactive aggression), and negatively associated Two functions of aggression are generally distinguished in the
with unintentional and retaliatory acts of aggression (i.e., reactive literature: reactive and proactive (Card & Little, 2006; Crick &
aggression; Prinstein & Cillessen, 2003). However, the association Dodge, 1996; Hubbard, McAuliffe, Morrow, & Romano, 2010;
between perceived popularity and aggression has recently been Polman, Orobio de Castro, Koops, van Boxtel, & Merk, 2007).
shown to be curvilinear, with both popular and unpopular adoles- Reactive aggression stems from the frustration-aggression model
cents exhibiting more proactive and reactive aggression than ad- (Berkowitz, 1993; Dollard, Doob, Miller, Mowrer, & Sears, 1939)
olescents of average status (Stoltz, Cillessen, van den Berg, & and is defined as a defensive, retaliatory, and angry reaction to an
Gommans, 2016). actual or perceived offense (Crick & Dodge, 1996). It derives
The current study therefore served two purposes. First, to rep- primarily from feelings of frustration (Dollard et al., 1939; Hub-
licate and extend previous findings on the associations between bard et al., 2010) and is often described as a dysregulated and
undercontrolled response caused by a range of social information
and emotion processing deficits (Card & Little, 2006; Hubbard et
al., 2010). Proactive aggression is rooted in social– cognitive
learning theory and refers to deliberate acts directed at obtaining a
X Yvonne H. M. van den Berg, William J. Burk, and Antonius H. N.
material or social reward (Bandura, 1973; Card & Little, 2006). It
Cillessen, Behavioural Science Institute, Radboud University.
The authors are grateful to the respondents, teachers, and school admin-
is therefore also called planned, instrumental, goal-oriented, or
istrators who made this research possible. “cold-blooded” aggression.
Correspondence concerning this article should be addressed to William Reactive and proactive aggression increase from childhood to
J. Burk, Behavioural Science Institute, Radboud University, P.O. Box adolescence, with a peak in middle adolescence (Fite, Colder,
9104, 6500 HE Nijmegen, the Netherlands. E-mail: w.burk@psych.ru.nl Lochman, & Wells, 2008). Concurrent associations between pro-

1
2 VAN DEN BERG, BURK, AND CILLESSEN

active and reactive aggression are moderate to strong across de- levels of popularity, both concurrently and over time. These stud-
velopment (Fite et al., 2008; Ojanen & Kiefer, 2013; Ostrov, ies both indicated that proactive aggression is more common
Murray-Close, Godleski, & Hart, 2013). Yet, the strength of the among popular youth, whereas reactive aggression seems to be
association varies considerably across studies (Polman et al., more common among unpopular youth. However, the findings
2007). Proactive and reactive aggression have also been found to from Prinstein and Cillessen (2003) and Stoltz and colleagues
have different precursors, correlates, and consequences (Hubbard (2016) were based on relatively small samples and a relatively
et al., 2010; Murray-Close & Ostrov, 2009; Ostrov et al., 2013). short time span (1 to 1.5 years). Moreover, the results were to some
This has led to the consensus that proactive and reactive aggres- extent inconsistent with one another. Therefore, there is a need for
sion are related, yet distinct types of aggression with unique replication with larger samples over a longer period of time.
precursors and consequences for child and adolescent development As popularity is a top priority in middle adolescence (LaFontana
(Poulin & Boivin, 2000). & Cillessen, 2010), it is particularly important to examine the
associations of popularity with proactive and reactive aggression
in this specific developmental period. Proactive and reactive ag-
Popularity and Functions of Aggression
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

gression both seem to peak in middle adolescence (Fite et al.,


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

The definitions of proactive and reactive aggression suggest that 2008; Ojanen & Kiefer, 2013). Yet, studies also show that high
both functions are distinctively related to youths’ social status. levels of proactive aggression remain rather stable or sometimes
Traditionally, social status has been conceptualized in terms of even further increase, at least among a subgroup of middle ado-
acceptance and likability in the peer group (i.e., sociometric pop- lescents (Ojanen & Kiefer, 2013; Vitaro, Brendgen, & Barker,
ularity; Coie, Dodge, & Coppotelli, 1982). Later, researchers have 2006). It is argued that this is partially driven by increasing
realized that youths who are highly accepted by their peers are not concerns about social status (Ojanen & Kiefer, 2013). Hence, it is
necessarily popular, powerful, or influential (LaFontana & Cil- in this developmental stage that adolescents are more concerned
lessen, 1998; Parkhurst & Hopmeyer, 1998). As a result, measures about their status within the larger peer group and are most likely
of perceived popularity were developed to assess a form of social to prioritize status over other social goals (Cillessen, Mayeux, Ha,
status reflecting impact, visibility, and social dominance among de Bruyn, & LaFontana, 2014; LaFontana & Cillessen, 2010).
peers (see, e.g., Cillessen, 2011). Previous studies investigating the Therefore, the first goal of this study was to examine prospective
association between social status and the functions of aggression associations between (changes in) popularity and changes in pro-
predominantly examined status in terms of peer acceptance. These active and reactive aggression in middle adolescence. We examine
studies have shown that reactive, but not proactive, aggression is the development of popularity and both functions of aggression
associated with lower levels of peer acceptance (i.e., peer rejec- across 3 years in middle adolescence, a time when attaining
tion) and impaired psychosocial functioning (Card & Little, 2006; popularity is a top priority and aggression is often seen as a means
Hubbard et al., 2010; Ostrov et al., 2013). Yet theoretically, it is to achieve status.
perceived popularity that should be related to proactive aggression.
The evolutionary perspective on aggression does not view ag-
Functions of Aggression Depend on Status Achieved
gression simply as an undesired behavior, but as a means to gain
resources when used effectively (Ellis et al., 2012; Hawley & In the studies by Prinstein and Cillessen (2003) and Stoltz and
Bower, 2018). As resources (e.g., material objects, information, colleagues (2016), curvilinear associations between popularity and
and social capital) are often limited, members within any given the function of aggression were also reported. These effects were
society have to compete for these resources. This can be done by interpreted as indicating that popular and unpopular adolescents
using prosocial or cooperative strategies, or by using more aggres- exhibit higher levels of both reactive and proactive aggression than
sive and coercive strategies. Irrespective of the strategies used, adolescents of average status. Moreover, it was argued that both
those who control the most resources are considered to be the most popular and unpopular youth display (increasing levels of) proac-
successful competitors and have the highest status within their tive aggression to gain status. This observation leads us to question
social group. As such, social status predicts high and increasing whether the prospective association between popularity and pro-
levels of aggression as a means of demonstrating individuals’ active and reactive aggression are actually the same for youth at
dominance and control over resources. Based on this perspective, the top and bottom of the status hierarchy. To investigate this idea,
proactive aggression may be especially related to social status in we operationalized popular status in terms of an adolescent’s
terms of perceived popularity (i.e., social dominance), but not initial membership in one of three popularity groups (popular,
necessarily in terms of peer acceptance (i.e., likability). Therefore, average, and unpopular) and changes in group membership over
we focused on (perceived) popularity as a predictor of both pro- time (obtain, maintain, or lose). Conceptualizing popularity as
active and reactive aggression. discrete categories provides the possibility to differentiate between
There is also empirical support for the idea that popularity is popular and unpopular adolescents and between those who main-
distinctively associated with both proactive and reactive aggres- tain, obtain, or lose their status. We propose that unpopular ado-
sion. In 2003, Prinstein and Cillessen demonstrated that popularity lescents will not gain status if they exhibit proactive aggression,
was concurrently associated with higher levels of proactive ag- and that popular adolescents will not lose status if they exhibit
gression and lower levels of reactive aggression. Yet, they found reactive aggression.
no prospective association between popularity and either proactive The reasons why the association between popularity and func-
or reactive aggression. In a recent study by Stoltz and colleagues tions of aggression might depend on adolescents’ initial status
(2016), proactive aggression was associated with higher levels of involve differences in social-emotional skills and differences in
popularity, whereas reactive aggression was associated with lower peers’ reaction to aggressive behaviors. The effective use of pro-
FUNCTIONS OF AGGRESSION AND POPULARITY 3

active aggression requires sufficient emotion regulation skills, Glick, & Smith, 2011). Yet, because of the inconsistent detection
social skillfulness, and social knowledge (Crick & Dodge, 1996; of gender differences in the associations between popularity and
Dodge, Lochman, Harnish, Bates, & Pettit, 1997; Hubbard et al., either measure of aggression (Prinstein & Cillessen, 2003; Stoltz et
2010). Popular adolescents typically have the skills necessary to al., 2016), no specific gender-related hypotheses were formulated.
strategically use proactive aggression as a means of exerting their
dominance and achieving their goals. Unpopular youth may not be Method
able to increase their status with proactive aggression as they have
insufficient skills to effectively use this strategic type of aggres-
sion. So, unpopular adolescents may not able to obtain higher Participants
status because of their inability to strategically use aggression to Data were derived from the Kandinsky Longitudinal Study, a
achieve their goals. longitudinal study designed to detect youths at risk of social and
The association between popularity and proactive and reactive emotional problems in secondary education. The study started in
aggression may also depend on adolescents’ initial status because 2010 when the head of a large secondary school requested a yearly
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

peers may respond differently to aggressive behaviors of popular assessment of students’ socioemotional well-being. For the next 6
This document is copyrighted by the American Psychological Association or one of its allied publishers.

and unpopular adolescents. Peers may accept and support aggres- years, the social and emotional functioning of all students in
sive acts from popular adolescents more than they accept the same Grades 7 through 10 (i.e., the first 4 years of secondary education
behaviors perpetrated by adolescents of lower status (Hawley & in The Netherlands) was assessed on an annual basis.
Bower, 2018). In fact, the use of proactive aggression by an For the current study, we used data collected from students
unpopular adolescent may backfire and result in a loss of status or attending 7th- through 9th-grade classrooms in Waves 1 through 7
at least contribute to the maintenance of their unpopular position. (2010 –2016).1 This resulted in a total sample of 1374 adolescents
Similarly, peers may accept (or at least not disapprove of) reactive (48.9% male) initially aged 12.67 years (SD ⫽ .42, range ⫽
aggression by popular adolescents, whereas the same behaviors are 11.09 –14.75 years) from five cohorts (n cohort2010 –2012 ⫽ 273; n
disapproved of or even punished when displayed by unpopular cohort2011–2013 ⫽ 275; n cohort2012–2014 ⫽ 282; n cohort2013–2015 ⫽
adolescents. Thus peers may react differently to the aggressive 258; n cohort2014 –2016 ⫽ 286). Most participants were born in the
behaviors exhibited by popular and unpopular adolescents by Netherlands (96.4%) to parents who were both born in the Nether-
rewarding popular adolescents for using aggression but sanction- lands (81.1%).
ing unpopular adolescents for the same behavior.
Measures
Current Study
Each year, popularity, proactive aggression, and reactive aggres-
The aim of the present study was to investigate how adolescents sion were assessed with computerized peer nominations within
who gain, maintain, or lose popularity differ in proactive and classrooms. Classmates were used as the reference group for the
reactive aggression over time. To accomplish this goal, we used a nominations because participants spent the majority of the school
large longitudinal sample of five same-aged cohorts followed day with the same group of classmates. Each nomination question
across three annual assessments. Two research questions were was presented at the top of the computer screen, followed by a
addressed. First, do popular status groups differ in their use of roster with the first names of all classmates. By clicking on names,
proactive aggression? We expected that: adolescents could nominate as many or as few classmates as they
H1: Adolescents who maintain their popular status will ex- wanted, with a minimum of one. They could not nominate them-
hibit the highest levels of proactive aggression. selves, as their own name was not presented on the personalized
screen. Same-sex and other-sex nominations were allowed (for
H2: Adolescents who obtain a popular status will be the only psychometric properties, see van den Berg & Cillessen, 2013).
group that increases in proactive aggression. Perceived popularity. Adolescents were asked to nominate
classmates “who were most popular” and “who were least popu-
H3: Adolescents who lose their popular status will be the only lar.” The number of nominations received for each question was
group that decreases in proactive aggression. counted and standardized within classrooms. A score for popular-
Second, do popular status groups differ in their use of reactive ity was computed as the difference between the standardized most
aggression? We expected that: popular and least popular scores, again standardizing the resulting
scores within classrooms (Cillessen & Marks, 2011). The popu-
H4: Adolescents who maintain their unpopular status will larity scores ranged from ⫺3.31 to 3.08 at Time 1 (T1),
exhibit the highest levels of reactive aggression. from ⫺3.32 to 2.64 at Time 2 (T2), and from ⫺3.07 to 3.02 at
Time 3 (T3).
H5: Adolescents who obtain an unpopular status will be the
Reactive aggression. Adolescents were asked, “Who feels
only group that increases in reactive aggression.
threatened or attacked easily, even though this might not have been
H6: Adolescents who lose their unpopular status will be the
only group that decreases in reactive aggression. 1
The first two waves are overlapping with the data used by Stoltz et al.
(2016). Because of this partial overlap in the samples, we performed all
Moderation by gender was explored, given the anticipated mean analyses with and without the first KLS cohort. The same pattern of
level gender differences in popularity and both forms of aggression statistically significant results emerged in the analyses with and without the
(Card, Hodges, Little, & Hawley, 2005; Card et al., 2008; Rose, first cohort, so we report the results that include data from all five cohorts.
4 VAN DEN BERG, BURK, AND CILLESSEN

intended? These classmates are not able to control their behavior procedure and at the same time during the school year (final 2
and feelings and react with aggressive behavior, like yelling or weeks of November and first week of December).
hitting” (Dodge & Coie, 1987; Stoltz et al., 2016). Nominations
received were counted and standardized within classrooms. The
Results
reactive aggression scores ranged from ⫺1.27 to 4.96 at T1,
from ⫺1.28 to 5.28 at T2, and from ⫺1.41 to 4.99 at T3.
Proactive aggression. Adolescents were asked, “Who tries to Imputation Strategy
reach their goals by using aggressive behavior? These classmates
intimidate, manipulate, or bully others to get admiration, respect or A total of 1,288 adolescents (94%) participated in the study in
objects” (Dodge & Coie, 1987; Stoltz et al., 2016). Nominations all three waves. The most common reason for nonparticipation was
received were counted and standardized within classrooms. The that the adolescent left the school. An additional 450 students did
proactive aggression scores ranged from ⫺.94 to 5.09 at T1, not complete the aggression items in 2011 because of a technical
from ⫺.81 to 5.00 at T2, and from ⫺1.45 to 4.99 at T3. error (i.e., the items were omitted from the online questionnaire for
some classrooms). We used a multiple imputation procedure to
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

account for the missing values. Specifically, we used the predictive


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

Procedure
mean matching algorithm from the MICE package (van Buuren &
Each year, the head of school formally requested the research Groothuis-Oudshoorn, 2011) in R (R Core Team, 2018) to obtain
and claimed responsibility for the parental consent procedure. The 20 imputations, each based on 20 between-imputation iterations.
school requested parental permission at the beginning of the school The imputed values were based on adolescent gender, popularity,
year for all assessments that they considered necessary for the proactive, and reactive aggression. In addition to visually inspect-
well-being of the students. The school distributed a letter to the ing the distributions of imputed values, three paired sample t tests
parents in which the purpose and procedures of the assessment was compared the observed and imputed values of popularity, proac-
described. The letter also asked parents to respond if they wanted tive, and reactive aggression. These analyses indicated that the
to exclude their child from participation. No parent objected to the imputed values did not differ from the observed values on any
participation of his or her child. Adolescents were asked to give study measure (all ps ⬎ .659).
assent at the start of each assessment. None of the participants
declined to fill out the questionnaire prior to or during the assess-
Descriptive Statistics
ment. Anonymized data were made available to the researchers for
scientific purposes. Procedures were approved by the Institutional Table 1 presents bivariate correlations among all study mea-
Review Board of the Faculty of Social Sciences of Radboud sures. The correlations were pooled across imputations using the
University (ECG2012-2505– 038; “Sociometry as a Method to micombine.cor function in the miceadds package (Robitzsch, Gr-
Measure Social Relationships Among Children and Adolescents”). und, & Henke, 2018). Gender was negatively correlated with all
Adolescents completed the computerized questionnaire on indi- measures except for popularity at Time 1 (T1), indicating that
vidual 10-in netbook computers during a 45- to 60-min classroom males were nominated as more popular and aggressive (both
session. Prior to each assessment, one of the researchers explained proactively and reactively) than females. The concurrent correla-
the goal and set up of the study. Participants were instructed that tions indicated that popularity was positively associated with pro-
the data would be processed anonymously and handled confiden- active aggression (range ⫽ .27 to .33), and negatively with reactive
tially. Adolescents were asked to keep their answers to themselves aggression (range ⫽ ⫺.12 to ⫺.14). The prospective correlations
and to be truthful in answering all questions. They were not of popularity with aggression at earlier and later assessments
allowed to talk to other students during the assessment, but they showed the same directional pattern, but popularity was more
could ask questions to the researchers and could stop participation consistently linked to proactive aggression (range ⫽ .18 to .25)
at any time. Each year, data was collected using an identical than to reactive aggression (range ⫽ ⫺.04 to ⫺.12).

Table 1
Bivariate Associations Between All Study Measures

Measure 1 2 3 4 5 6 7 8 9

1. Gender (males ⫽ 0)
2. Popularity T1 ⫺.05
3. Popularity T2 ⫺.09ⴱⴱ .80ⴱⴱ
4. Popularity T3 ⫺.10ⴱⴱ .65ⴱⴱ .73ⴱⴱ
5. Proactive aggression T1 ⫺.30ⴱⴱ .32ⴱⴱ .24ⴱⴱ .24ⴱⴱ
6. Proactive aggression T2 ⫺.30ⴱⴱ .25ⴱⴱ .33ⴱⴱ .27ⴱⴱ .40ⴱⴱ
7. Proactive aggression T3 ⫺.25ⴱⴱ .18ⴱⴱ .21ⴱⴱ .27ⴱⴱ .34ⴱⴱ .37ⴱⴱ
8. Reactive aggression T1 ⫺.25ⴱⴱ ⫺.12ⴱⴱ ⫺.08ⴱ ⫺.04 .50ⴱⴱ .23ⴱⴱ .26ⴱⴱ
9. Reactive aggression T2 ⫺.16ⴱⴱ ⫺.12ⴱⴱ ⫺.14ⴱⴱ ⫺.05 .19ⴱⴱ .42ⴱⴱ .20ⴱⴱ .31ⴱⴱ
10. Reactive aggression T3 ⫺.14ⴱⴱ ⫺.12ⴱⴱ ⫺.10ⴱⴱ ⫺.13ⴱⴱ .16ⴱⴱ .17ⴱⴱ .39ⴱⴱ .27ⴱⴱ .27ⴱⴱ
Note. N ⫽ 1,374. T1 ⫽ Time 1; T2 ⫽ Time 2; T3 ⫽ Time 3.

p ⬍ .01. ⴱⴱ p ⬍ .001.
FUNCTIONS OF AGGRESSION AND POPULARITY 5

Popularity trajectories were created by initially identifying three the interaction with time, F(12, 258.06) ⫽ 3.89, p ⬍ .001. The
popularity groups (popular, average, and unpopular) at T1 and at three-way interaction between popularity trajectory, time, and gender
T3 in each of the imputed data sets using ⫾ 1 SD to differentiate was not statistically significant, F(12, 405.2) ⫽ 1.54, p ⫽ .109. Figure
groups. In each imputed dataset, seven popularity trajectories were 1 presents the estimated means and 95% confidence intervals (CIs) of
created using the T1 and T3 groups. Three trajectories described each popularity trajectory separately for the three measures of proac-
adolescents with stable levels of popularity. Stable popular (n ⫽ tive aggression. Between-groups comparisons tested whether adoles-
94, range ⫽ 91–97) included adolescents who were popular at T1 cents in the stable popular group were more proactively aggressive
and T3; stable average (n ⫽ 770, range ⫽ 767–775) included than those in the other groups (i.e., Hypothesis 1) at T1 and T3. At T1,
adolescents who were average at T1 and T3; and stable unpopular the stable popular group had higher levels of proactive aggression
(n ⫽ 100, range ⫽ 97–102) included adolescents who were un- than all other groups (all ps ⬍ .05) except for the lose popularity
popular at T1 and T3. Four trajectories described adolescents with group. At T3, the stable popular group had higher levels of proactive
popularity scores that changed from T1 to T3. Lose popularity aggression than all other trajectories (all ps ⬍ .001) except for the
(n ⫽ 112, range ⫽ 109 –117) included adolescents who were obtain popularity group (see Figure 1). Adolescents in the stable
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

popular at T1 but not popular by T3; obtain popularity (n ⫽ 112, unpopular and lose popularity groups also differed from the other
This document is copyrighted by the American Psychological Association or one of its allied publishers.

range ⫽ 109 –115) included adolescents who were not popular at groups in proactive aggression at T1, and adolescents in the obtain
T1 and became popular by T3; lose unpopularity (n ⫽ 76, range ⫽ popularity group differed from all other groups at T3.
71–78) included adolescents who were unpopular at T1 but not Within-group comparisons tested whether adolescents who ob-
unpopular by T3; and obtain unpopularity (n ⫽ 110, range ⫽ tained popular status increased in proactive aggression from T1 to
106 –114) included adolescents who were not unpopular at T1 and T3 (Hypothesis 2) and those who lost popular status decreased in
became unpopular by T3. proactive aggression over time (Hypothesis 3). These comparisons
Chi-square analyses examined whether the composition of the indicated that the adolescents in the obtain popularity group in-
popularity trajectories differed as a function of gender. The chi- creased in proactive aggression across all three assessments (ps ⬍
square values for each imputed dataset were pooled using the .001) and those in the lose popularity group decreased in proactive
micombine.chisquare function in the miceadds package (Robitzsch aggression across all three assessments (ps ⬍ .001). In addition,
et al., 2018). This analysis indicated that the composition of the adolescents in the lose unpopularity group increased in proactive
popularity trajectories differed for males and females, F(6, aggression, and adolescents in the obtain unpopularity group de-
4016.16) ⫽ 12.91, p ⬍ .001. Cell-wise tests indicated that five of creased in proactive aggression from T1 to T3 (ps ⬍ .001). Thus,
the seven trajectories significantly differed in gender composition: the omnibus tests indicated that the popularity trajectories differed
Males were overrepresented in the stable popular (60%), obtain in proactive aggression over time and these differences were not
popularity (60%), and lose unpopularity (61%) groups. Females further moderated by gender. Follow-up analyses indicated that (a)
were overrepresented in the obtain unpopularity (62%) and stable adolescents who maintained their popular status were perceived as
average (55%) groups. the most proactively aggressive, (b) adolescents who obtained
popular status and those who lost unpopular status increased their
Popularity Trajectories Predicting proactive aggression, and (c) adolescents who lost popular status
and those who obtained unpopular status decreased their proactive
Changes in Aggression
aggression.
Two linear mixed models were performed using the lmer func- For reactive aggression the statistically significant main effect
tion with the package lme4 (Bates, Mächler, Bolker, & Walker, of popularity trajectories, F(6, 33.9) ⫽ 14.25, p ⬍ .001, was
2015) to examine changes in proactive and reactive aggression as qualified by the interaction with time, F(12, 7151.75) ⫽ 3.09, p ⬍
a function of popularity trajectories. In both models, popularity .001. The three-way interaction between popularity trajectory,
trajectories, gender, and time (and all interactions between these time, and gender was not statistically significant, F(12, 330.75) ⫽
three predictors) were included as fixed effects. Reactive aggres- 0.92, p ⫽ .523. Figure 2 presents the estimated means and 95% CIs
sion was included as a time-varying covariate in the model pre- of each popularity trajectory separately for the three measures of
dicting proactive aggression, and proactive aggression was in- reactive aggression. Between-groups comparisons tested whether
cluded as a time-varying covariate in the model predicting reactive adolescents in the stable unpopular group were more reactively
aggression. Individual differences in aggression were modeled aggressive than those in the other groups (i.e., Hypothesis 4) at T1
with a random intercept and cohort differences in the associations and T3. At T1, the stable unpopular group had higher levels of
between popularity trajectories and aggression were modeled with reactive aggression than all other groups (all ps ⬍ .001). At T3, the
a random slope. Type 3 conditional F tests were computed with the stable unpopular group had higher levels of reactive aggression
Kenward-Roger approximation for degrees of freedom. The mi- than all other trajectories (all ps ⬍ .001) except for the obtain
combine.F function in the miceadds package (Robitzsch et al., unpopularity trajectory (see Figure 2). Adolescents in the lose
2018) was used to pool the omnibus F tests. Estimated means, unpopularity group also differed from all other groups in reactive
standard errors, and pairwise comparisons (with Tukey adjust- aggression at T1, and adolescents in the obtain unpopularity group
ment) were obtained with the emmeans and pairs functions in the also differed from all other groups at T3.
emmeans package using the recover_data.mire and emm_basis- Within-group comparisons tested whether adolescents who ob-
.mira extensions developed for pooling results from multiply im- tained unpopular status increased in reactive aggression from T1 to
puted mixed effect models (Lenth, 2018). T3 (Hypothesis 5) and those who lost unpopular status decreased
For proactive aggression, the statistically significant main effect of in reactive aggression over time (Hypothesis 6). Adolescents in the
popularity trajectories, F(6, 64.69) ⫽ 9.98, p ⬍ .001, was qualified by obtain unpopularity group increased in reactive aggression from
6 VAN DEN BERG, BURK, AND CILLESSEN

1
Proactive aggression (z-score)

0.5

Time 1
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Time 2
This document is copyrighted by the American Psychological Association or one of its allied publishers.

-0.5
Time 3

-1

Popularity trajectories

Figure 1. Estimated means and 95% confidence intervals for proactive aggression of the seven popularity
trajectories.

T1 to T3 and from T2 to T3 (ps ⬍ .001) but did not increase in elementary to secondary education. This is a time when adoles-
reactive aggression from T1 to T2 (p ⫽ .256). Adolescents in the cents form new groups with previously unacquainted peers, and
lose unpopularity group decreased in reactive aggression across all when questions of gaining, maintaining, and losing status are of
three assessments (ps ⬍ .001). In addition, adolescents in the primary importance (LaFontana & Cillessen, 2010). Our results
stable popularity group also decreased in reactive aggression from replicated previous work in which popular adolescents exhibited
T1 to T3. Thus, the omnibus tests indicated that the popularity the highest levels of proactive aggression, and unpopular adoles-
trajectories differed in reactive aggression over time and these cents exhibited the highest levels of reactive aggression. Our
differences were not further moderated by gender. Follow-up results also extend previous work by demonstrating that prospec-
analyses indicated that (a) adolescents who maintained their un- tive associations between popularity and reactive, but not proactive
popular status were perceived as the most reactively aggressive, aggression, varied as a function of adolescents’ initial status.
(b) adolescents who obtained unpopular status increased in reac- Specifically, adolescents in groups characterized by increases in
tive aggression over time, and (c) adolescents who lost their status (obtain popularity and lose unpopularity) increased in pro-
unpopular status decreased in reactive aggression. active aggression and those in groups characterized by decreases in
status (lose popularity and obtain unpopularity) decreased in pro-
Discussion active aggression. In addition only adolescents who obtained un-
popularity increased in reactive aggression, and those who lost
In this study, we examined how adolescents who gain, maintain their unpopular status decreased in reactive aggression.
or lose popular or unpopular status differed in proactive and
reactive aggression over time. We proposed that prospective as-
Popularity and the Distinct Functions of Aggression
sociations between popularity and the functions of aggression
would depend on status already achieved. To test this proposition, This study builds on two earlier studies on proactive and reac-
we created groups of adolescents who maintained, gained, or lost tive aggression and popularity in adolescence (Prinstein & Cil-
popular and unpopular status across three school years and exam- lessen, 2003; Stoltz et al., 2016). We replicated that high levels of
ined group differences in proactive and reactive aggression over proactive aggression are generally associated with high status and
time. Popularity and aggression were assessed yearly from early to that high levels of reactive aggression are generally associated
middle adolescence, immediately following the transition from with low status in a larger sample and over a longer time period.
FUNCTIONS OF AGGRESSION AND POPULARITY 7

1
Reactive aggression (z-score)

0.5

Time 1
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Time 2
This document is copyrighted by the American Psychological Association or one of its allied publishers.

-0.5
Time 3

-1

Popularity trajectories

Figure 2. Estimated means and 95% confidence intervals for reactive aggression of the seven popularity
trajectories.

In addition, popular youth who maintained their high status dis- who lost their unpopular status increased in proactive aggression;
played the highest levels of proactive aggression across time. those who became unpopular and those who lost their popular
Unpopular youth who retained their low status over time exhibited status decreased in proactive aggression. In other words, changes
the highest levels of reactive aggression. These results illustrate the in popular and unpopular status were associated with changes in
roots of both functions of aggression in social– cognitive learning proactive aggression. Taken together, these findings suggest that
theory and the frustration-aggression model (Berkowitz, 1993; proactive aggression has universal benefits for popularity, whereas
Crick & Dodge, 1996; Dodge et al., 1997; Hubbard et al., 2010). reactive aggression has differential costs.
Proactive aggression requires social skillfulness and emotion reg- The results of this study are in line with the evolutionary
ulation to effectively assert dominance and control. When used perspective on the function of proactive aggression (Ellis et al.,
effectively, strategic use of aggression is rewarded in terms of 2012; Hawley & Bower, 2018). According to this viewpoint, the
status maintenance, thus further reinforcing the use of proactive effective, strategic, and proactive use of aggression provides one
aggression over time. In contrast, when adolescents respond ag- way in which adolescents can exert social dominance to gain and
gressively due to a lack of social skills or emotion regulation, peers control resources. Although we had expected that only popular
will respond in disapproving and negative ways underscoring adolescents would have the knowledge and skills necessary to
adolescents’ low status. effectively use proactive aggression, our findings also indicated
The current study extends our understanding of the prospective that adolescents who were initially unpopular and subsequently
associations between popularity and the functions of aggression by gained status (i.e., lose unpopularity group) also increasingly used
showing that changes in reactive aggression depended on adoles- proactive aggression over time. Thus, gains in status were associ-
cents’ initial status, and changes in proactive aggression did not. ated with increasing levels of proactive aggression, irrespective of
That is, adolescents who lost their unpopular status were the only ones already achieved status in the peer group.
group that decreased in reactive aggression over time. Adolescents Although our hypotheses regarding the differential benefits of
who became unpopular were the only group that increased in proactive aggression were partially unexpected, our results did
reactive aggression. Thus, adolescents who obtained or lost un- indicate that the use of reactive aggression has differential costs.
popular status changed in reactive aggression over time. Obtaining That is, changes in unpopular status (and not popular status) were
or losing popular status was not related to changes in reactive related to changes in reactive aggression. This suggests that ado-
aggression. A different pattern of group differences emerged for lescents of higher status do not face the same social sanctions for
proactive aggression. Adolescents who became popular and those using aggression in a reactive manner as those of lower status. This
8 VAN DEN BERG, BURK, AND CILLESSEN

is the first study to demonstrate that initial status shapes the frequency and severity of the aggressive acts. This distinction has
functional use of aggression. However, there is also evidence that been successfully used in research investigating the functions of
suggests that the use of some forms of aggression may not have the aggression for bullying behavior (see, e.g., van Noorden, Bu-
same consequences for all youth. Ettekal and Ladd (2015) exam- kowski, Haselager, Lansu, & Cillessen, 2016) and may prove to be
ined the costs and benefits of various forms of aggression for equally helpful in further understanding the links between the
children’s peer acceptance, rejection, and friendships. Using a functions of aggression and popularity. Third, we categorized
person-oriented approach (i.e., identifying aggression trajectories) adolescents into popularity trajectories and examined group dif-
they found that children with moderate levels of aggression (es- ferences in aggression over time. We created the categories based
pecially relational aggression) became better liked over time and on a commonly used statistical criterion (i.e., ⫾1 SD), yet there
were nominated more often as friends. Surprisingly, they also was an inevitable loss of information in the conversion of contin-
found that children who abstained from any form of aggression uous to categorical scores. In addition, we did not use statistical
had significant decreases in peer acceptance and friendships over procedures to identify co-occurring trajectories of status and ag-
time. The present study, as well as the one by Ettekal and Ladd gression such as latent growth curve mixture modeling, instead
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(2015), demonstrate how person-oriented and variable-oriented electing to form the trajectories directly from the popularity cate-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

approaches can be used in a complementary manner to gain a gories. This strategy allowed us to directly test our hypotheses, it
better understanding of the complex associations between social provided meaningfully sized groups, and groups characterized by
status and aggression. specific types of changes. Yet, it is unclear to what extent alter-
native strategies would have produced comparable results. Fourth,
the adolescents in our sample were quite homogeneous in age and
The Role of Gender
ethnic background and all attended a single school in the Nether-
With regard to gender differences, we found that males were lands. It remains an open question whether our findings generalize
nominated as more popular, as well as more proactively and to more diverse samples and age groups.
reactively aggressive than females. Furthermore, males were over- Despite these limitations, this study provides several promising
represented in popularity trajectories characterized by gains in and viable directions for future research. First, although we fo-
status (stable popular, obtain popularity, lose unpopularity) and cused on the functions of aggression, investigating the combina-
females were overrepresented in trajectories characterized by the tion of forms and functions is a logical next step in understanding
maintenance of average status (stable average) and loss of status the complex dynamics of status and aggression. In a study with
(obtain unpopularity). However, gender did not further moderate preschool children, Ostrov and colleagues (2013) found that pro-
the associations between popularity groups and changes in either active relational aggression was uniquely associated with de-
of the aggression measures over time. That is, the popularity group creases in peer rejection, whereas reactive relational aggression
differences in proactive and reactive aggression reported in this was associated with increases in peer rejection over time. Prinstein
study did not differ for adolescent males and females. and Cillessen (2003) found that proactive overt aggression was
Although the mean-level gender differences in popularity and concurrently associated with popularity, whereas reactive overt
both functions of aggression are not surprising (Card et al., 2005, aggression was associated with unpopularity. Moreover, reactive
2008; Rose et al., 2011), the lack of moderating effects of gender aggression was differently associated with popularity depending
on the popularity-aggression link indicates that the functions of on its form. Thus, combining forms and functions of aggression
aggression for gaining, maintaining, and losing social status are will reveal further specific associations with popularity as well as
not gender-specific. This is in line with previous research arguing other social behaviors.
that the underlying motives and processes regarding status acqui- Another direction for future research is to examine the under-
sition and maintenance apply equally to males and females (Haw- lying processes that explain how and why changes in popularity
ley, 1999). correspond to changes in aggression. For instance, we do not know
whether adolescents who scored high on proactive aggression were
the ones who actually strived for popularity and thought they could
Limitations and Future Directions
achieve status by means of aggression. In line with the definition
This study contributed to research on the popularity-aggression of proactive aggression (Card & Little, 2006) and the evolutionary
link (cf. Mayeux, Houser, & Dyches, 2011) and to the relatively perspective on aggression (Ellis et al., 2012; Hawley & Bower,
new focus on the function of aggression for youth’s social status 2018), we asked adolescents “who uses proactive aggression to
(Prinstein & Cillessen, 2003; Stoltz et al., 2016). Despite the achieve or maintain a certain status.” Yet, we did not consider their
strengths of this study, several caveats deserve mentioning. own social goals, the importance they place on obtaining or main-
First, we used peer nominations to measure both functions of taining popularity, or the ways in which they thought they could
aggression and popularity. Shared method variance therefore may acquire or protect their own status. In future studies, researchers
have inflated the magnitude of associations. Measuring popularity may include participants’ own motives for displaying certain be-
and aggression by questioning those who frequently interact with haviors and whether they actually aspire to become (more) popular
each other and are insiders in the peer culture has high face validity (Dawes & Xie, 2014; Ojanen & Findley-Van Nostrand, 2014; van
and is considered a reliable and valid technique (Cillessen & den Broek, Deutz, Schoneveld, Burk, & Cillessen, 2016). By doing
Marks, 2017; Stoltz et al., 2016). Still, future studies may also so, we can empirically test whether adolescents who strive for
consider alternative methods such as observational or self-report popularity and think they can acquire status by means of aggres-
instruments. A second and related point is that our measures of sion, are the ones who most frequently use proactive aggression.
proactive and reactive aggression did not distinguish between the Similarly, reactive aggression was defined and operationalized as
FUNCTIONS OF AGGRESSION AND POPULARITY 9

an angry or frustrated reaction to an actual or perceived offense Card, N. A., Stucky, B. D., Sawalani, G. M., & Little, T. D. (2008). Direct
(Crick & Dodge, 1996). Youth who display such reactions often and indirect aggression during childhood and adolescence: A meta-
have problems in regulating their emotions or have hostile biases analytic review of gender differences, intercorrelations, and relations to
when interpreting social information (Card & Little, 2006; Hub- maladjustment. Child Development, 79, 1185–1229. http://dx.doi.org/10
bard et al., 2010). Assessing these and other sociocognitive func- .1111/j.1467-8624.2008.01184.x
Cillessen, A. H. N. (2011). Toward a theory of popularity. In A. H. N.
tions would allow explicit testing of whether adolescents with such
Cillessen, D. Schwartz, & L. Mayeux (Eds.), Popularity in the peer
deficits or biases are the ones who retain low status or are at risk system (pp. 273–299). New York, NY: Guilford Press.
of obtaining unpopular status. Cillessen, A. H. N., & Marks, P. E. L. (2011). Conceptualizing and
Finally, previous studies have argued that the functionality of measuring popularity. In A. H. N. Cillessen, D. Schwartz, & L. Mayeux
aggression is context specific (Ellis et al., 2012). In this study, (Eds.), Popularity in the peer system (pp. 25–56). New York, NY:
aggression and popularity were assessed among adolescents within Guilford Press.
the classroom context. This is a salient context for studying these Cillessen, A. H. N., & Marks, P. E. L. (2017). Methodological choices in
constructs, because it is a context in which adolescents spend much peer nomination research. New Directions for Child and Adolescent
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

of their time. If proactive aggression is beneficial for obtaining Development. Advance online publication. http://dx.doi.org/10.1002/cad
This document is copyrighted by the American Psychological Association or one of its allied publishers.

status in the classroom, as our results suggest, this does not .20206
necessarily mean that proactive aggression is also beneficial or Cillessen, A. H. N., Mayeux, L., Ha, T., de Bruyn, E. H., & LaFontana,
K. M. (2014). Aggressive effects of prioritizing popularity in early
functional outside of the classroom. The functions of aggression
adolescence. Aggressive Behavior, 40, 204 –213. http://dx.doi.org/10
may also be specific to this developmental period. The relatively
.1002/ab.21518
low levels of reactive aggression exhibited by popular adolescents Coie, J. D., Dodge, K. A., & Coppotelli, H. (1982). Dimensions and types
and the relatively low levels of proactive aggression displayed by of social status: A cross-age perspective. Developmental Psychology, 18,
unpopular adolescents seems to indicate that links between status 557–570. http://dx.doi.org/10.1037/0012-1649.18.4.557
and aggression have been established by early adolescence. A Crick, N. R., & Dodge, K. A. (1996). Social information-processing
further understanding of whether or when these patterns emerge mechanisms in reactive and proactive aggression. Child Development,
and potentially disappear is needed. 67, 993–1002. http://dx.doi.org/10.2307/1131875
Dawes, M., & Xie, H. (2014). The role of popularity goal in early
adolescents’ behaviors and popularity status. Developmental Psychol-
Conclusion ogy, 50, 489 – 497. http://dx.doi.org/10.1037/a0032999
Dodge, K. A., & Coie, J. D. (1987). Social-information-processing factors
This study replicated and extended previous work on the asso-
in reactive and proactive aggression in children’s peer groups. Journal of
ciations between peer status and functions of aggression. Associ- Personality and Social Psychology, 53, 1146 –1158. http://dx.doi.org/10
ations between popularity and both measures of aggression were .1037/0022-3514.53.6.1146
robust, concurrently within each year, and also when examining Dodge, K. A., Lochman, J. E., Harnish, J. D., Bates, J. E., & Pettit, G. S.
changes in popularity and changes in proactive and reactive ag- (1997). Reactive and proactive aggression in school children and psy-
gression over time. Our findings provide a more nuanced under- chiatrically impaired chronically assaultive youth. Journal of Abnormal
standing of these concurrent and prospective links, by revealing Psychology, 106, 37–51. http://dx.doi.org/10.1037/0021-843X.106.1.37
that popular adolescents maintain their high status and less popular Dollard, J., Doob, N., Miller, N. E., Mowrer, O. H., & Sears, R. R. (1939).
adolescents gain status through the use of proactive aggression Frustration and aggression. New Haven, CT: Yale University Press.
(universal benefits); whereas only less popular adolescents lose http://dx.doi.org/10.1037/10022-000
status through the use of reactive aggression (differential costs). Ellis, B. J., Del Giudice, M., Dishion, T. J., Figueredo, A. J., Gray, P.,
Griskevicius, V., . . . Wilson, D. S. (2012). The evolutionary basis of
These findings have the potential to move research on the associ-
risky adolescent behavior: Implications for science, policy, and practice.
ations between popularity and aggression in a new direction: while
Developmental Psychology, 48, 598 – 623. http://dx.doi.org/10.1037/
certain behaviors may be beneficial for some, the same behaviors a0026220
may come at a cost for others. Ettekal, I., & Ladd, G. W. (2015). Costs and benefits of children’s physical
and relational aggression trajectories on peer rejection, acceptance, and
friendships: Variations by aggression subtypes, gender, and age. Devel-
References opmental Psychology, 51, 1756 –1770. http://dx.doi.org/10.1037/
Bandura, A. (1973). Aggression: A social learning analysis. Englewood dev0000057
Cliffs, NJ: Prentice Hall. Fite, P. J., Colder, C. R., Lochman, J. E., & Wells, K. C. (2008). Devel-
Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear opmental trajectories of proactive and reactive aggression from fifth to
mixed-effects models using lme4. Journal of Statistical Software, 67, ninth grade. Journal of Clinical Child and Adolescent Psychology, 37,
1– 48. http://dx.doi.org/10.18637/jss.v067.i01 412– 421. http://dx.doi.org/10.1080/15374410801955920
Berkowitz, L. (1993). Aggression: Its causes, consequences, and control. Hawley, P. H. (1999). The ontogenesis of social dominance: A strategy-
New York, NY: McGraw-Hill. based evolutionary perspective. Developmental Review, 19, 97–132.
Card, N. A., Hodges, E. V. E., Little, T. D., & Hawley, P. H. (2005). http://dx.doi.org/10.1006/drev.1998.0470
Gender effects in peer nominations for aggression and social status. Hawley, P. H., & Bower, A. R. (2018). Evolution and peer relations:
International Journal of Behavioral Development, 29, 146 –155. http:// Considering the functional roles of aggression and prosociality. In W. M.
dx.doi.org/10.1080/01650250444000414 Bukowski, B. Laursen, & K. H. Rubin (Eds.), Handbook of peer inter-
Card, N. A., & Little, T. D. (2006). Proactive and reactive aggression in actions, relationships, and groups (2nd ed., pp. 106 –122). New York,
childhood and adolescence: A meta-analysis of differential relations NY: Guilford Press.
with psychosocial adjustment. International Journal of Behavioral De- Hubbard, J. A., McAuliffe, M. D., Morrow, M. T., & Romano, L. J. (2010).
velopment, 30, 466 – 480. http://dx.doi.org/10.1177/0165025406071904 Reactive and proactive aggression in childhood and adolescence: Pre-
10 VAN DEN BERG, BURK, AND CILLESSEN

cursors, outcomes, processes, experiences, and measurement. Journal of Poulin, F., & Boivin, M. (2000). Reactive and proactive aggression:
Personality, 78, 95–118. http://dx.doi.org/10.1111/j.1467-6494.2009 Evidence of a two-factor model. Psychological Assessment, 12, 115–
.00610.x 122. http://dx.doi.org/10.1037/1040-3590.12.2.115
LaFontana, K. M., & Cillessen, A. H. N. (1998). The nature of children’s Prinstein, M. J., & Cillessen, A. H. N. (2003). Forms and functions of
stereotypes of popularity. Social Development, 7, 301–320. http://dx.doi adolescent peer aggression associated with high levels of peer status.
.org/10.1111/1467-9507.00069 Merrill-Palmer Quarterly, 49, 310 –342. http://dx.doi.org/10.1353/mpq
LaFontana, K. M., & Cillessen, A. H. N. (2010). Developmental changes .2003.0015
in the priority of perceived status in childhood and adolescence. Social R Core Team. (2018). R: A language and environment for statistical
Development, 19, 130 –147. http://dx.doi.org/10.1111/j.1467-9507.2008 computing. Vienna, Austria: R Foundation for Statistical Computing.
.00522.x Retrieved from https://www.R-project.org/
Lenth, R. (2018). Emmeans: Estimated marginal means, aka least-squares Robitzsch, A., Grund, S., & Henke, T. (2018). miceadds: Some additional
means (R package Version 1.2.1) [Computer software]. Retrieved from multiple imputation functions, especially for mice (R package Version
https://CRAN.R-project.org/package⫽emmeans 3.0 –16) [Computer software]. Retrieved from https://CRAN.R-project
Mayeux, L., Houser, J. J., & Dyches, K. D. (2011). Social acceptance and .org/package⫽miceadds
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

popularity: Two distinct forms of peer status. In A. H. N. Cillessen, D. Rose, A. J., Glick, G. C., & Smith, R. L. (2011). Popularity and gender:
Two cultures of boys and girls. In A. H. N. Cillessen, D. Schwartz, & L.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Schwartz, & L. Mayeux (Eds.), Popularity in the peer system (pp.


79 –102). New York, NY: Guilford Press. Mayeux (Eds.), Popularity in the peer system (pp. 103–122). New York,
NY: Guilford Press.
Murray-Close, D., & Ostrov, J. M. (2009). A longitudinal study of forms
Stoltz, S., Cillessen, A. H. N., van den Berg, Y. H. M., & Gommans, R.
and functions of aggressive behavior in early childhood. Child Devel-
(2016). Popularity differentially predicts reactive and proactive aggres-
opment, 80, 828 – 842. http://dx.doi.org/10.1111/j.1467-8624.2009
sion in early adolescence. Aggressive Behavior, 42, 29 – 40. http://dx.doi
.01300.x
.org/10.1002/ab.21603
Ojanen, T., & Findley-Van Nostrand, D. (2014). Social goals, aggression,
van Buuren, S., & Groothuis-Oudshoorn, K. (2011). mice: Multivariate
peer preference, and popularity: Longitudinal links during middle
imputation by chained equations in R. Journal of Statistical Software,
school. Developmental Psychology, 50, 2134 –2143. http://dx.doi.org/10
45, 1– 67. http://dx.doi.org/10.18637/jss.v045.i03
.1037/a0037137
van den Berg, Y. H. M., & Cillessen, A. H. N. (2013). Computerized
Ojanen, T., & Kiefer, S. (2013). Instrumental and reactive functions and sociometric and peer assessment: An empirical and practical evaluation.
overt and relational forms of aggression: Developmental trajectories and International Journal of Behavioral Development, 37, 68 –76. http://dx
prospective associations during middle school. International Journal of .doi.org/10.1177/0165025412463508
Behavioral Development, 37, 514 –517. http://dx.doi.org/10.1177/ van den Broek, N., Deutz, M. H. F., Schoneveld, E. A., Burk, W. J., &
0165025413503423 Cillessen, A. H. N. (2016). Behavioral correlates of prioritizing popu-
Ostrov, J. M., Murray-Close, D., Godleski, S. A., & Hart, E. J. (2013). larity in adolescence. Journal of Youth and Adolescence, 45, 2444 –
Prospective associations between forms and functions of aggression and 2454. http://dx.doi.org/10.1007/s10964-015-0352-7
social and affective processes during early childhood. Journal of Exper- van Noorden, T. H. J., Bukowski, W. M., Haselager, G. J. T., Lansu,
imental Child Psychology, 116, 19 –36. http://dx.doi.org/10.1016/j.jecp T. A. M., & Cillessen, A. H. N. (2016). Disentangling the frequency and
.2012.12.009 severity of bullying and victimization in the association with empathy.
Parkhurst, J. T., & Hopmeyer, A. (1998). Sociometric popularity and Social Development, 25, 176 –192. http://dx.doi.org/10.1111/sode.12133
peer-perceived popularity: Two distinct dimensions of peer status. The Vitaro, F., Brendgen, M., & Barker, E. D. (2006). Subtypes of aggressive
Journal of Early Adolescence, 18, 125–144. http://dx.doi.org/10.1177/ behaviors: A developmental perspective. International Journal of Be-
0272431698018002001 havioral Development, 30, 12–19. http://dx.doi.org/10.1177/016
Polman, H., Orobio de Castro, B., Koops, W., van Boxtel, H. W., & Merk, 5025406059968
W. W. (2007). A meta-analysis of the distinction between reactive and
proactive aggression in children and adolescents. Journal of Abnormal Received October 23, 2018
Child Psychology, 35, 522–535. http://dx.doi.org/10.1007/s10802-007- Revision received June 14, 2019
9109-4 Accepted June 26, 2019 䡲

You might also like