DRD4 Genotype and the Developmental Link of Peer Social Preference with Conduct Problems and Prosocial Behavior Across Ages 9–12 Years
DRD4 Genotype and the Developmental Link of Peer Social Preference with Conduct Problems and Prosocial Behavior Across Ages 9-12 Years
J. Marieke Buil 0 1 2
Hans M. Koot 0 1 2
Tjeert Olthof 0 1 2
Kelly A. Nelson 0 1 2
Pol A. C. van Lier 0 1 2
0 EMGO Institute for Health and Care Research , Amsterdam , The Netherlands
1 Department of Developmental Psychology, VU University Amsterdam , Van der Boechorststraat 1, 1081 BT Amsterdam , The Netherlands
2 Avera Institute for Human Genetics , 3720 W. 69th Street, Suite 200, Sioux Falls, SD , USA
The peer environment is among the most important factors for children's behavioral development. However, not all children are equally influenced by their peers, which is potentially due to their genetic make-up. The dopamine receptor D4 gene (DRD4) is a potential candidate gene that may influence children's susceptibility to the peer environment. In the present study, we explored whether variations in the DRD4 gene moderated the association between children's social standing in the peer group (i.e., social preference among classmates) with subsequent conduct problems and prosocial behavior among 405 (51 % females) elementary school children followed annually throughout early adolescence (ages 9-12 years). The behavioral development of children with and without the DRD4 7-repeat allele was compared. The results indicated that children who had higher positive social preference scores (i.e., who were more liked relative to disliked by their peers) showed less conduct problem development in subsequent years relative to children who had lower positive social preference scores. In contrast, children who had more negative preference scores (i.e., who were more disliked relative to liked among peers) showed more conduct problem development in subsequent years, relative to children who had less negative preference scores. However, these effects only occurred when children had a 7-repeat allele. For children who did not have a 7-repeat allele, the level of social preference was not associated with subsequent conduct problems. No evidence for gene-environment interaction effects for prosocial behavior was found. The implications for our understanding of conduct problem development and its prevention are discussed.
Gene-environment interaction; DRD4; Peer social preference; Conduct problems; Prosocial behavior; Differential susceptibility
In school, children have to function in a classroom for a
significant amount of time every day, across the better part
of their childhood and later adolescent years. As in every
social setting, children evaluate classmates and form
opinion on who they do and do not like. As a consequence
of this evaluation, some children will become highly
preferred and liked among many of their peers. These highly
preferred children have been found to develop high-quality
friendships (Parker and Asher 1993), have positive
relationships with teachers (Hughes et al. 2006), and generally
show favorable developmental outcomes such as prosocial
behavior (Bierman and Erath 2006). However, the dark
side of the peer evaluation process is that some children
become disliked and poorly preferred by their classmates,
which is a robust predictor of maladjustment. For instance,
these children are at risk of peer victimization and
friendlessness (Van Lier and Koot 2010) and poor support
or rejection by teachers (Leflot et al. 2011). As such, it may
come as no surprise that children who are poorly preferred
by their peers are at risk of developing behavioral problems
(Ladd 2006; Van Lier and Koot 2010).
Thus, there is a vast body of research linking childrens
social standing among peers, also known as peer social
preference (Coie et al. 1982), to childhood adjustment and
maladjustment. However, individual differences in the
predictive links are striking. Recent findings have
suggested that the genetic make-up of children may be of
relevance in understanding why children are more or less
affected by their social environment (for a meta-analysis,
see Bakermans-Kranenburg and van IJzendoorn 2011).
That is, several studies have indicated that the dopamine
receptor D4 gene (DRD4) may render children susceptible
to environmental influences for better and for worse
(Bakermans-Kranenburg and van IJzendoorn 2011, p. 39).
According to this viewpoint, carriers of the 7-repeat allele
(DRD4-7r) may be disproportionally susceptible for
developing negative behavioral outcomes in an adverse
environment, but are also more likely to respond with
positive behavioral outcomes when in a favorable
environment (Belsky and Hartman 2014). In the present study,
we aimed to investigate the possible moderating role of
DRD4 in the prospective association between low and high
levels of peer social preference and the development of
conduct problems and prosocial behavior, among children
attending elementary school who were followed annually
from age 9 to 12 years.
According to the differential susceptibility hypothesis
(Belsky 1997; Belsky and Hartman 2014), some genetic
variants may render individuals more malleable to negative
as well as positive environments with respect to subsequent
development, while other individualsdepending on their
genetic make-upare altogether less influenced by their
environment. In a nutshell, this viewpoint proposes that, in
order to increase reproductive fitness it makes evolutionary
sense that some children are more susceptible to their
environment than others (Belsky 1997; Belsky and Hartman
2014). That is, parents may (subconsciously or
consciously) aim to modify childrens behavior so that it
matches the environmental requirements. If the future
environment is predicted correctly, a beneficial
behavior-environment match occurs that may support the offsprings
health and reproductive fitness. However, given that future
environmental circumstances are uncertain, for some
children a mismatch occurs, potentially resulting in adverse
outcomes. Thus, if within a family some children are born
with a genetic disposition that renders them highly
susceptible to their environment and others have a genetic
disposition that renders them less susceptible, the
probability that for all offspring such a detrimental mismatch
takes place decreases (example adapted from Belsky 1997).
A potential candidate gene that may further our
understanding of individual differences in sensitivity to the
environment is the dopamine receptor D4 gene, DRD4
(Bakermans-Kranenburg and van IJzendoorn 2011; Belsky
and Hartman 2014). DRD4 regulates dopamine receptor
activity in the brain, particularly in brain regions of the
mesocorticolimbic dopamine pathway (Oak et al. 2000).
The neurotransmitter dopamine plays a major role in
reward, punishment, attention and motivation mechanisms
related to social interaction and learning. Furthermore,
dopamine may signal the salience of social events and is a
key factor in the imprinting of motivational importance to
environmental factors (Trainor 2011).
The coding DNA sequence of DRD4 is highly
polymorphic, resulting in receptor variants that may be
functionally different. In this regard, the 48-bp tandem repeat
(48-bp VNTR) in the third exon, consisting of 211
repeats, has received much research attention in behavior
genetics. It has been shown that DRD4 has higher potency
for dopamine-mediated coupling to adenylyl cyclase in the
presence of the short 2-repeat and 4-repeat alleles, than
when receptors are encoded by the 7-repeat allele, known
as DRD4-7r (Oak et al. 2000; Schoots and Van Tol 2003).
Decreased postsynaptic inhibition due to the 7-repeat allele
results in lower dopaminergic tone and a suboptimal
response to dopamine. This is associated with heightened
reward-related reactivity in the ventral striatum and
reward-related behaviors like impulsivity (Forbes et al.
2009). In addition, the mesocorticolimbic dopamine
pathway is associated with the functioning of the anterior
cingulate cortex, which is related to processing punishment
and reward stimuli. Changes in dopamine levels due to the
DRD4 polymorphism could thus enhance
social-environmental signals related to reward and punishment (Posner
and Rothbart 2009). Indeed, subjects with the 7-repeat
allele show increased reactivity to social-environmental
stimuli compared to subjects without this allele, as
evidenced by findings from brain imaging, observational and
experimental studies in humans and animals (Grady et al.
2013; Sheese et al. 2007). When confronted with emotional
stimuli, carriers of the DRD4-7r allele were found to show
more brain activity than non-carriers in brain regions
associated with attention to and appraisal of negative
emotional stimuli, as well as in brain regions involved in
preparation for action (Gehricke et al. 2015). To the best of
our knowledge, as of yet no studies have used functional
brain imaging to investigate whether brain regions that are
involved in reactivity and attention with regard to negative
stimuli also apply to positive stimuli. However,
observational research has indicated that individuals with the
7-repeat allele show heightened sensitivity to positive
parenting environments when compared to individuals
without this allele (Bakermans-Kranenburg and van
IJzendoorn 2011). Together, these findings may suggest
that individuals with a 7-repeat allele of the DRD4 gene are
more susceptible to their environment than individuals
without this allele, irrespective of whether this environment
is positive or negative. Furthermore, some authors
suggested that the dopaminergic system is key to the
development of social behavior (Insel 2003). This statement is
supported by the fact that on a behavioral level DRD4-7r
has been related to aggression in children (Schmidt et al.
2002), to conduct problems and oppositional behavior in
individuals with Attention Deficit Hyperactivity Disorder
(ADHD; Holmes et al. 2002; Kirley et al. 2004), and to
diminished levels of prosocial behavior (Anacker et al.
2013; DiLalla et al. 2009; Jiang et al. 2013).
A recent meta-analysis showed that children with less
efficiently functioning dopamine-related genetic variants
(of which DRD4 was the most studied gene) do worse in
negative parental rearing environments than children
without such alleles (Bakermans-Kranenburg and van
IJzendoorn 2011). At the same time, the authors concluded
that children with susceptibility alleles are also likely to
profit most from positive rearing environments
(Bakermans-Kranenburg and van IJzendoorn 2011). Despite that
the results presented in that meta-analysis generally
supported the differential susceptibility hypothesis, the study
of differential susceptibility of DRD4 to the social
environment is far from complete.
First, although geneenvironment interaction (G 9 E)
studies of DRD4 in the parenting context are fairly
common, only a few studies focused on the peer environment
(i.e., DiLalla et al., 2009; Kretschmer et al. 2013). As said,
children in elementary school function in the presence of
their peers for a large proportion of their day.
Consequently, the peer environment becomes increasingly
important for the development of school-aged children
(Sroufe et al. 2009). None of the studies that investigated
the peer environment 9 DRD4 interaction effects focused
on the elementary school period. DiLalla et al. (2009)
found that preschoolers carrying the DRD4-7r allele
showed more aggression during peer-play in an
environment where there was little peer aggression, while in a
highly aggressive environment all children showed
aggressive behavior regardless of genotype. No evidence of
G 9 E was found for the association between peers
prosocial behavior and childrens own prosocial behavior
in that study. Kretschmer et al. (2013) focused on
victimization and social well-being during adolescence as
predictors of delinquency. These authors found that, in
contrast to previous findings and their own hypotheses, the
adolescents who did not have the DRD4-7r allele, as
opposed to those who did have this allele, were more
susceptible to the effects of victimization and social
wellbeing. Thus, information on the elementary school peer
environment is lacking and the scarce studies with regard
to moderation by DRD4 genotype in the relation between
peer experiences and (mal)adjustment have produced
Second, many previous studies have studied
environmental variables that not all children will be exposed to on
a daily basis and for the better part of the week, such as
bully-victimization, intrusive parenting, or peer aggression
(e.g., DiLalla et al. 2009; Kretschmer et al. 2013; Propper
et al. 2007). It is currently not known whether moderating
effects of DRD4 also extend to peer experiences that
children will encounter on each typical school day. In the
present study we therefore focused on childrens social
preference among peers as the environmental factor of
interest. Peer social preference in the classroom refers to
the extent to which children are liked relative to disliked by
their classmates. It is the result of a natural evaluation
process that occurs in every social setting, for every
individual within that setting (Coie et al. 1982; Rubin et al.
2006). Establishing a positive social standing in the larger
peer-group is a key developmental task for children in
elementary school, which facilitates a healthy behavioral
development (Sroufe et al. 2009). Indeed, the impact of
low social preference within the peer group on behavioral
misconduct in children has been well documented (for
overviews, see Parker et al. 2006; Rubin et al. 2006).
However, and in accordance with the for better and for
worse hypothesis, the influence of peer relations is
multidirectional: being mostly disliked among peers may
elevate the risk for the development of conduct problems and
may hinder prosocial development; in contrast, being
mostly liked may buffer against the development of
conduct problems and may promote prosocial behavioral
development (Ladd 2006; Twenge et al. 2007; Wentzel 2014;
Wentzel and McNamara 1999; Witvliet et al. 2009).
Therefore, by focusing on social preference as the
environmental peer-factor of interest we aim to expand
previous results found in the field of gene 9 peer environment
Third, and related to the previous argument, none of the
previous studies focused on both negative and positive
environments with regard to both negative and positive
outcomes. The study by Kretschmer et al. (2013) focused
on negative and positive peer environmental factors with
respect to predicting negative behavioral outcomes. The
study by DiLalla et al. (2009) focused on a positive peer
environment with respect to predicting positive behavioral
outcomes and a negative environment with respect to
predicting negative behavioral outcomes. Other studies
also focused on either the positive environment or the
negative environment and/or either positive outcomes or
negative outcomes (e.g., see examples in the overview of
Bakermans-Kranenburg and van IJzendoorn 2011).
However, less negative behavioral outcomes or even the
absence of negative behavioral outcomes does not necessarily
mean that behavioral outcomes are positive. This also
applies vice versa: less positive behavioral outcomes or the
absence of positive outcomes does not necessarily mean
that behavioral outcomes are negative. The same holds for
the environment: the absence of a negative environment or
a less negative environment does not necessarily mean that
the environment is positive, and vice versa. Ideally, the
study of differential susceptibility includes both negative
and positive environments as well as both negative and
positive behavioral outcomes to test for all possibilities:
(a) a negative environment predicting more positive
behavioral outcomes and less negative behavioral outcomes
and (b) a positive environment predicting less positive
behavioral outcomes and more negative behavioral
outcomes. To this end, we focused on peer social preference as
our environmental factor of interest and conduct problems
and prosocial behavior as our behavioral outcomes of
interest. Peer social preference encompasses both a risk (i.e.,
negative social preference scores: children who are more
disliked relative to liked) and a protective end (i.e., positive
social preference scores: children who are more liked
relative to disliked). Thus, this allows for a comprehensive
test of the differential susceptibility hypothesis. That is,
moderation by DRD4 genotype in both the for better and
the for worse direction can be tested by including both
positive and negative peer environmental factors with
respect to predicting both positive and negative outcomes.
Lastly, many previous studies suffered from design
limitations because most were cross-sectional or
longitudinal prediction studies that were built upon the assumption
that childrens environment predicts subsequent behavior
and not vice versa. However, previous studies have shown
that associations between social preference and behavior
may be bidirectional: childrens social standing among
peers may influence their behavior and their behavior may
influence their social preference among peers (e.g., Van
Lier and Koot 2010). Thus, when developmental models do
not account for the possibility of these bidirectional effects,
the direction of influence between environmental and
behavioral factors may be obscured. Furthermore, by using
the participants as their own controls, our longitudinal
study in which the behavioral and environmental factors
are assessed in parallel over 4 years enables investigating
whether behavior has changed from a prior baseline level
after experiencing low or high social preference.
Present Study and Hypotheses
Using a sample of mainstream elementary school children
(N = 405) in which social preference, prosocial behavior,
and conduct problems were assessed in parallel, annually
across ages 912 years (four waves), we aimed to extend
previous research on the moderating role of DRD4 in four
ways. First, we focused on the peer environment in
elementary school children, thereby extending studies on
parental environmental factors as well as studies focused
on the peer environment in kindergarten and adolescence.
Second, we focused on a peer environmental factor that all
children experience on a daily basis for the better part of
the week, namely peer social preference. We thereby
expand previous research that used peer factors that likely not
all children are exposed to. Third, by focusing on both
negative and positive peer environmental factors in
predicting both negative and positive behavioral outcomes, we
tested the differential susceptibility hypothesis in a
comprehensive manner. Lastly, we investigated potential
G 9 E effects in a longitudinal design where children were
followed over 4 years, which enabled us to investigate the
direction of influence between the behavioral and
We started by investigating whether positive social
preference scores and negative social preference scores
would be prospectively associated with conduct problems
and prosocial behavioral development, above and beyond
possible direct effects of DRD4 on the environmental and
behavioral variables, as well as above and beyond potential
opposite effects (i.e., behavior affecting social preference).
We hypothesized that children who had higher positive
preference scores would have lower levels of conduct
problems and higher levels of prosocial behavior in
subsequent years, relative to children with lower levels of
positive preference scores. Furthermore, we expected these
effects to be mirrored for children who had negative social
preference scores. That is, we hypothesized that children
who had more negative preference scores would have
higher levels of conduct problems and lower levels of
prosocial behavior in subsequent years, relative to children
with less negative preference scores (hypothesis 1). Within
these models, direct associations between DRD4 and social
preference scores as well as between DRD4 and behavioral
outcomes were explored.
Next we examined our main hypothesis, namely whether
the prospective association between peer social preference
and behavioral development varied as a function of DRD4
polymorphisms. In line with the differential susceptibility
hypothesis, we tested whether the potential moderation by
DRD4 occurred for better and for worse (hypothesis 2).
Specifically, we hypothesized that children who had higher
positive preference scores would have lower levels of
conduct problems and higher levels of prosocial behavior
in subsequent years, but in both cases particularly when
they had a DRD4-7r allele (i.e., G 9 E for better). In
addition, we expected that particularly for children with a
DRD4-7r allele more negative preference scores would be
related to subsequent higher levels of conduct problems
and lower levels of prosocial behavior (i.e., G 9 E for
Participants were children attending 48 different
mainstream elementary schools and were part of two longitudinal
research projects on childrens social, emotional and
behavioral development in the Netherlands. These research
projects were conducted by the department of
Developmental Psychology, VU University Amsterdam. Parental
consent for participation was obtained for a total of 1091
children. In the first project, schools were recruited from two
urban areas in the western part of the Netherlands and one
rural area in the eastern part of the Netherlands. A
convenience sample was utilized in which the first 30 schools that
accepted our invitation to participate in the project were
included. In the other project, eighteen schools from the
northern and the eastern part of the Netherlands were
recruited via municipal health services. In both projects, all
children were followed annually across elementary school.
Additional information on the participants, design, and
procedures is provided elsewhere (Gooren et al. 2011;
Menting et al. 2011). The ethic review boards of the Erasmus
University Rotterdam and the VU University Amsterdam
approved the projects. In first and second grade, a preventive
intervention targeting problem behavior (either the Good
Behavior Game; Barrish et al. 1969; or PATHS curriculum;
Kusche and Greenberg 1994) was implemented in which
approximately 60 % of the children participated, with the
remaining 40 % serving as controls. To prevent
confounding by intervention effects, data covering ages 912 years
(grades 36, four waves) were used in the present study.
Moreover, all estimates were controlled for potential
longterm intervention effects and three-way interactions
including condition (intervention or control; G 9 E 9
condition) were tested. More detailed information about both
interventions can be found in Appendix 1.
At age 13, children were asked to provide DNA through
a saliva sample. Children and parents who granted
permission were eligible for inclusion in the present study
(N = 406; 51 % girls). DRD4 genotyping was successful
for 405 out of the 406 subjects. Of these, 143 (35 %)
subjects carried one or two 7-repeat alleles (referred to as
DRD4-7r) and 262 (65 %) subjects carried no 7-repeat
alleles (referred to as DRD4-no7). Of the DRD4-no7
group, all but 2 children carried either a 2-repeat allele or a
4-repeat allele. More details on the distribution of the
DRD4 polymorphisms and the assignment to groups is
provided in Appendix 3.
Eighteen percent of the children came from low
socioeconomic status (SES) families. Furthermore, 87 % of
the present sample had a Dutch/Caucasian background,
3.8 % were Moroccan, 3.8 % were Surinamese, 2 % were
from the Netherlands Antilles, and 3.4 % of the children
came from other ethnical backgrounds (i.e., Turkey,
Somalia, Pakistan, Iraq, Congo-Kinshasa, and Sri Lanka).
Given that the DRD4-environment interaction may be
dependent on race (e.g., Propper et al. 2007), we examined
whether results changed when only native Dutch (i.e.,
Caucasian) children remained in the sample. In addition,
because the developmental relation between peer
experiences and subsequent behavioral development may differ
for boys and girls (Moffitt et al. 2001; Van Lier and Koot
2010; Witvliet et al. 2009) and that moderating effects of
DRD4 may be influenced by the childs sex (Froehlich
et al. 2007), we investigated potential sex differences in the
moderation by DRD4 (i.e., G 9 E 9 sex).
Participants who declined participation in DNA
collection did not differ from those who conceded with
participation on average levels of conduct problems, (F(1,
973) = 2.49, p = .12) or negative social preference scores
(F(1, 1089) = 1.48, p = .22) over ages 912 years.
However, children who declined participation compared to
children who participated had slightly lower average levels
of prosocial behavior (F(1, 972) = 11.44, p \ .01,
g2 = .01; M = 2.87, SD = 0.57 for children who
participated, M = 2.74, SD = 0.62 for children who declined
participation), as well as slightly lower levels of positive
social preference scores (F(1, 1010) = 6.27, p \ .05,
g2 \ .01; M = 0.23, SD = 0.16 for children who
participated, M = 0.20, SD = 0.17 for children who declined
participation) over ages 9 to 12 years. During the follow-up
period used in the present study, data of 91 % of the
children were complete for at least two measurement
moments. Missing data was due to retention, moving to
another school, or because of absence during the
measurements. Children with missing data did not differ
from children with complete data on any of the study
variables in third grade, indicating that there was no
evidence for selective attrition during the period investigated
in the present study.
Teacher Ratings of Conduct Problems
Teacher ratings of conduct problems were assessed
annually with the conduct problems scale from the Problem
Behavior at School Interview (PBSI; Erasmus 2000). The
PBSI is a face-to-face interview in which teachers rated
pupils behavior on a five-point Likert-scale ranging from 0
(never applicable) to 4 (often applicable). Conduct
problems were assessed by 12 items (range a over the
assessments = .90.92). Sample items include: attacks other
children physically, bullies, steals, destroys
property belonging to other children, is absent from school
without permission, curses or swears. Item scores were
averaged, resulting in a scale ranging from 0 to 4.
Teacher Ratings of Prosocial Behavior
Teacher ratings of prosocial behavior were assessed
annually with the prosocial behavior scale from the Social
Experiences Questionnaire (SEQ-T; Crick and Grotpeter
1996). During a face-to-face interview teachers rated
pupils behavior on a 5-point Likert scale ranging from 0
(never applicable) to 4 (often applicable). Prosocial
behavior was assessed by 4 items (range a over the
assessments = .75.83). Sample items include: Comforts a
child who is sad and Is nice to other children. Item
scores were averaged, resulting in a scale ranging from 0 to
Peer Nominations on Social Preference
Peer nominations on social preference were obtained by
asking children to nominate an unlimited number of
children in their classroom whom they liked most and whom
they liked least. The liked least scores of each child were
subtracted from his or her liked most scores to obtain a
social preference score. This score was divided by the total
number of children in the classroom, minus one (it was not
allowed to nominate oneself), resulting in a score ranging
from -1 (disliked by all classmates and liked by none) to
?1 (liked by all classmates and disliked by none). This
procedure was adapted from the protocol described by Coie
et al. (1982). Social preference is generally regarded as a
reliable and valid measure of sociometric status (Rubin
et al. 2006). We then differentiated between children with
positive social preference scores, that is children who were
more liked relative to disliked and children with negative
social preference scores, that is children who were more
disliked relative to liked. Negative social preference scores
were then multiplied by -1 such that higher scores
reflected a more negative social preference score. Children
who were equally liked as disliked or who were not
nominated at all (between 3.3 and 6.5 % of all children
throughout ages 912 years) received a score of zero.
was dummy coded as 0 = female,
Household Socioeconomic Status (SES)
SES was measured through parental occupation in third
grade. Fathers and mothers occupations were classified
into one of five levels (0 = unemployed, 1 = elementary
level, 2 = lower level, 3 = medium level, 4 = higher
level). Levels of occupation were assigned according to the
Dutch Working Population Classifications of Occupations
Scheme (Statistics Netherlands 2001), which is based upon
the International Standard Classification of Occupations
(ISCO; International Labour Organization 1987a, b). The
highest occupation level (from father or mother) was
considered to reflect household SES. Household SES was
then dummy coded as 0 = medium to higher level SES,
1 = unemployed to lower level SES.
Intervention status was dummy coded as 0 = no
intervention, 1 = intervention.
Genotyping of VNTR in Exon 3 of DRD4
DNA was extracted from saliva using the OraganeTM DNA
Self-collection Kit according to the manufacturers
instructions (DNAGenotek, Ottawa, Ontario, CAN). The 48
base pair VNTR in exon 3 of DRD4 (211 repeats) was
genotyped using PCR and fragment analysis on a 3130
Genetic Analyzer (Life Technologies, Carlsbad, CA). The
PCR assay was a modification of the method by Boor and
colleagues (Bo or et al. 2002). In accordance with previous
studies (e.g., Kretschmer et al. 2013), children were coded
as DRD4-7r (at least one allele had 7-repeats) or
DRD4no7 (no 7-repeat alleles).
Autoregressive cross-lagged models (Joreskog 1970) were
used to test our two hypotheses. Models were fitted in
Mplus 6.11, Los Angeles, California (Muthen and Muthen
19982011). We aimed to test links between social
preference scores, conduct problems and prosocial behavior in
two separate models. That is, we specified one model for
links between positive social preference scores and
behavioral development and another model for links between
negative social preference scores and behavioral
development. Within each model, autoregressive paths from ages 9
to 12 years tested for stability within the environmental
and behavioral constructs, while cross-lagged paths
assessed the developmental links between these constructs (see
Fig. 1 for an illustration). All estimates were controlled for
potential long-term intervention effects, SES status and
Given that statistical power is a major concern in modern
behavioral genetics (Duncan and Keller 2011), we
conducted an a priori Monte Carlo simulation study to ensure
that power was sufficient given our models and sample size
before starting with testing our hypotheses. Statistical
power is the probability of detecting a significant result
given that the alternative hypothesis (in our study: that
particularly children with a 7-repeat allele are susceptible
to the peer environment for better and for worse) is true.
Low statistical power is problematic, because it implies
that true findings are likely to be missed (type II error) and
because low power increases the proportion of significant
results that are published, but that are actually false (type I
Monte Carlo Simulation
In an a priori Monte Carlo analysis, data are generated
from a population with hypothesized parameter values.
Then, a large number of samples are drawn and a model is
estimated for each sample. Parameter values and standard
errors are averaged over the samples (Muthen and Muthen
2002). We expected effects for negative social preference
and positive social preference to be similar, thus we only
investigated power for the model including positive social
preference. We used 10,000 replications to ensure that
stability would be reached. Data for a multiple-group
model were generated using the following population
values (see also Appendix 2). For the DRD4-7r group as
well as for the DRD4-no7 group, means and variances of
variables were standardized to 0 and 1 respectively; the
standardized regression coefficients for autoregressive
paths of social preference, conduct problems and prosocial
behavior were all 0.60; standardized regression coefficients
of lagged paths from behavioral outcomes to social
preference were 0.05 and -0.05 for prosocial behavior and
conduct problems respectively; and standardized residual
correlations were 0.10 between social preference and
prosocial behavior and -0.10 for social preference and
conduct problems and for conduct problems and prosocial
behavior. For the DRD4-no7 group the standardized
regression coefficients of the lagged paths from social
preference to prosocial behavior as well as to conduct problems
were 0. These values were chosen based upon Keith and
colleagues consideration that within the social sciences
estimates (i.e., standardized regression coefficients) \0.05
are too small to interpret, estimates C0.05 are small but
meaningful, estimates C0.10 are moderate, and estimates
C0.25 are large (Keith 2006; Keith and Cool 1992).
The focus of the power investigation in the
multiplegroup autoregressive cross-lagged model was the
standardized regression coefficient of the lagged paths from
social preference to prosocial behavior and to conduct
problems for the DRD4-7r group. Different standardized
regression coefficients were estimated, starting from 0.05
(which is a small, but meaningful effect; Keith 2006) until
a power of 0.80 by p \ .05 was reached. Results are in
Appendix 2. These indicated that a power of 0.80 (p \ .
05) would be reached when the standardized regression
coefficients would be 0.12 for the link between positive
social preference and subsequent prosocial behavior and
-0.12 for the link between positive social preference and
subsequent conduct problems. A beta of 0.12 indicates a
moderate effect in the social sciences (Keith 2006), which
we deemed both reasonable and relevant. Under the
condition of no effect (i.e., b = 0) for the DRD4-no7 group,
this results in a significant difference in slopes at p \ .01
when standard errors are 0.10 for the DRD4-7r group and
0.01 for the DRD4-no7 group (which is a rather larger
difference in SEs and thus a stringent test of differences
between slopes). Furthermore, coverage for the parameters
of interest was 0.94, which indicates that the 95 %
confidence intervals of 94 % of the 10,000 replications include
the population value of 0.12 (prosocial behavior) and
-0.12 (conduct problems; see Table 4 in Appendix 2).
Hence, we assumed power to be sufficient to test our
hypothesis on G 9 E effects.
After sufficient power was assured, our two hypotheses
were tested as follows. We first tested for the prospective
influence of social preference on subsequent behavioral
development. To this end, we started with a model that
included autoregressive paths and cross-lagged paths, in
addition to cross-sectional correlations between social
preference and the behavioral phenotypes (models 1; see
example in Fig. 1). We also included direct effects of
genotype on the environmental and behavioral variables.
This model allowed us to test bidirectional effects (i.e.,
whether positive/negative social preference scores added to
Fig. 1 Illustration of the model used for hypotheses testing. This model was tested for positive social preference and negative social preference
behavioral development above and beyond possible
prospective associations between behavioral development
and subsequent environmental changes), cross-sectional
correlations, and direct effects of DRD4 (hypothesis 1). We
tested these models separately for positive social
preference scores and negative social preference scores, but the
development of prosocial behavior and conduct problems
was estimated simultaneously. We then continued by
testing whether recurring autoregressive and cross-lagged
paths could be constrained to be equal over time in order to
create parsimonious models (models 2).
Next, we tested our second and main hypothesis, namely
whether DRD4 moderated the prospective link between
social preference and behavioral development. The
following hierarchy of nested model comparisons was applied
to test for potential differences between DRD4-7r and the
DRD4-no7 groups. Multiple-group models were used in
which children with the DRD4-7r allele were compared to
children with DRD4-no7 alleles. First of all, all parameters
were freely estimated between the groups (models 3); next,
we tested whether pathways that were not part of our
hypotheses (i.e., autoregressive paths and paths from the
behavioral constructs to the environment) were equal
between groups (models 4); and lastly, we investigated our
hypothesized G 9 E effects by testing whether paths
between social preference and prosocial behavior (models 5)
and between social preference and conduct problems
(models 6) were equal between groups. As said, two
models were tested: one for positive social preference
scores and one for negative social preference scores. In
order to support our second hypothesis, constraining
autoregressive paths and paths from the behavioral constructs
to social preference to be equal between DRD4-7r carriers
and DRD4-no7 carriers (models 4) should not significantly
decrease model fit, while constraining the pathways
between social preference and behavioral phenotypes to be
equal for DRD4-7r carriers and DRD4-no7 carriers
(models 5 and 6) should result in a significant drop in fit. In each
model testing step, constraints that did not result in a
significant drop in model fit were remained in subsequent
Full Information Maximum likelihood estimation with
robust standard errors (FIML, MLR-estimator) was used to
account for missing data. We accounted for clustering of
data within schools by using a sandwich estimator
(Williams 2000). The SatorraBentler Chi square difference test
was used to compare nested models (Satorra, 2000). Model
fit was determined via the Comparative Fit Index (CFI;
with values [.95 indicating acceptable fit), and the
Standardized Root Mean Squared Residual (SRMR; with values
B.08 being acceptable) (Hu and Bentler 1998; Marsh et al.
2004). We tested for potential sex-differences and
differences due to intervention status in the moderation by
DRD4 using three-way interactions (G 9 E 9 sex and
G 9 E 9 condition, respectively). Furthermore, using the
equation provided by Duncan and Keller (2011) we
calculated the False Discovery Rate (FDR) from Monte Carlo
power analyses. The FDR indicates the proportion of false
discoveries (i.e., the proportion of false support for our
hypotheses when this support actually represents type I
Distribution of the DRD4 polymorphisms was comparable
to reported global repeat frequencies (see Appendix 3;
Chang et al. 1996). Allele frequencies of DRD4
polymorphisms were analyzed from HardyWeinberg equilibrium
(HWE) using v2 tests. No deviations from HWE were
detected, v2 (2) = 0.20, p = .90.
Table 1 gives the means and SDs for study variables for
boys and girls. Repeated measures analyses of variance
(ANOVAs) indicated that from ages 9 to 12 years, boys had
on average higher levels of conduct problems (F(1,
394) = 57.83, p \ .001, g2 = .13), and lower levels of
prosocial behavior (F(1, 395) = 80.50, p \ .001, g2 = .17),
than girls. In addition, boys had slightly lower levels of
positive social preference scores (F(1, 362) = 15.30,
p \ .001, g2 = .04), and slightly higher levels of negative
social preference scores (F(1, 363) = 10.69, p \ .01,
g2 = .03), than girls. Correlations between study variables
in Table 1 indicated significant cross-time correlations of
conduct problems, prosocial behavior, positive and negative
social preference in the expected directions. Furthermore,
repeated measures ANOVAs indicated that DRD4-7r and
DRD4-no7 carriers did not differ in their average levels of
conduct problems, prosocial behavior or social preference
throughout ages 9 to 12 years.
We started by investigating the prospective associations
between peer social preference and behavioral outcomes
over time. No moderation by DRD4 genotype was tested at
this stage. Links between positive social preference scores,
conduct problems and prosocial behavior and negative
social preference scores, conduct problems and prosocial
behavior were tested in two separate models (see Fig. 1).
We fitted bivariate cross-lagged autoregressive models
with stability paths and directional paths from social
preference to behavior and vice versa, in addition to
crosssectional correlations. Direct effects between DRD4 and
social preferences and between DRD4 and behavioral
outcomes were also included in the models.
Results of model fitting are presented in Table 2. The
two models fitted the data adequately according to fit
indices (models 1; CFIs C .95, SRMRs B .06).
Constraining recurring autoregressive and lagged paths to be
equal over time (model 2) did not result in worsened model
fit for any of the two models (see Table 2). Therefore, these
time-constraints were retained in the models. Estimates for
models 2 are displayed in Table 3. For conduct problem
development, neither positive social preference scores nor
negative social preference scores were related to
subsequent conduct problem development, although trends were
observed (i.e., p B .08). Furthermore, the paths from
conduct problems to subsequent positive social preference as
well as negative social preference were non-significant,
although in the latter link again a trend was observed (i.e.,
p B .07.)
For prosocial behavioral development, higher positive
social preference scores were related to higher subsequent
prosocial behavior and more negative social preference
scores were related to lower levels of subsequent prosocial
behavior. Furthermore, higher levels of prosocial behavior
were related to higher levels of subsequent positive social
preference, while the paths between prosocial behavior and
negative social preference scores were non-significant.
These effects were found above and beyond stability paths
and cross-sectional correlations, and all estimates were
controlled for sex, SES and intervention status.
Furthermore, neither the direct effects of DRD4 on social
preference, nor the direct relationships between DRD4 and
behavioral outcomes were significant.
Hypothesis 2: Differential Susceptibility of DRD4
to the Environment For Better and for Worse
We then tested whether the magnitude of the prospective
links between positive and negative social preference
scores, prosocial behavior and conduct problems (see
Fig. 1 for an illustration), were different for DRD4-7r and
DRD4-no7 children (hypothesis 2). Multiple group models
were used (DRD4-no7 versus DRD4-7r). Table 2 shows fit
indices for models in which all paths were estimated freely
between DRD4 groups (models 3), models in which the
paths that were not part of our hypothesis were constrained
to be equal between the DRD4 groups (models 4), and
models in which developmental pathways from social
preference to the behavioral outcomes were constrained to
be equal between DRD4 groups (models 5 and 6).
Comparisons of fit indices showed evidence for
moderation by DRD4 in the link between social preference and
subsequent conduct problems only. As can be seen in
1. Base model
2. Time constraints
3. No constraints
4. Non-hypothesized paths equal
5. GxE: positive social preference ? prosocial behavior equal
6. GxE: positive social preference ? conduct problems equal
1. Base model
2. Time constraints
3. No constraints
4. Non-hypothesized paths equal
5. GxE: negative social preference ? prosocial behavior equal
6. GxE: negative social preference ? conduct problems equal
Dv2 statistics are based on the SatorraBentler Chi square difference test
Table 3 Coefficients for paths
between positive social
preference, negative social
preference and behavioral
Positive social preference
Positive social preference predicting prosocial behavior
Prosocial behavior predicting positive social preference
Positive social preference predicting conduct problems
Conduct problems predicting positive social preference
Negative social preference
Negative social preference predicting prosocial behavior
Prosocial behavior predicting negative social preference
Negative social preference predicting conduct problems
Conduct problems predicting negative social preference
Table 2 Geneenvironment interactions between DRD4 and positive as well as a negative peer social preference in predicting conduct problems
and prosocial behavior: fit statistics and nested model comparisons
As recurring paths were constrained to be similar over time, these results apply to all recurring paths in the
Table 2, multiple group models in which paths between
social preference and subsequent conduct problems were
estimated freely between the DRD4-7r and DRD4-no7
groups (models 5), are the best fitting models for children
with positive as well as for children with negative social
preference scores. Results from analyses for prosocial
behavior indicate that neither positive nor negative social
preference scores had a differential effect on prosocial
behavior as a function of DRD4 (see Table 2).
Estimates of geneenvironment interaction effects for
conduct problem development are in Fig. 2. Figure 2
shows that positive social preferences scores were
prospectively associated with lower levels of conduct
problems, but only among DRD4-7r carriers. These effects
were mirrored for children with negative social preference
scores. That is, being more disliked than liked among peers
was associated with more conduct problems, but again only
among DRD4-7r carriers. No relation was found between
the positive or negative social preference scores and
conduct problems for DRD4-no7 children. Note that no G 9 E
interaction effect was found for prosocial behavior. Hence,
estimates for associations between social preference and
Fig. 2 Multiple-group (DRD4-7r vs. DRD4-no7) model of positive
social preference (a) and negative social preference (b) predicting
conduct problems. Results are a graphical presentation of models 5.
Entries reflect standardized regression coefficients. Paths that were
different for the DRD4-7r and DRD4-no7 children have two
coefficients: upper entries are estimates for DRD4-7r, lower entries
are estimates for DRD4-no7. All entries are controlled for sex, SES
and intervention status. Dashed lines represent non-significant
pathways. *Significant at p \ .05, **significant at p \ .01,
***significant at p \ .001
prosocial behavior were similar for the DRD4-7r and
DRD4-no7 groups (i.e., similar to findings of the total
sample) and can be found in Table 3.
We ran a number of additional tests to test the
robustness of our findings. First, potential effects of ethnicity
were tested. Specifically, we investigated whether results
were similar when only native Dutch children remained in
the sample (N = 342; n = 127 for DRD4-7r, n = 215 for
DRD4-no7). Results of these tests indicated that removing
non-Dutch children from the sample did not influence the
results for nested model comparisons. Second, we tested
whether the moderating role of DRD4 in the prediction of
conduct problems from social preference scores were
influenced by the childrens sex. To this end, we investigated
the effects of three-way interactions (G 9 E 9 sex) on
conduct problem and prosocial behavior development,
which were all non-significant. Thus the moderation of
DRD4 in the association between social preference
(positive or negative), prosocial behavior, and conduct problems
did not differ between boys and girls. Third, we tested
whether the moderating role of DRD4 in the prediction of
prosocial behavior and conduct problems from social
preference scores was influenced by whether or not
children had participated in an intervention. To this end, we
investigated the effects of three-way interactions
(G 9 E 9 intervention status) on conduct problem and
prosocial behavior development, which were all
non-significant. Thus the moderation of DRD4 in the association
between social preference (positive or negative), prosocial
behavior, and conduct problems was not dependent upon
intervention status. Lastly, we performed post Monte Carlo
power analyses (10,000 repetitions) using our sample
estimates to calculate the False Discovery Rate (FDR) in our
study. Power for our parameters of interest was .0.95 and
1.00 for predicting conduct problem development from
positive and negative social preference respectively, which
equaled a FDR of 0.05 and 0.01 for positive and negative
social preference respectively. This indicates that 5 % of
evidence for our hypotheses for positive social preference
and 1 % of evidence for our hypotheses for negative social
preference with regard to conduct problem development,
may actually be type 1 errors.
The main aim of the current study was to investigate whether
the dopamine receptor D4 gene (DRD4) moderated the
association of positive social preference (i.e., children that were
more liked than disliked among classmates) and negative
social preference (i.e., children that were more disliked than
liked among classmates) among peers with subsequent
positive and negative behavioral development. This study was one
of the first to investigate differential susceptibility of DRD4 to
a common peer environmental experience that covers positive
as well as negative aspects of the peer environment. Our first
hypothesis that social preference would be related to
behavioral development in subsequent years for the group in total
was only partially supported. That is, throughout ages
912 years children with higher positive social preference
scores showed a larger increase in prosocial behavior in
subsequent years than children with lower positive social
preference scores. This effect was mirrored for negative social
preference scores: children with more negative social
preference scores showed a larger decrease in prosocial behavior in
subsequent years compared to children with less negative
social preference scores. Contrary to our expectations, we did
not find strong evidence for developmental links between
social preference (either positive or negative) and conduct
problems in subsequent years for the group in total, although a
trend was observed for these developmental links. Our second
hypothesis that developmental links between social
preference and behavioral outcomes would be moderated by DRD4
for better and for worse was also partially supported. As we
hypothesized, we found that throughout ages 912 years
children with higher positive social preference scores showed
a larger decrease in subsequent conduct problem development
relative to children with lower positive preference scores and
that children with more negative social preference scores
showed an larger increase in subsequent conduct problem
development relative to children with less negative social
preference scores, but in both cases only when they carried a
DRD4-7r allele. When children did not have this allele, their
conduct problem development was not influenced by their
social preference among peers. In contrast and contrary to our
expectations, prosocial behavioral development was
influenced by negative as well as positive social preference among
peers regardless of the genetic make-up of the children. Taken
together, these findings provide evidence in support of the
differential susceptibility hypothesis of DRD4 for conduct
problem development, but not for the development of
Our findings add to existing knowledge on individual
differences in the impact of peer environmental aspects,
dependent upon childrens genetic make-up. It concurs with
previous studies on bully-victimization (Kretschmer et al.
2013) and peer aggression (DiLalla et al. 2009), in that
dopamine-related genes are of importance in understanding
the impact of peer environmental factors on behavioral
development. Specifically, the results we found in children
followed from age 9 to 12 years are in line with DiLalla et al.
(2009) who focused on geneenvironment interplay in
kindergarten and found that children with the DRD4-7
repeat allele were particularly susceptible to their peer
environment. Interestingly, Kretschmer et al. (2013) found an
opposite effect for adolescents aged 1318 years of age.
That is, their results suggested that it are the DRD4-no7
repeat carriers and not the 7-repeat carriers who are
particularly susceptible to the negative as well as the positive
environment. A possible explanation for these differential
effects for younger versus older children may be that
adolescence is a developmental period in which major
neurological and biological changes occur, which may influence
the effect of DRD4 polymorphisms on behavior/outcomes
(Kretschmer et al. 2013). Our findings extend these previous
studies by showing that the DRD4-7r allele may not only
affect how children respond to these rather extreme peer
experiences, but also influences how children respond to
common peer evaluations that all children encounter on a
daily basis over the elementary school years. In addition,
together with the studies of DiLalla et al. (2009) and
Kretschmer et al. (2013), the present results warrant
attention to the specific developmental period that is under
investigation as results from geneenvironment interactions
may change throughout development.
It is important to note that our findings on
geneenvironment interplay only held for conduct problems and not
for prosocial behavior. In line with differential susceptibility
theorizing that DRD4 moderation of environmental effects
would be for better and for worse, we expected this
moderation to be domain general in that both the
development of conduct problems and the development of prosocial
behavior would be affected. However, our results suggest
that this moderation is domain specific. Specifically, our
results suggest that DRD4 effects likely depend on the
specific environment-behavioral phenotype relation that is
investigated. In line with this suggestion, DiLalla et al.
(2009) found DRD4 to only moderate the effect of peer
aggression on childrens aggressive behavior, but DRD4 did
not moderate the effect of peer prosocial behavior on
childrens prosocial behavior. As such, the present findings and
those of DiLalla et al. (2009) both contribute to a rapidly
accumulating body of knowledge that will eventually inform
us about the extent to which differential susceptibility
effects are domain general or domain specific.
The present findings suggest that Belskys (1997)
differential susceptibility theory may not only apply to rearing
practices, but also to the peer environment. When susceptible
childrens position within the peer group is threatened by
peer rejection or low preference, one way to strengthen their
position is through the use of dominance-oriented social
strategies, including aggression (Reijntjes et al. 2013). This
is likely to increase individuals social dominance position
which improves their chances for obtaining attractive
resources and (in the future) makes them attractive for mating
(Pellegrini and Long 2003), thus improving their chances for
reproduction. For susceptible children who are socially
preferred by their peers, behaving aggressively to strengthen
their dominance position in the peer group is not necessary
and given dangerous side-effects (like becoming injured
from fighting) may even be undesirable, thus explaining the
decrease in subsequent conduct problem development for
socially-preferred susceptible children.
Children who were less susceptible (i.e., DRD4-no7
carriers) seemed to be unaffected by their peer environment
in that their conduct problem development was not
influenced by their social standing among peers. Perhaps children
with dopamine-related alleles that are not related to
decreased postsynaptic inhibition (e.g., children with
DRD4no7 alleles) have better self-regulatory skills. There is
indeed some evidence pointing in this direction (Fan et al.
2003; Fossella et al. 2002; Posner and Rothbart 2009). Better
self-regulatory skills may facilitate effective socialization
and may enable children to inhibit inappropriate responses
like conduct disordered behavior and to behave in
accordance with social demands from parents, teachers, and peers.
In line with Belskys (1997) reasoning regarding differential
susceptibility to parenting, it makes evolutionary sense that
some children are particularly vulnerable to their peer
environment and adapt their behavior accordingly, while
others are not influenced by their peers. Future research may
elaborate on this suggestion by investigating differential
susceptibility of children with DRD4-7r alleles to the peer
environment in relation to other behavioral strategies that
may strengthen their position in the peer group, such as the
combined use of both aggressive and cooperative strategies
(Hawley 1999) and behaving as a bully (Olthof et al. 2011).
This study is not without limitations. First of all, although
we used a normative sample, the selection of schools was not
at random. Children included in our study came from
families with higher SES status than is generally reported for
the Dutch population (Statistics Netherlands 2012).
Furthermore, children whose parents did not consent to having
their childs DNA collected had slightly lower positive social
preference scores was well as slightly lower levels of
prosocial behavior than children that did participate in the
DNA collection. Although the reported differences were
small, we cannot be certain that the results generalize to the
broader Dutch population. Second, we used teacher reports
on childrens prosocial behavior and conduct problems.
Teachers may not be aware of these behaviors outside the
school context and children may hide certain conduct
problems, such as stealing, from their teacher. Although previous
studies have indicated that teachers are valid informants of
childrens conduct problems and prosocial behavior (Becker
et al. 2004; Hart et al. 1994), our results should only be
interpreted within the school context. Third, influences of
peers as assessed in this study were limited to peers within
the classroom. However, poor relations with peers outside
the classroom may also affect childrens behavior. Although
others have shown that influences of peers outside of the
school context are limited for elementary school children
(Kupersmidt et al. 1995), we cannot be certain that peers
outside the classroom have not influenced our results. Fourth,
by investigating the influence of social preference on
subsequent behavioral phenotypes while taking into account the
stability of these constructs as well as concurrent links
between environment and behavior, we were able to identify
the actual change in behavioral phenotypes that can be
ascribed to peer environment, genetic effects, and their
interplay. However, we want to stress that no causality can be
inferred from this design. Fifth, although we took both the for
better and the for worse side of the differential susceptibility
hypothesis into account, we could not directly examine
whether the same children who do worse than comparisons in
adverse peer environments, also do better when they
experience supportive peer environments. Future studies may
want to include designs that allow studying the same children
in various environmental conditions, such as an
experimental study in which the same children encounter peer
exclusion as well as inclusion situations (Rutter et al. 2001).
In addition, from our study it cannot be inferred which brain
processes and neurocognitive functions that are associated
with the DRD4 gene account for our differential
susceptibility findings. This is of particular importance given the
different results that have been found for kindergarten and
elementary school children versus older adolescents. Future
studies may want to investigate these brain processes and
neurocognitive functioning that are associated with
differential susceptibility (Ellis and Boyce 2011), ideally within a
developmental framework in which potential differences in
brain processes and functioning throughout development can
be studied. As a last and perhaps most important limitation
we want to note that we were not able to directly replicate our
results in an independent sample using the same measures.
Therefore, our results should be interpreted with caution
The DRD4 7-repeat allele may render children and young
adolescents susceptible to their everyday peer environment
for better and for worse with regard to subsequent conduct
problem development. We found that, throughout ages
912 years, children who experienced a more positive peer
environment at a given age showed less conduct problem
development 1 year later when compared to children who
experienced a less positive environment; vice versa,
children who experienced a more negative peer environment
showed more conduct problem development in subsequent
years relative to children who experienced a less negative
environment. However, in both situations these effects only
held when children had a DRD4-7 repeat allele. Integral
strengths of this study were the use of a peer environmental
factor that included both a protective and a risk end to
assess how a positive and negative daily peer environment
may influence the development of conduct problems and
prosocial behavior and whether allelic variations within the
DRD4 gene may moderate these developmental relations.
Other strengths include the use of multiple informants and
our longitudinal design. Our findings enhance further
understanding of the developmental relationship between
youths social standing among peers and subsequent
behavioral development and advance current knowledge on
why some, but not all, children and adolescents are
influenced by peer experiences. We suggest that part of the
individual differences in responding to the peer
environment may be explained by differences in the genetic
makeup of these individuals.
Furthermore, our findings have implications for
preventive interventions for those children at risk for conduct
problem development. The peer environment, regardless
whether this environment is positive or negative, affects
conduct problem development for those children who are
susceptible to it. Preventive interventions that succeed in
prohibiting the development of poor peer preference or that
improve disliked childrens appraisal among peers to a more
neutral level, may decrease the development of conduct
problems in susceptible children. Although research on
endophenotypes related to susceptibility is still in its infancy,
future discoveries of endophenotypes associated with
susceptibility may advance the early screening of at-risk
children that likely will profit from improvements in peer
appraisal. At the same time, as others have suggested
(Bakermans-Kranenburg and van IJzendoorn 2011), early
detection of those children who likely will not benefit from
preventions targeting the peer environment may ideally lead
to more individual-based interventions and thus more
effective strategies of targeting conduct problem development.
Acknowledgments We would like to thank our colleagues Franca
Leeuwis (VU University, Department of Developmental Psychology,
Amsterdam, The Netherlands) for contributing to the DNA collection
process, Karin J. H. Verweij (VU University, Department of Biological
Psychology, Amsterdam, The Netherlands) for providing feedback on
the manuscript, and Brittany Evans (VU University, Department of
Developmental Psychology, Amsterdam, The Netherlands) for editing
our manuscript. This study was financially supported by the
Netherlands Organization for Health Research and Development (ZonMw)
Grants #26200002 and #50-50110-96-514 and the Netherlands
Organization for Scientific Research (NWO) Grant #120620029.
Author contributions JMB conceived of the study, participated in
its design and coordination, drafted the manuscript, performed the
analyses and the main interpretation of the data; PAC v L conceived
of the study, participated in its design and coordination, participated
in interpretation of the data, and extensively reviewed the
manuscript.; KN wrote the paragraph on DNA-subtraction; TO reviewed
the manuscript and wrote (parts of) one paragraph; HMK conceived
of the study, participated in its design, participated in interpretation of
the data, and reviewed the manuscript. All authors read and approved
the final manuscript.
Informed consent All procedures followed were in accordance
with the ethical standards of the responsible committee on human
experimentation (institutional and national) and with the Helsinki
Declaration of 1975, as revised in 2000. Informed consent was
obtained from all individuals for being included in the study.
Open Access This article is distributed under the terms of the
Creative Commons Attribution 4.0 International License (http://
creativecommons.org/licenses/by/4.0/), which permits unrestricted
use, distribution, and reproduction in any medium, provided you give
appropriate credit to the original author(s) and the source, provide a
link to the Creative Commons license, and indicate if changes were
Appendix 1: Description of Interventions
Approximately 60 % of the children in the present study
participated in a preventive intervention targeting problem
behavior. Half of the children in the intervention group
received the Good Behavior Game intervention (Barrish
et al. 1969) and the other 50 % of the intervention children
received the PATHS curriculum intervention (Kusche and
Greenberg 1994). The interventions were implemented in
first and second grade of elementary school.
Good Behavior Game (GBG)
The GBG (Barrish et al. 1969) is a classroom-based
preventive intervention aimed at creating a safe and
predictable classroom environment, by promoting adaptive,
prosocial classroom behavior. Positively formulated class
rules are chosen by the teacher and the students together.
To facilitate positive peer interaction, teachers assign
children to teams of 4 to 5 members, equally composed of
children with and without disruptive behavior. Team
members are encouraged to work together and behave
adaptively. All teams receive a set of cards at the beginning
of the game period in which children work on regular
school tasks (e.g., instruction, working alone, reading).
Each time a member violates a rule, the teacher takes a
card away from that team. Teams as a whole are rewarded
(e.g. by extra leisure time, stickers, compliments) for
adaptive behavior when at least one card remains at the end
of the game period. Game periods lasted between 10 and
60 minutes. During and after the game, compliments are
given to the students and teams when deemed appropriate
(Dolan et al. 1989).
PATHS (Kusche and Greenberg 1994) is a program that
targets the development of social and emotional
competence in order to decrease the risk of behavioral and social
problems. Emotional, cognitive, and social skills are
promoted through lessons taught by the teacher. PATHS
emphasizes techniques to promote positive interaction
amongst students and to reduce peer rejection. For
instance, children are taught to adequately express and
understand peers emotions by using so-called emotion
cards. Also, children learn problem-solving and
angermanagement techniques that are generalized throughout the
classroom and the school context. Furthermore, the child
of the week receives particular attention and is allowed to
help the teacher throughout the week.
Appendix 2: Model Specifications and Outcomes
Power Analysis Using Monte Carlo Simulations
See Table 4.
Autoregressive paths positive social preference
Autoregressive paths prosocial behavior
Autoregressive paths conduct problems
Positive social preference predicting prosocial behavior
Positive social preference predicting conduct problems
Prosocial behavior predicting positive social preference
Conduct problems predicting positive social preference
Correlations positive social preference and prosocial behavior
Correlations positive social preference and conduct problems
Correlations prosocial behavior and conduct problems
Estimates of paths reflect standardized regression coefficients. Correlations between constructs reflect residual error correlations. Means of all
constructs were estimated to be 0 and variances of all constructs were estimated to be 1. Recurring paths were constrained to be similar over time,
hence estimates hold for all recurring paths. Estimates \ 0.05 are considered too small to interpret, estimates C0.05 are small but meaningful,
estimates C0.10 are moderate, estimates C0.25 are large (Keith 2006)
Appendix 3: Distribution of the DRD4
Polymorphisms and Assignment to Groups
See Table 5.
DRD4-no7 includes participants with no 7-repeat alleles. DRD4-7r
includes participants with at least one 7-repeat allele. The three most
common repeat frequencies in our sample were the 4-repeat (66 %),
the 7-repeat (19 %), and the 2-repeat (7 %)
J. Marieke Buil, M.Sc. is a Ph.D.-student at VU University
Amsterdam, the Netherlands. She received a M.Sc. in Educational
Sciences, from Leiden University, the Netherlands. Her major
research interests include: Behavioral problems and development;
delinquency; peer-relations, and longitudinal research.
Hans M. Koot, Ph.D. is a Professor at VU University Amsterdam, the
Netherlands. He received his doctorate in Medicine from Erasmus
University Rotterdam, the Netherlands. His major research interests
include: Emotional and behavioral development; behavioral and
emotional problems; psychopathology; delinquency; quality of life;
intellectual disability; epidemiology; longitudinal research,
Tjeert Olthof, Ph.D. is an Assistant Professor at VU University
Amsterdam, the Netherlands. He received his doctorate in Psychology
from Radboud University Nijmegen, the Netherlands. His major
research interests include: Development of self-conscious emotions;
self-conscious emotions and anti-social behavior; peer relations and
Kelly A. Nelson, M.Sc. is a Research Associate/genetics laboratory
technician at the Avera Institute for Human Genetics. She received an
M.Sc. in Biological Sciences, from South Dakota State University in
Brookings, USA. Her major research interests include:
pharmacogenomics, gene expression, and human microbiome.
Pol A. C. van Lier, Ph.D. is a Professor at VU University,
Amsterdam, the Netherlands. He received his doctorate in Medicine
from Erasmus University Rotterdam, the Netherlands. His major
research interests include: peer relations; emotional and behavioral
development; behavioral and emotional problems; psychopathology;
delinquency; longitudinal research, prevention.
Anacker , K. , Enge , S. , Reif , A. , Lesch , K. , & Strobel , A. ( 2013 ). Dopamine D4 receptor gene variation impacts self-reported altruism . Molecular Psychiatry , 18 , 402 - 403 .
Bakermans-Kranenburg , M. J. , & van IJzendoorn, M. H. ( 2011 ). Differential susceptibility to rearing environment depending on dopamine-related genes: New evidence and a meta-analysis . Development and Psychopathology , 23 , 39 - 52 .
Barrish , H. H. , Saunders , M. , & Wolf , M. M. ( 1969 ). Good Behavior Game: Effects of individual contingencies for group consequences on disruptive behavior in a classroom . Journal of Applied Behavior Analysis , 2 , 119 - 124 .
Becker , A. , Woerner , W. , Hasselhorn , M. , Banaschewski , T. , & Rothenberger , A. ( 2004 ). Validation of the parent and teacher SDQ in a clinical sample . European Child & Adolescent Psychiatry , 13 , ii11 - ii16 .
Belsky , J. ( 1997 ). Variation in susceptibility to environmental influence: An evolutionary argument . Psychological Inquiry , 8 , 182 - 186 .
Belsky , J. , & Hartman , S. ( 2014 ). Gene-environment interaction in evolutionary perspective: differential susceptibility to environmental influences . World Psychiatry , 13 , 87 - 89 .
Bierman , K. L. , & Erath , S. A. ( 2006 ). Promoting social competence in early childhood: Classroom curricula and social skills coaching programs . In K. McCartney & D. Phillips (Eds.), Blackwell handbook of early childhood development (pp. 595 - 615 ). Oxford: Blackwell Publishing Ltd.
Boor, K., Ronai , Z. , Nemoda , Z. , Gaszner , P. , Sasvari-Szekely, M. , Guttman , A. , & Kalasz, H. ( 2002 ). Noninvasive genotyping of dopamine receptor D4 (DRD4) using nanograms of DNA from substance-dependent patients . Current Medicinal Chemistry , 9 , 793 - 797 .
Chang , F.-M. , Kidd , J. R. , Livak , K. J. , Pakstis , A. J. , & Kidd , K. K. ( 1996 ). The world-wide distribution of allele frequencies at the human dopamine D4 receptor locus . Human Genetics , 98 , 91 - 101 .
Coie , J. D. , Dodge , K. A. , & Coppotelli , H. ( 1982 ). Dimensions and types of social status: A cross-age perspective . Developmental Psychology , 18 , 557 - 570 .
Crick , N. R. , & Grotpeter , J. K. ( 1996 ). Children's treatment by peers: Victims of relational and overt aggression . Development and Psychopathology , 8 , 367 - 380 .
DiLalla , L. F. , Elam , K. K. , & Smolen , A. ( 2009 ). Genetic and geneenvironment interaction effects on preschoolers' social behaviors . Developmental Psychobiology , 51 , 451 - 464 .
Dolan , L. J. , Kellam , S. G. , Werthamer , L. , & Kellam , S. G . (Eds.). ( 1989 ). The good behavior game manual . Baltimore , MD : The John Hopkins Prevention Research Center.
Duncan , L. E. , & Keller , M. C. ( 2011 ). A critical review of the first 10 years of candidate gene-by-environment interaction research in psychiatry . The American Journal of Psychiatry , 168 , 1041 - 1049 .
Ellis , B. J. , & Boyce , W. T. ( 2011 ). Differential susceptibility to the environment: Toward an understanding of sensitivity to developmental experiences and context . Development and Psychopathology , 23 , 1 - 5 .
Erasmus , M. C. ( 2000 ). Problem behavior at school interview . Rotterdam: Department of Child and Adolescent Psychiatry.
Fan , J. , Fossella , J. , Sommer , T. , Wu , Y. , & Posner , M. I. ( 2003 ). Mapping the genetic variation of executive attention onto brain activity . Proceedings of the National Academy of Sciences , 100 , 7406 - 7411 .
Forbes , E. , Brown , S. , Kimak , M. , Ferrell , R. , Manuck , S. , & Hariri , A. ( 2009 ). Genetic variation in components of dopamine neurotransmission impacts ventral striatal reactivity associated with impulsivity . Molecular psychiatry , 14 , 60 - 70 .
Fossella , J. , Sommer , T. , Fan , J. , Wu , Y. , Swanson , J. M. , Pfaff , D. W. , & Posner , M. I. ( 2002 ). Assessing the molecular genetics of attention networks . BMC Neuroscience , 3 , 1 - 11 .
Froehlich , T. E. , Lanphear , B. P. , Dietrich , K. N. , Cory-Slechta , D. A. , Wang , N. , & Kahn , R. S. ( 2007 ). Interactive effects of a DRD4 polymorphism, lead, and sex on executive functions in children . Biological Psychiatry , 62 , 243 - 249 .
Gehricke , J.-G. , Swanson , J. M. , Duong , S. , Nguyen , J. , Wigal , T. L. , Fallon , J. , & Moyzis , R. K. ( 2015 ). Increased brain activity to unpleasant stimuli in individuals with the 7R allele of the DRD4 gene . Psychiatry Research: Neuroimaging , 231 , 58 - 63 .
Gooren , E. M. , van Lier , P. A. , Stegge , H. , Terwogt , M. M. , & Koot , H. M. ( 2011 ). The development of conduct problems and depressive symptoms in early elementary school children: The role of peer rejection . Journal of Clinical Child & Adolescent Psychology , 40 , 245 - 253 .
Grady , D. L. , Thanos , P. K. , Corrada , M. M. , Barnett , J. C. , Ciobanu , V. , Shustarovich , D. , & Rubinstein , M. ( 2013 ). DRD4 genotype predicts longevity in mouse and human . The Journal of Neuroscience , 33 , 286 - 291 .
Hart , E. L. , Lahey , B. B. , Loeber , R. , & Hanson , K. S. ( 1994 ). Criterion validity of informants in the diagnosis of disruptive behavior disorders in children: A preliminary study . Journal of Consulting and Clinical Psychology , 62 , 410 - 414 .
Hawley , P. H. ( 1999 ). The ontogenesis of social dominance: A strategy based evolutionary perspective . Developmental Review , 19 , 97 - 132 .
Holmes , J. , Payton , A. , Barrett , J. , Harrington , R. , McGuffin , P. , Owen , M. , & Kirley , A. ( 2002 ). Association of DRD4 in children with ADHD and comorbid conduct problems . American Journal of Medical Genetics , 114 , 150 - 153 .
Hu , L. , & Bentler , P. M. ( 1998 ). Fit indices in covariance structure modeling: Sensitivity to underparameterized model misspecification . Psychological Methods , 3 , 424 - 453 .
Hughes , J. N. , Zhang , D. , & Hill , C. R. ( 2006 ). Peer assessments of normative and individual teacher-student support predict social acceptance and engagement among low-achieving children . Journal of School Psychology , 43 , 447 - 463 .
Insel , T. R. ( 2003 ). Is social attachment an addictive disorder? Physiology & Behavior, 79 , 351 - 357 .
International Labour Organization . ( 1987a ). International standard classification of occupations: ISCO-88. Gene`ve: ILO.
International Labour Organization . ( 1987b ). Revision of the international standard classification of occupations, part I: Background, principles and draft resolution . Gene`ve: ILO.
Jiang , Y. , Chew , S. H. , & Ebstein , R. P. ( 2013 ). The role of D4 receptor gene exon III polymorphisms in shaping human altruism and prosocial behavior . Frontiers in Human Neuroscience , 7 , 195 .
Joreskog, K. G. ( 1970 ). A general method for analysis of covariance structures . Biometrika , 57 , 239 - 251 .
Keith , T. Z. ( 2006 ). Multiple regression and beyond . Boston: Allyn and Bacon.
Keith , T. Z. , & Cool , V. A. ( 1992 ). Testing models of school learning: Effects of quality of instruction, motivation, academic coursework, and homework on academic achievement . School Psychology Quarterly, 7 , 207 - 226 .
Kirley , A. , Lowe , N. , Mullins , C. , McCarron , M. , Daly , G. , Waldman , I. , & Hawi , Z. ( 2004 ). Phenotype studies of the DRD4 gene polymorphisms in ADHD: Association with oppositional defiant disorder and positive family history . American Journal of Medical Genetics Part B: Neuropsychiatric Genetics , 131 , 38 - 42 .
Kretschmer , T. , Dijkstra , J. K. , Ormel , J. , Verhulst , F. C. , & Veenstra , R. ( 2013 ). Dopamine receptor D4 gene moderates the effect of positive and negative peer experiences on later delinquency: The Tracking Adolescents' Individual Lives Survey study . Development and Psychopathology , 25 , 1107 - 1117 .
Kupersmidt , J. B. , Burchinal , M. , & Patterson , C. J. ( 1995 ). Developmental patterns of childhood peer relations as predictors of externalizing behavior problems . Development and Psychopathology , 7 , 825 - 843 .
Kusche, C. A., & Greenberg , M. T. ( 1994 ). The PATHS curriculum . Seattle: Developmental Research and Programs.
Ladd , G. W. ( 2006 ). Peer rejection, aggressive or withdrawn behavior, and psychological maladjustment from ages 5 to 12: An examination of four predictive models . Child Development , 77 , 822 - 846 .
Leflot , G. , van Lier , P. A. , Verschueren , K. , Onghena , P. , & Colpin , H. ( 2011 ). Transactional associations among teacher support, peer social preference, and child externalizing behavior: A fourwave longitudinal study . Journal of Clinical Child and Adolescent Psychology , 40 , 87 - 99 .
Marsh , H. W. , Hau , K. T. , & Wen , Z. ( 2004 ). In search of golden rules: comment on hypothesis-testing approaches to setting cutoff values for fit indexes and dangers in overgeneralizing Hu and Bentler's findings . Structural Equation Modeling: A Multidisciplinary Journal , 11 , 320 - 341 .
Menting , B. , Van Lier , P. A. C. , & Koot , H. M. ( 2011 ). Language skills, peer rejection, and the development of externalizing behavior from kindergarten to fourth grade . Journal of Child Psychology and Psychiatry , 52 , 72 - 79 .
Moffitt , T. E. , Caspi , A. , Rutter , M. , & Silva , P. A. ( 2001 ). Sex differences in antisocial behavior: Conduct disorder, delinquency, and violence in the Dunedin Longitudinal Study . New York : Cambridge University Press.
Muthen, L. K. , & Muthen, B. O. ( 1998 - 2011 ). Mplus user's guide: Statistical analysis with latent variables (6th ed.). Los Angeles, CA: Muthen & Muthen.
Muthen, L. K. , & Muthen, B. O. ( 2002 ). How to use a Monte Carlo study to decide on sample size and determine power . Structural Equation Modeling , 9 , 599 - 620 .
Oak , J. N. , Oldenhof , J. , & Van Tol , H. H. M. ( 2000 ). The dopamine D4 receptor: One decade of research . European Journal of Pharmacology , 405 , 303 - 327 .
Olthof , T. , Goossens , F. A. , Vermande , M. M. , Aleva , E. A. , & van der Meulen, M. ( 2011 ). Bullying as strategic behavior: Relations with desired and acquired dominance in the peer group . Journal of School Psychology , 49 , 339 - 359 .
Parker , J. G. , & Asher , S. R. ( 1993 ). Friendship and friendship quality in middle childhood: Links with peer group acceptance and feelings of loneliness and social dissatisfaction . Developmental Psychology , 29 , 611 - 621 .
Parker , J. G. , Rubin , K. H. , Erath , S. A. , Wojslawowicz , J. C. , & Buskirk , A. A. ( 2006 ). Peer relationships, child development and adjustment: A developmental psychopathology perspective . In D. Cichetti & D. J. Cohen (Eds.), Developmental psychopathology, vol 1: Theory and method (2nd ed. , pp. 419 - 493 ). Hoboken, NJ: Wiley.
Pellegrini , A. D. , & Long , J. D. ( 2003 ). A sexual selection theory longitudinal analysis of sexual segregation and integration in early adolescence . Journal of Experimental Child Psychology , 85 , 257 - 278 .
Posner , M. I. , & Rothbart , M. K. ( 2009 ). Toward a physical basis of attention and self-regulation . Physics of Life Reviews , 6 , 103 - 120 .
Propper , C. , Willoughby , M. , Halpern , C. , Carbone , M. , & Cox , M. ( 2007 ). Parenting quality, DRD4, and the prediction of externalizing and internalizing behaviors in early childhood . Developmental Psychobiology , 49 , 619 - 632 .
Reijntjes , A. , Vermande , M. , Goossens , F. A. , Olthof , T. , van de Schoot , R. , Aleva , L. , & van der Meulen, M. ( 2013 ). Developmental trajectories of bullying and social dominance in youth . Child Abuse and Neglect , 37 , 224 - 234 .
Rubin , H. K. , Bukowski , W. M. , & Parker , J. G. ( 2006 ). Peer interactions, relationships, and groups . In N. Eisenberg (Ed.), Handbook of Child Psychology (Vol. 3 , pp. 571 - 645 )., Social, emotional, and personality development Hoboken: Wiley, NJ.
Rutter , M. , Pickles , A. , Murray , R. , & Eaves , L. ( 2001 ). Testing hypotheses on specific environmental causal effects on behavior . Psychological Bulletin , 127 , 291 - 324 .
Satorra , A. ( 2000 ). Scaled and adjusted restricted tests in multisample analysis of moment structures . In R. D. H. Heijmans , D. S. G. Pollock , & A. Satorra (Eds.), Innovations in multivariate statistical analysis . A Festschrift for Heinz Neudecker (pp. 233 - 247 ). London: Kluwer.
Schmidt , L. A. , Fox , N. A. , Rubin , K. H. , Hu , S. , & Hamer , D. H. ( 2002 ). Molecular genetics of shyness and aggression in preschoolers . Personality and Individual Differences , 33 , 227 - 238 .
Schoots , O. , & Van Tol , H. ( 2003 ). The human dopamine D4 receptor repeat sequences modulate expression . The Pharmacogenomics Journal , 3 , 343 - 348 .
Sheese , B. E. , Voelker , P. M. , Rothbart , M. K. , & Posner , M. I. ( 2007 ). Parenting quality interacts with genetic variation in dopamine receptor D4 to influence temperament in early childhood . Development and Psychopathology , 19 , 1039 - 1046 .
Sroufe , L. A. , Egeland , B. , Carlson , E. A. , & Collins , W. A. ( 2009 ). The development of the person: The Minnesota Study of risk and adaptation from birth to adulthood . New York : Guilford Press.
Statistics Netherlands . ( 2001 ). Standaard Beroepenclassificatie 1992 ( SBC 1992), verbeterde editie 2001 [Eng: Working Population Classifications of Occupations Scheme] . Retrieved December 6th 2014 , from http://www.cbs.nl/nl-NL/menu/methoden/classifi caties/overzicht/sbc/default.htm
Statistics Netherlands . ( 2012 ). Statline. Voorburg/Heerlen: The Netherlands Central Bureau of Statistics. Retrieved from http:// statline.cbs.nl/StatWeb/publication/?VW=T&DM=SLNL&PA= 71812ned&LA=NL
Trainor , B. C. ( 2011 ). Stress responses and the mesolimbic dopamine system: social contexts and sex differences . Hormones and Behavior , 60 , 457 - 469 .
Twenge , J. M. , Baumeister , R. F. , DeWall , C. N. , Ciarocco , N. J. , & Bartels , J. M. ( 2007 ). Social exclusion decreases prosocial behavior . Journal of Personality and Social Psychology , 92 , 56 - 66 .
Van Lier , P. A. C. , & Koot , H. M. ( 2010 ). Developmental cascades of peer relations and symptoms of externalizing and internalizing problems from kindergarten to fourth-grade elementary school . Development and Psychopathology , 22 , 569 - 582 .
Wentzel , K. R . ( 2014 ). Prosocial behavior and peer relations in adolescence . In L. M. Padilla-Walker & C. Gustavo (Eds.), Prosocial development: A multidimensional approach (pp . 178 - 200 ). New York : Oxford University Press.
Wentzel , K. R ., & McNamara , C. C. ( 1999 ). Interpersonal relationships, emotional distress, and prosocial behavior in middle school . The Journal of Early Adolescence , 19 , 114 - 125 .
Williams , R. L. ( 2000 ). A note on robust variance estimation for cluster-correlated data . Biometrics , 56 , 64 - 646 .
Witvliet , M. , Van Lier , P. A. C. , Cuijpers , P. , & Koot , H. M. ( 2009 ). Testing links between childhood positive peer relations and externalizing outcomes through a randomized controlled intervention study . Journal of Consulting and Clinical Psychology , 77 , 905 - 915 .