A preview of this full-text is provided by American Psychological Association.
Content available from Developmental Psychology
This content is subject to copyright. Terms and conditions apply.
Change in Parenting, Change in Student–Teacher Relationships, and
Oxytocin Receptor Gene (OXTR): Testing a Gene-⫻-Environment (G⫻E)
Hypothesis in Two Samples
Beate Wold Hygen
Norwegian University of Science and Technology and NTNU
Social Research, Trondheim, Norway
Jay Belsky and Zhi Li
University of California, Davis
Frode Stenseng
Norwegian University of Science and Technology Ismail Cuneyt Güzey
Norwegian University of Science and Technology and St. Olav
University Hospital
Lars Wichstrøm
Norwegian University of Science and Technology and NTNU Social Research, Trondheim, Norway
Prior research suggests that parenting affects children’s relationships, including those with teachers,
although there is variation across individuals in such effects. Given evidence suggesting that oxytocin
may be particularly important for the quality of social relationships, we tested the hypotheses (a) that
change in parenting from 4 to 6 years of age influences and predicts change in the student–teacher
relationship from 6 to 8 years of age and (b) that this effect is moderated by a polymorphism related to
the child’s oxytocin receptor gene (OXTR), rs53576. In 2 studies, participants included, respectively, 652
socioeconomically diverse Norwegian children from a community sample (50.8% male; mean age of
54.9 months at first assessment) and 559 such children from 8 different U.S. locales (49.0% male;
approximately 54 months at the first assessment). Norwegian results showed that change in parenting
predicted change in student–teacher relationships, but only in the case of children homozygous for the
A allele of rs53576 and in a manner consistent with differential-susceptibility theory: for AA carriers,
when parenting changed for the worse, so did children’s relationship with teachers, whereas when
parenting changed for the better, the teacher–child relationships improved accordingly. Such G⫻E
findings could not be replicated in the American sample. Results are discussed in terms of 2 contrasting
models of Person-⫻-Environment interaction (differential susceptibility and diathesis stress) and poten-
tial reasons for failure to replicate.
Keywords: diathesis stress, differential susceptibility, OXTR rs53576, parenting, student–teacher
relationship
Evidence indicates that the quality of student–teacher relation-
ships has wide-ranging implications for children’s functioning and
development (Hamre & Pianta, 2001), including behavior prob-
lems (Silver, Measelle, Armstrong, & Essex, 2005), school en-
gagement and academic achievement (Ladd, Birch, & Buhs, 1999;
Roorda, Koomen, Spilt, & Oort, 2011), peer relations (De Laet et
al., 2014), and social competence (Skalická et al., 2015). Such
findings underscore the importance of illuminating the determi-
nants of student–child-relationship quality. Available evidence
suggests that positive parent–child relations may lay the ground-
work for children’s positive relations with their teachers (Booth,
Kelly, Spieker, & Zuckerman, 2003;DeMulder, Denham,
Schmidt, & Mitchell, 2000); in the present study, we evaluated the
effects of parenting in two separate samples, one Norwegian
Editor’s Note. Marc Bornstein served as the action editor for this article. —JSE
This article was published Online First May 4, 2017.
Beate Wold Hygen, Department of Psychology, Norwegian University
of Science and Technology, and NTNU Social Research, Trondheim,
Norway; Jay Belsky and Zhi Li, Department of Human
Ecology, University of California, Davis; Frode Stenseng, Regional Centre
for Child and Youth Mental Health and Child Welfare, Norwegian
University of Science and Technology; Ismail Cuneyt Güzey, Department
of Neuroscience, Norwegian University of Science and Technology, and
Department of Psychiatry, St. Olav University Hospital; Lars Wichstrøm,
Department of Psychology, Norwegian University of Science and Tech-
nology, and NTNU Social Research, Trondheim, Norway.
This research was funded by Grants 191144/V50 and 228685/H10 from
the Research Council of Norway.
Correspondence concerning this article should be addressed to Beate
Wold Hygen, Department of Psychology, Norwegian University of Science
and Technology, N-7491 Trondheim, Norway. E-mail: beate.hygen@
samfunn.ntnu.no
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Developmental Psychology © 2017 American Psychological Association
2017, Vol. 53, No. 7, 1300–1315 0012-1649/17/$12.00 http://dx.doi.org/10.1037/dev0000333
1300
(Study 1) and the other from the United States (Study 2). Notably,
we focused on change in parenting predicting change in teacher–
child relationships because evidence of change in predictors fore-
casting change in outcomes is more indicative of causal processes
in observational research like that which is the focus of this report
than simple cross-time predictor-outcome associations. Thus, we
predicted that parenting that becomes more positive over time
would lead to increasingly positive change in the teacher–child
relationship, whereas parenting that becomes increasingly negative
would result in deterioration in teacher–child relations over time.
Given the increasing evidence that environmental effects, in-
cluding those involving parenting, vary according to the child’s
dispositional qualities (Belsky, Bakermans-Kranenburg, & Van
Ijzendoorn, 2007;Belsky & Pluess, 2009,2013;Ellis, Boyce,
Belsky, Bakermans-Kranenburg, & van Ijzendoorn, 2011), often
as a result of their genotype (Bakermans-Kranenburg & van Ijzen-
doorn, 2011), we investigated genetic moderation of parenting
effects on teacher–child relations in the research reported herein.
Based on research showing that oxytocin (Meyer-Lindenberg,
Domes, Kirsch, & Heinrichs, 2011) and variations in the oxytocin
receptor gene (OXTR) are involved in social interaction and soci-
ality (Feldman, Monakhov, Pratt, & Ebstein, 2016;Li et al., 2015),
we specifically evaluated genetic moderation via this polymor-
phism. More specifically, we evaluated whether effects of change
in parenting from 4 to 6 years of age on change in the student–
teacher relationship from 6 to 8 years of age vary as a function of
OXTR rs53576. We also evaluated whether such genetic modera-
tion of parenting is consistent with the differential susceptibility or
diathesis stress model of Person-⫻-Environment interaction. Be-
cause research has repeatedly shown that children’s behavior also
affects parenting (e.g., Reitz, Dekovic, Meijer, & Engels, 2006),
we adjusted for children’s externalizing and internalizing problems
and their observed behavior toward their parents.
Determinants of the Student–Teacher Relationship
Children vary considerably in their relationships with their
teachers. Although a range of background factors may promote or
hinder the development of rewarding student–teacher relation-
ships, such as child gender (Jerome, Hamre, & Pianta, 2009),
disabilities (Murray & Greenberg, 2001), and temperament (Ru-
dasill, Reio, Stipanovic, & Taylor, 2010), as well as family income
(Wyrick & Rudasill, 2009), these factors are not likely targets of
interventions to improve this relationship. The parent–child rela-
tionship, however, is more appropriate given evidence that it
affects children’s relationships with teachers (Howes & Hamilton,
1992;O’Connor & McCartney, 2006) and is subject to change
(Wyatt Kaminski, Valle, Filene, & Boyle, 2008).
Attachment theory (Bowlby, 1973) and social learning theory
(Bandura, 1977) offer some insight into why parent–child rela-
tions may influence the student–teacher relationship.
According to attachment theory (Bowlby, 1973), repeated ex-
posure to sensitive parenting fosters an internal working model in
which others, including teachers, are benign if not benevolent and
are predictable and in which student–teacher interactions are pre-
dicted to be rewarding. Insensitive or even frightening parenting,
however, should promote internal working models in which others
cannot be trusted and social exchanges will be unpleasant and not
mutually beneficial. From the perspective of attachment theory,
one would expect sensitive (and security-inducing) parenting to
lay the groundwork for a positive student–teacher relationship,
with the opposite true of insensitive parenting. In line with this
view, securely attached children have been found to develop
positive relationships with their teachers, whereas those with in-
secure attachments are at heightened risk of developing problem-
atic ones (DeMulder et al., 2000;O’Connor & McCartney, 2006;
Rydell, Bohlin, & Thorell, 2005). It must be acknowledged, how-
ever, that such theoretically anticipated associations have not
emerged in all relevant research (Goossens & van IJzendoorn,
1990).
From the standpoint of social learning theory (Bandura, 1977),
which emphasizes learning processes involving modeling and re-
inforcement (rather than internal working models), it is expected
that children whose parents behave in supportive ways and praise
positive social behavior will themselves behave in ways that
promote positive teacher–child relationships. In contrast, parents
who treat their children harshly or neglectfully may model aggres-
sive or unresponsive behavior that the child learns and generalizes
to other adults, thereby adversely affecting relationships with
teachers. Moreover, insensitive and especially harsh parenting is
expected to instigate coercive parent–child interaction processes
that (inadvertently) reinforce negative child behavior, which can
itself undermine other relationships (Patterson, Debaryshe, &
Ramsey, 1989), including teacher–child ones. Not surprisingly,
social learning-related research indicates that negative parenting
(e.g., criticizing, making negative statements, and/or rejecting) is
associated with child behavioral problems (Webster-Stratton &
Hammond, 1999), whereas positive parenting (e.g., responsive-
ness, praise) promotes social skills and reduces behavioral prob-
lems (Webster-Stratton, Reid, & Hammond, 2001). Notably, there
is also evidence that children with such behavioral problems
develop poorer relationships with their teachers (Henricsson &
Rydell, 2004;Murray & Zvoch, 2011).
Despite emphasizing different mechanisms responsible to ac-
count for why there should be an association between the quality
and functioning of parent–child and teacher–student relationships,
both theoretical frameworks just considered, attachment theory
and social learning theory, share the view that experiences in one
relationship, especially that with parents, influence experiences in
other relationships, including student–teacher ones. It is for this
reason that we predicted that the development of parent–child
relationships, operationalized in terms of change over time from 4
to 6 years, would predict the development of teacher–child rela-
tionships from 6 to 8 years of age.
Variation in Environmental Effects
Even if the processes under consideration prove generally op-
erative, parenting may not affect all children equally. Indeed, it has
long been assumed that some children are more susceptible to the
negative effects of problematic parenting or other conditions of
adversity than other children because of their own personal char-
acteristics, be they genetic, temperamental, and/or physiological in
character. In fact, this assumption is central to the long-standing
and widely embraced diathesis-stress model of Person-⫻-
Environment interaction (Zuckerman, 1999).
The alternative and more recently introduced differential-
susceptibility model extends thinking about variation in response
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1301
OXTR, PARENTING, AND STUDENT–TEACHER RELATIONSHIPS
to environmental conditions by stipulating that the very children
most likely to be adversely affected by problematic parenting will
be most likely to benefit from supportive care (Belsky, 2005;
Belsky et al., 2007;Belsky & Pluess, 2009,2013;Ellis, Boyce,
Belsky, Bakermans-Kranenburg, & van IJzendoorn, 2011). Ac-
cording to this evolutionary inspired perspective, children with
certain personal attributes that make them especially sensitive to
environmental conditions should develop both better and worse
teacher–child relationships, depending on the quality of their par-
enting experience, than children whose personal attributes make
them less sensitive to such experiences and exposures. In the
present report, we evaluated whether the diathesis stress or differ-
ential susceptibility model better accounts for variation in teacher–
child relationships as a function of parenting—in two separate
samples. To this end, we focused on one particular genetic poly-
morphism that has been found to be influential in research on
relationships and the relational process (Carter, 2014), OXTR
rs53576.
Oxytocin Receptor Gene
Oxytocin (OXT) has been related to a variety of socioemotional
behaviors, including anxiety, aggression, and social recognition
(Lee, Macbeth, Pagani, & Young, 2009). A review of relevant
evidence by Meyer-Lindenberg and associates (2011) revealed that
OXT affects how individuals interpret social cues, thereby influ-
encing social cognition and behavior, including attachment, social
recognition, and social exploration.
OXT functions as a hormone and a neurotransmitter, expressing
its actions via the OXTR (Carter, 2014). The genetically polymor-
phic OXTR is located on chromosome 3p25, and one of these
polymorphisms (rs53576) can influence social behavior (Krueger
et al., 2012). The single nucleotide polymorphism (SNP) rs53576
is an intronic variation, caused by a guanine (G) to adenine (A)
replacement (G and A allele). Although our knowledge regarding
the specific functionality of this variation is limited, the rs53576
genotype has been shown to impact brain structure and function-
ality such as hypothalamus gray matter and amygdala volume.
When exposed to emotionally salient cues, individuals homozy-
gous for the G allele showed heightened task-related amygdala
activation. Carriers of the A allele evinced the lowest activation
and increased coupling of the hypothalamus and amygdala, in
which the connectivity between the two was mirrored by the dose
of the minor allele (Tost et al., 2010).
Even though little is known about the specific function of OXTR
SNP rs53576, it has been the subject of numerous genotype-
phenotype association studies, some of which might link it, indi-
rectly, to student–teacher relationships, including support seeking
(Chen et al., 2011), trust (Krueger et al., 2012), and even parental
sensitivity (Bakermans-Kranenburg & van Ijzendoorn, 2008).
More specifically, some findings indicate that carriers of the A
allele (i.e., AA or AG) are, relative to G homozygotes, less
empathic and more reactive to stress (Rodrigues, Saslow, Garcia,
John, & Keltner, 2009), exhibit lower trust-related behaviors
(Krueger et al., 2012), and prove to be less sensitive parents
(Bakermans-Kranenburg & van Ijzendoorn, 2008;Klahr, Klump,
& Burt, 2015). Moreover, at least one of two meta-analyses has
documented associations between rs53576 and sociality (e.g., ex-
traversion, social skills, and prosocial behavior). Specifically, Li et
al. (2015) observed that although G homozygotes did not differ
from A carriers with regard to the quality of their close relation-
ships, G homozygotes did score higher on general sociality—a
result that Bakermans-Kranenburg and van IJzendoorn’s (2014)
meta-analysis did not discern. Such inconsistency across investi-
gations, including even meta-analyses, is not unusual in research
on genotype-phenotype associations.
Gene-ⴛ-Environment Interaction
Indeed, evidence such as that reviewed above highlighting in-
consistency in research on the main effects of parenting and of
candidate genes stimulated Caspi et al.’s (2002,2003) pioneering
research on Gene-⫻-Environment (G⫻E) interactions. Over the
past decade, numerous reports have appeared documenting such
G⫻E interplay (e.g., Babineau et al., 2015;Bresin, Sima Finy, &
Verona, 2013;Brown et al., 2013;Hygen et al., 2015;Hygen,
Guzey, Belsky, Berg-Nielsen, & Wichstrøm, 2014); however, the
G⫻E field has been subjected to controversy and critique (e.g.,
Duncan & Keller, 2011). Nevertheless, prior research indicates
that the OXTR SNP rs53576 interacts with various environmental
factors in predicting diverse behavioral and psychological pheno-
types. Most relevant to the present inquiry is research showing that
rs53576 moderates the effect of maltreatment on regulatory be-
havior and disorganized attachment (Bradley et al., 2011), percep-
tion of social support (Hostinar, Cicchetti, & Rogosch, 2014), and
borderline personality features (Cicchetti, Rogosch, Hecht, Crick,
& Hetzel, 2014). Moreover, rs53576 has been found to moderate
the effect of unsupportive parenting on children‘s coping styles
(McInnis, McQuaid, Matheson, & Anisman, 2015) and the effect
of maternal depression on youths’ depressive symptoms (Thomp-
son, Hammen, Starr, & Najman, 2014). Such G⫻E findings in-
volving both parenting and rs53576 motivated the current inves-
tigation of the interaction of these two predictors in forecasting
change in the student–teacher relationship.
As it turns out, the just-cited G⫻E studies involving rs53576 do
not consistently indicate which allelic subgroup or subgroups
prove more and less sensitive to environmental effects. In some
studies it is A carriers who emerge as most vulnerable to the
negative effects of adversity (Thompson et al., 2014), whereas in
others it is G carriers who prove to be at environmental risk
(Bradley et al., 2011;Hostinar et al., 2014). Moreover, some
research indicates that the rs53576 moderation of environmental
effects varies by gender, with GG boys and A-carrier girls proving
more susceptible to the anticipated negative effects of adversity
(Cicchetti et al., 2014). This inconsistency may be the result of the
diverse predictors and outcomes examined in different studies or other
methodological factors that vary across inquiries. Nevertheless, the
fact that no allelic subgroup stands out as consistently susceptible or
vulnerable directly informs our approach to statistical analysis, given
our concern for alternative models of G⫻E interaction.
When prior evidence indicates that particular allelic subgroups
proves more or less susceptible to environmental effects in a more,
rather than less, consistent manner (e.g., short allele carriers of the
5-HTTLPR polymorphism), it is appropriate to use Widaman et al.’s
(2012;Belsky, Pluess, & Widaman, 2013) competitive model-fitting
approach when seeking to determine whether a differential-
susceptibility or diathesis-stress framework best fits the data. This
approach involves foregoing traditional exploratory tests of G⫻E
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1302 HYGEN, BELSKY, LI, STENSENG, GU
¨ZEY, AND WICHSTRØM
interaction using regression models and moving directly to model
fitting. However, in situations such as the present in which prior
research does not provide a strong basis for predicting which allelic
subgroup will prove most environmentally sensitive, exclusive reli-
ance on competitive model fitting is not appropriate. Thus, in the
current inquiry, we first used the traditional regression approach to
identify which allelic subgroups; if any, prove more and less sensitive
to the effects of change in parenting on change in the student–teacher
relationship before using model fitting to evaluate whether any de-
tected G⫻E interaction proves more consistent with differential sus-
ceptibility or diathesis stress thinking.
Recent years have witnessed an outpouring of concern regarding
the replicability of scientific findings (Jasny, Chin, Chong, & Vign-
ieri, 2011;Ryan, 2011). Perhaps nowhere has this issue emerged so
forcefully in the human behavioral sciences as in research involving
measured genes. Much of this concern originated in this disappoint-
ment that arose when initial and significant genotype-phenotype as-
sociations could not be detected in subsequent studies (Hamer, 2002;
Insel & Collins, 2003). When attention turned to G⫻E interaction,
following the publication of pioneering work in this area by Caspi and
associates (Caspi et al., 2002,2003), it was not long before issues of
replication arose here as well. The situation no doubt became espe-
cially confusing when different meta-analyses of the supposedly same
G⫻E interaction yielded radically different conclusions. Whereas
Risch et al. (2009) claimed that the interaction between the serotonin
transporter, 5-HTTLPR, and stressful life events did not reliably
predict depression in women, Karg and collegues (2011) drew the
opposite conclusion in their meta-analyses, so did Uher and McGuffin
(2008,2010) in a review of human G⫻E findings reported over the
first decade of such work (see also Kaufman, Gelernter, Kaffman,
Caspi, & Moffitt, 2010;Rutter, 2010). In so doing, the latter scholars
highlighted the need to pay special attention to the issue of measure-
ment (see also Caspi, Hariri, Holmes, Uher, & Moffitt, 2010). More
recently, Duncan and Keller (2011) concluded that most findings
were likely to be false positive (see also Moore & Thoemmes, 2016).
In light of the concerns raised and observations made about G⫻E
findings in particular, we conducted a second study using data from 8
of 10 sites (that collected genetic data) involved in a large scale,
American study, National Institute of Child Health and Human De-
velopment (NICHD). NICHD Study of Early Child Care and Youth
Development (NICHD Early Child Care Research Network, 2005), to
determine whether it was possible to replicate the G⫻E finding which
emerged in Study 1 conducted in Norway. Important to appreciate is
that measurement across the two studies were not identical, even if
highly similar, and that policies and practices of schools in the two
countries cannot be presumed to be the same.
Study 1
Method
Participants and recruitment. The Trondheim Early Secure
Study (TESS) is a prospective study conducted in the city of
Trondheim, Norway. The study population comprised the 2003
and 2004 birth cohorts and their parents. Parents and their chil-
d(ren) were invited to participate in the study through an invitation
letter sent to their home. The letter also included the Strengths and
Difficulties Questionnaire (SDQ) version 4–16 (Crone, Vogels,
Hoekstra, Treffers, & Reijneveld, 2008), which was used for
screening. Completed SDQs were returned by 3,358 of 3,456,
families (97.2%) at the Well Child Clinics (when the child at-
tended a routine health check for 4-year-olds). Only parents with
adequate proficiency in Norwegian were included in the study.
Eligible parents (n⫽3,016) acquired information of the study
through procedures approved by the Regional Committee for Med-
ical and Health Research Ethics. Parents of 2,475 children pro-
vided written consent to participate.
Variability, and thus statistical power, was increased by over-
sampling children with mental health problems. To accomplish
this task, the SDQ total scores were divided into four strata (0–4,
5–8, 9–11, and 12– 40), and defined proportions of children in
each stratum (.37, .48, .70, and .89, respectively) were selected to
participate using a random number generator. The selection prob-
abilities increased with increasing SDQ scores. Of the 1,250 par-
ents drawn to participate, we tested 936 (74.9%). Data were
collected at the Norwegian University of Science and Technology,
where the children met together with one parent, usually the
mother (84.4%). Trained personnel performed all interviews and
observations. Altogether, 762 children participated in the
follow-up assessment 2 years later (T2), whereas 699 children
participated 2 years after that (T3). The day-care teacher (at age 4)
who knew the child best was instructed to fill in a questionnaire
measuring the student–teacher relationship, and 989 teachers com-
pleted the scheme at age 4, whereas 788 and 607 teachers com-
pleted the scheme in 1st and 3rd grade, respectively. At age 4, the
day-care teachers had known the child for an average of 1.7 years
(SD ⫽.9). At first grade (age 6), the teacher had known the child
for .7 years (SD ⫽.4) and for 2.1 years (SD ⫽1.4) in third grade.
Among these children, 652 were successfully genotyped for the
OXTR rs53576 polymorphism and thus formed the basis of the
analysis sample.
There were 332 (49.2%) females in the present sample; most
lived with their biological parents, most of whom were married
(see Table 1). Those who had a poor teacher–child relationship at
Time 1 were less likely to be genotyped (T1; Odds ratio [OR]⫽
0.98, 95% confidence interval [CI] ⫽0.96–1.00, p⫽.03). Posi-
tive parenting at Time 1 made it less likely to drop out at Time 2
(OR ⫽0.98, [0.96–1.00], p⫽.03) and Time 3 (OR ⫽.95,
[.91–1.00], p⫽.03). Hence, because of the initial values of the
outcome variable being predictive of attrition, we suspect that data
were missing not at random (MNAR), and we performed the
analyses accordingly—see Statistical Analysis.
Measures.
Student–teacher relationship. The student–teacher relation-
ship was measured using the Student–Teacher Relationship Scale
(STRS; Pianta, 2001) at 4, 6, and 8 years of age applying a 5-point
Likert-type scale ranging from 1 ⫽definitely does not apply to 5 ⫽
definitely applies. The total scale consists of 28 items and provides
three dimensions measuring teacher-perceived closeness (11
items), dependency (5 items), and conflict (12 items) with the
child. Sample items include “I share an affectionate, warm rela-
tionship with this child” (closeness), “This child is overly depen-
dent on me” (dependency), and “This child easily becomes angry
with me” (conflict). The STRS is a widely used instrument of
preschool- and elementary-aged children (e.g., Birch & Ladd,
1997;Howes & Ritchie, 1999;Skalická et al., 2015). Because we
had no hypothesis regarding which aspect of the student–teacher
relationship would be (differentially) affected by the parent–child
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1303
OXTR, PARENTING, AND STUDENT–TEACHER RELATIONSHIPS
relationship, we applied the STRS total score. Cronbach’s ␣s were
.81, .84, and .83 at T1 through T3, respectively.
Parenting. Parenting was measured using the Dyadic Parent–
Child Interaction System (DPICS; Eyberg, Nelson, Duke, &
Boggs, 2005), which is an observation measure designed to cap-
ture the quality of parent–child interactions. The DPICS is used to
assess progress in Parent–Child Interaction Therapy (PCIT; Brink-
meyer & Eyberg, 2003), a parent-training program commonly used
to address behavioral problems in children (Bjørseth & Wich-
strøm, 2016;Nixon, Sweeney, Erickson, & Touyz, 2003) but
applied to internalizing problems as well, such as notably depres-
sion (modified version; Lenze, Pautsch, & Luby, 2011) and anx-
iety (Choate, Pincus, Eyberg, & Barlow, 2005). To tap parent–
child interaction under diverse conditions the DPICS codes the
interaction in three standardized 5-min sessions with very different
demands on the parent (and the child). First, in Child-led play, the
parent is instructed to let the child lead the interaction, which
might facility sensitivity and behavioral descriptions, following the
child’s lead. In the second situation, Parent-led play, the parent is
instructed to lead the play and make the child follow, facilitating
demands, instructions, and more controlling behavior from the
parent. Third, in Clean-up, the parent is instructed to have the child
clean up the material used by herself or himself. Thus, in this
situation parents are requested to enforce a rule and handle conflict
id the child does not behave according to the instruction. Corre-
lations between parent behavior in the three situations ranged from
modest (r⫽.13) to moderate (r⫽.43). A sumscore of parenting
in the three situations was created. Each session is videotaped, and
the recordings were coded by observers naive to the study hypoth-
eses and any information about the family. Before coding, the
observers underwent substantial training, and reliability checks
(80% agreement required to proceed) were conducted after every
25th video coding throughout the process.
According to the DPICS coding manual, each utterance from the
parent and child is assigned a code, and there are 27 parent codes
capturing positive, neutral, and various forms of alleged negative
parenting. Because our aim was to contrast differential suscepti-
bility with diathesis stress, we opted for a measure of parenting
that ranged from positive to negative, and not merely the absence
of positivity or negativity. Researchers have grouped the DPICS
codes in various ways, but one of the most common ways is to
assign positive communication to “Do-skills” (that are promoted in
PCIT) consisting of labeled (e.g., you did know the right answer)
and unlabeled praises (e.g., much better!), behavioral descriptions
(e.g., you are singing a song), and reflections (e.g., child: My doll
has blue eyes, parent: She does have blue eyes). Arguably, ques-
tions and commands from parents do not necessarily represent
negative communication in all circumstances. Therefore, to avoid
such ambiguity, we used only the negative talk (e.g., you are being
naughty) and negative touch (e.g., hitting, kicking, grabbing, or
shaking the child) categories. We henceforth created a communi-
cation composite that consisted of the ratio of positive communi-
cation to negative communication. A low score indicated parental
communication dominated by negative communication, a score of
1 indicated as many positive as negative utterances, and a score ⬎1
indicated communication dominated by positive communication.
The coders were blind to all of the information about the family
and the same coder coded the three sessions. The blinded coders
recoded tapes from 106 families, yielding an intraclass correlation
coefficient (ICC) of .80 for our communication measure across
multiple pairs of coders.
Table 1
Sample Characteristics (n ⫽652)
Variable MSDMinimum Maximum N
Demographics
Child age at T1 (months) 54.87 2.98 48.23 67.81 610
Male children (%) 50.8% 332
Age of parent at clinic (in years) 35.02 4.75 21.00 57.00 620
Women brought child to clinic (%) 84.4% 523
Relation to the child
Biological parents (%) 98.2% 609
Adoptive parents (%) 1.1% 7
Stepparents (%) .2% 1
Foster parents .5% 3
Marital status
Married (%) 55.7% 343
Lives together, ⬎6 months (%) 34.3% 211
Lives together, ⬍6 months (%) 1.1% 7
Divorced (%) 6.2% 38
Separated (%) 1.5% 9
Never lived together (%) 1.0% 6
Widow (%) .3% 2
Ethnicity
Ethnicity male parent (%) Norwegian 94.4 587
Ethnicity female parent (%) Norwegian 96.3% 599
Genotype n⫽652
Genotype A/A (%) 11.8 77
Genotype A/G (%) 46.8 305
Genotype G/G (%) 41.4 270
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1304 HYGEN, BELSKY, LI, STENSENG, GU
¨ZEY, AND WICHSTRØM
Genotype. Genotyping of OXTR rs53576 was performed using
KASP assays by LGC Genomics (http://www.lgcgroup.com/kasp),
which are based on competitive allele-specific PCR and enable
high-throughput genotyping of specific SNPs. The KASP geno-
typing reaction consists of three components: the sample DNA,
KASP Assay mix and KASP Master mix. The SNP-specific KASP
Assay mix and the universal KASP Master mix are added to DNA
samples, and a thermal cycling reaction is performed followed by
an endpoint fluorescent read that enables detection of the specific
allele/s present in a DNA sample for the chosen SNP. The KASP
Master mix contains the two universal FRET reporter cassettes,
which are labeled with FAM and HEX. The KASP Master mix
also contains ROX passive reference dye, Taq polymerase, free
nucleotides and MgCl2 in an optimized buffer solution. The KASP
Assay mix is specific to the target polymorphism and comprises
two competitive, allele-specific forward primers and one common
reverse primer. Each forward primer incorporates an additional tail
sequence that corresponds to one of two universal FRET cassettes
present in the KASP master mix. During thermal cycling of the
reaction, the target polymorphism is amplified and fluorescent
labels are incorporated via successive rounds of PCR. Once the
KASP reaction is complete, the resulting fluorescence is measured
on a BMG PHERAstar plate reader. The raw data are analyzed and
scored on a Cartesian plot, also known as a cluster plot, to assign
a genotype to each DNA sample using LGCs proprietary Kraken
software.
Covariates. At ages 4 and 6 years, child internalizing (␣⫽
.83, ␣⫽.82, respectively) and externalizing problems (␣⫽.89,
␣⫽.86) were measured by parental ratings on the Child Behavior
Check List (CBCL; Achenbach, 1991); the 1.5–5 years version
was used at age 4 and the 6–18 year version at age 6. Children’s
behavior toward their parents were observed while interacting with
their parents in the Child led play, Parent led play, and Clean-up
situations described above and assessed using the DPICS. A ratio
of positive (Positive touch, Laughter, and Prosocial behavior) to
negative behavior (Negative talk, Commands, Whining, Yelling,
and Negative touch) was computed at age 4 and 6, respectively
(ICC ⫽.60).
Statistical analysis. To assess the effect of level and change
in parent communication from 4 to 6 years of age on level and
change in the student–teacher relationship from 4 to 8 years, we
used piecewise growth modeling within a structural equation
framework using Mplus 7.31 (1998–2012). The piecewise growth
model yields three growth factor-related outcomes, the intercept,
which was set at age 4, the first period of growth in the student–
teacher relationship from age 4 to 6, and the second period of
growth in the student–teacher relationship from age 6 to 8. Like-
wise, as predictors, we modeled the level (set at age 4) and change
(from age 4 to 6) in parent communication. To model growth, the
residuals of the STRS and parent communication at each time
point were set to zero. The two periods of growth in the student–
teacher relationship were regressed on the intercept (age 4) of
parenting, and the second growth in the student–teacher-
relationship was regressed on the growth in parent communication
(from 4 to 6 years) as well. The intercept of the student–teacher
relationship and parent communication indices were allowed to
correlate. To adjust for expected regression toward the mean
effects, the growth in the student–teacher relationship and parent
communication was regressed on the respective intercepts.
The second stage of analysis proceeded, as outlined at the end of
the introduction, in two steps because of the failure of prior G⫻E
work to highlight which allelic subgroups were more or less
sensitive to environmental effects. Thus, we first conducted an
exploratory G⫻E test using piecewise regression analyses from 4
to 6 years and from 6 to 8 years to determine whether G⫻E was
operative and which allelic group(s) appeared most developmen-
tally malleable. If a significant G⫻E effect emerged, we proceeded
to the second step involving the competitive, model-fitting ap-
proach developed by Widaman and associates (2012;Belsky et al.,
2013) to evaluate the relative fit of two competing G⫻E accounts,
diathesis stress versus differential susceptibility. To examine
whether the child’s behavior toward the parent and child’s degree
of internalizing and externalizing problems confounded the results,
the child’s initial values of these potential confounders at age 4 as
well as the change from age 4 to 6 years were entered as covariates
in three separate models, one for each potential confounder. Three
separate models were conducted because entering all covariates in
one model was not applicable because there were more parameters
than the sample size allowed for; thus, such a model could not be
identified.
Under diathesis-stress theorizing, the predicted interaction
should be ordinal in form. Consider a biallelic polymorphism with
three possible genotypes, containing 0, 1, or 2 putative “risk”
alleles. According to diathesis-stress, a regression model with a
Linear G ⫻Linear E interaction should reveal four outcomes: (a)
a small or zero effect of the environment for the (“resilient”) group
with 0 risk alleles, (b) a stronger, significant effect of the environ-
ment for the “vulnerability” group with 2 risk alleles, (c) a mid-
dling outcome by the group with 1 risk allele, and (d) a crossover
point of the linear functions at or near the most positive value for
the environment.
Differential susceptibility leads to a contrasting prediction re-
garding the form of the G⫻E interaction. The alternate alleles
under differential susceptibility are recast as “plasticity” and “non-
plasticity” alleles, rather than risk and resilience alleles, respec-
tively. The predicted interaction would still have a small (or nil)
effect of the environment for the least-plastic group, a stronger,
significant effect of the environment for the plastic group, and a
moderate effect for the moderately plastic group. However, the
crossover point of these three linear functions would be near the
middle of the distribution of scores on the environmental variable,
thus revealing a “for better and for worse” pattern (Belsky et al.,
2007), with “better” outcomes (i.e., more positive student–teacher
relationship) predicted for the most-plastic group under more
favorable environmental conditions (i.e., greater positive change in
parenting) and “worse” outcomes (i.e., poorer student–teacher
relationship) for the most-plastic group under less favorable con-
ditions (i.e., less positive/more negative change in parenting).
The location of the crossover point for the predicted outcomes
is, therefore, the crucial parameter that distinguishes predictions
for the G⫻E interaction for the competing diathesis-stress and
differential-susceptibility positions. Widaman et al. (2012) pro-
posed a reparameterized regression model that makes the cross-
over point one of the parameters to be estimated. One major
benefit of the reparameterization is that the point estimate of the
crossover point is accompanied by a SE, so that an interval
estimate can be calculated. Among other things, the reparameter-
ized model allows model fit under differential-susceptibility and
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1305
OXTR, PARENTING, AND STUDENT–TEACHER RELATIONSHIPS
diathesis-stress conditions to be statistically contrasted, with the
better fitting model offered as the optimal representation of the
data.
Widaman et al. (2012;Belsky et al., 2013) distinguished four
reparameterized models that can provide tests of key parameters
consistent with (a) weak and (b) strong differential-susceptibility
and (c) weak and (d) strong diathesis-stress predictions. In the case
of weak models, all of the subgroups show significant associations
between predictor and outcome, but it proves stronger for some
groups than others. In strong models, the degree to which there is
a predictor-outcome association also varies as a function of allelic
condition, but the association is nonsignificant for at least one
subgroup. Thus, whereas strong models presume that some indi-
viduals are not susceptible to environmental effects (i.e., no asso-
ciation between predictor and outcome), weak models presume
that all individuals are susceptible, while some individuals are
presumed to be more susceptible than others.
Therefore, to determine whether an interaction effect reflects
diathesis-stress or differential susceptibility, we evaluated whether
the regression slopes for the A homozygotes and the G carriers
crossed within the range of available data of parent communication
using the competitive model-testing procedures of Widaman et al.
(2012) and Belsky et al. (2013). Recall that there is not sufficient
available evidence to single out AA, AG, or GG carriers as a priori
plastic or resilient or grouping hetero or homozygotes. Treating all
G carriers as members of a single group was therefore based on
preliminary analyses revealing that the effects of parenting on
STRS did not vary across hetero-and homozygote G carriers (see
Table 3).
A robust maximum likelihood estimator was applied. The anal-
yses were performed on cases that were genotyped and had infor-
mation on the dependent variable, that is, STRS, and missing data
on the independent variables were handled according to a full
information maximum likelihood procedure (FIML). The attrition
analyses indicated that data were missing not at random (MNAR).
FIML uses all available information and arrives at less biased
estimates than complete case analysis in the present MNAR situ-
ation (Schafer & Graham, 2002). To examine whether G⫻E was
present, we tested the difference in pairs of regression slopes from
parent communication to the student–teacher relationship in the
various allelic groups by comparing the model fit of a model in
which the slopes were fixed to be equal with the model fit when
they were freely estimated, using a Wald test.
With a screen-stratified sample, all parameters were weighted
with a factor corresponding to the number of subjects in the
population in a particular stratum divided by the number of par-
ticipants in that stratum (i.e., low-screen scorers were “weighted
up,” and high scorers were weighted down). Two-sided pval-
ues ⬍0.05 were regarded as statistically significant, and 95% CIs
are reported when relevant.
Results
Preliminary descriptive analyses. Table 2 displays the de-
scriptive statistics of and correlations among all variables included
in the primary piecewise growth analysis. Measures of parenting
and the student–teacher relationship were moderately skewed.
However, and more important to our inquiry, the distribution of the
change in parenting and in student–teacher relationship were not
skewed, S
parenting
⫽⫺.47, S
STRS12
⫽⫺.06, and S
STRS23
⫽.03.
There was a decline in student–teacher relationship scores from
day-care (age 4) to Grade 1 (age 6) as seen in the mean value of
the yearly change score, M⫽⫺1.23, p⬍.001, but not from age
6to8,M⫽.09, p⫽.62. A slight drop in parenting scores from
age 4 to 6 was also seen, M⫽⫺.48, p⫽.015.
Furthermore, parenting at age 4 was not correlated with STRS at
age 4, but a small correlation emerged between positive parenting
at age 4 and 6 and better STRS scores at ages 6 and 8—indicating
that it is not until children start school that parenting is predictive
of student–teacher relationships. Moreover, day care teachers re-
ported better relationships with children homozygous for the A
allele (i.e., AA) at age 4, whereas teachers reported better rela-
tionships with GG homozygotes at age 8; the GGs also received
more positive communication from parents at age 4 than the other
allelic groups. As shown in Table 2, the latter association emerged
in the context of AG heterozygotes receiving less positive com-
munication.
Primary prediction analyses. Table 3 presents the results of
the piecewise regression analyses. Recall that these analyses eval-
uated whether rs53576 moderated (1) the effect of level of positive
parenting at age 4 on (a) student–teacher-relationships level at age
4 of and on (b) change in the student–teacher relationship from age
4 to 6 and (c) from age 6 to 8, and also whether this polymorphism
moderated (2) the effect change in positive parenting from age 4 to
6 on change in a positive student–teacher relationship from age 6
to 8. The results revealed that in the case of AA homozygotes, a
higher level of positive parenting at age 4 was associated with
better student–teacher relationships at age 4 and predicted im-
proved student–teacher relationships from age 4 to 6 (p⫽.03) and
from age 6 to 8 (p⫽.06). Although these significant (or near-
Table 2
Descriptions of and Correlations Between Study Variables (TESS)
Study variables Mor % SD Skewness 1 2 3 4 5 6 7
1. Parenting–4 years 7.30 8.88 1.74
2. Parenting–6 years 6.35 7.65 1.84 .29
ⴱⴱⴱ
3. Student–teacher relationship–4 years 118.08 7.45 ⫺1.22 .07 .04
4. Student–teacher relationship–6 years 115.56 8.18 ⫺1.45 .08
ⴱ
.09
ⴱ
.23
ⴱⴱⴱ
5. Student–teacher relationship–8 years 116.09 9.73 ⫺1.57 .08
ⴱ
.11
ⴱⴱ
.27
ⴱⴱⴱ
.55
ⴱⴱⴱ
6. OXTR–AA 11.8% .02 .00 .07
ⴱ
.01 ⫺.07
7. OXTR–AG 46.8% ⫺.12
ⴱⴱ
.00 ⫺.07 ⫺.02 ⫺.05 ⫺.33
ⴱⴱⴱ
8. OXTR–GG 41.4% .11
ⴱ
.00 .02 .01 .09
ⴱ
⫺.31
ⴱⴱⴱ
⫺.80
ⴱⴱⴱ
Note. TESS ⫽Trondheim Early Secure Study; OXTR ⫽oxytocin receptor gene.
ⴱ
p⬍.05.
ⴱⴱ
p⬍.01.
ⴱⴱⴱ
p⬍.001.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1306 HYGEN, BELSKY, LI, STENSENG, GU
¨ZEY, AND WICHSTRØM
significant) effects did not emerge in the case of AG and GG
carriers, the differences in effects between AA carriers and the
other genotypes were not significant; that is, there was no (statis-
tically significant) genetic moderation.
Notably, however, the effect of change in parent communication
from 4 to 6 years on change in teacher–student relationships from
age 6 to 8 proved to be genetically moderated, with the parenting
effect on teacher–student relationships among AA carriers being
significantly stronger than that among AG and GG carriers. Given
that AA carriers differed from both AG and GG carriers and that
the latter two allelic subgroups did not differ from each other, in
the next stage of analysis, A homozygotes were contrasted with G
carriers.
To adjust for potential confounders (internalizing behavior, ex-
ternalizing behavior, and child behavior toward parent) we entered
the level of the respective potential confounder at age 4 as well as
change in the potential confounder from age 4 to 6 in our model.
Important to appreciate before moving on to the next stage of
analysis is that adjustment for potential confounders made no
substantial impact on the documented effects of parenting on later
student–teacher relationship or the genetic moderation of this
effect. More specifically, when child behavior toward the parent
was controlled, the difference in the effect of change in parenting
from age 4 to 6 on change in student–teacher relationship from age
6 to 8 remained between the AA and GG groups, Wald ⫽5.77,
df ⫽1, p⫽.02 as well as between the AA and AG groups,
Wald ⫽4.15, df ⫽1, p⫽.04. This was also the case when
controlling for externalizing problems, AA and GG groups:
Wald ⫽6.65, df ⫽1, p⫽.01; AA and AG groups: Wald ⫽4.23,
df ⫽1, p⫽.04, and mostly so when controlling for internalizing
problems, AA and GG groups: Wald ⫽64.49, df ⫽1, p⫽.03; AA
and AG groups: Wald ⫽2.59, df ⫽1, p⫽.11. As regards the
latter difference between the AA and AG groups, this difference in
regression coefficients with adjustment of .27 (95% CI [⫺.058,
.59]) compared with a difference in regression coefficients of .40
with no adjustment for internalizing problems. Hence, because the
latter difference falls within the CI of the first difference, this
change in the effect of change in parenting on change in the
student–teacher relationship when adjustment for internalizing
problems was implemented was not significant.
Given the possibility that evocative GE correlations (rGE) might
explain G⫻E findings (Plomin, DeFries, McClearn, & McGuffin,
2008), and that one cross-sectional study indicated that OXTR
might influence child behavior, which in turn affect parenting
(Kryski et al., 2014) and considering that OXTR evidenced some
weak main effects with predictors and outcome (see Table 2)we
examined whether our main finding regarding the AA carriers
could be attributed to rGE. If this was the case there should be an
indirect effect from AA-status to parenting (either intercept or
change) and from parenting to student–teacher relationship. We
found no indication of such rGE involving level of parenting at age
4 (indirect effect ⫽.02, p⫽.76) or change in parenting from age
4 to 6 (indirect effect ⫽.00, p⫽.94).
To determine whether the detected G⫻E interaction proved to
be more consistent with the diathesis stress or the differential
susceptibility framework, we evaluated whether the cross-over
point of the regression lines for the effect of change in positive
parenting from age 4 to 6 on change in the student–teacher rela-
tionship from 6 to 8 years among AA and G carriers differed from
Table 3
Effect of Level (At Age 4 Years) on Level of (At Age 4) and Change in Positive Parenting From Age 4 To 6 Years and Age 6 To 8 Years and the Effect of Change in
Positive Parenting From Age 4 to 6 Years on Change in the Student–Teacher Relationship From 6 To 8 Years According to Variations in the Rs53576 Genotype (TESS)
Predictors and outcome
AA (n⫽77) AG (n⫽305) GG (n⫽270) Pairwise comparisons between genotypes
95% CI p-value 95% CI p-value 95% CI p-value
AA vs. AG AA vs. GG AG vs. GG
Wald (p-value) Wald (p-value) Wald (p-value)
Level of student–teacher relationship at 4 years
a
Level of parenting at 4 years
b
.21 [.04, .38] .016 .08 [⫺.01, .18] .09 ⫺.01 [⫺.17, .16] .94 .96 (.33) 2.15 (.14) .73 (.39)
Change from 4 to 6 years in student–teacher relationship
a
Level of parenting at 4 years
b
.19 [.02, .36] .031 .04 [⫺.02, .10] .21 .03 [⫺.07, .13] .58 1.35 (.24) 1.74 (.18) .09 (.76)
Change from 6 to 8 years in student–teacher relationship
a
Level of parenting at 4 years
b
.34 [⫺.02, .70] .06 .02 [⫺.18, .22] .83 .07 [⫺.10, .23] .43 1.51 (.22) 1.44 (.23) .06 (.81)
Change in parenting from 4 to 6
years
b
.58 [.18, .97] .004 .07 [⫺.10, .25] .41 .03 [⫺.13, .18] .74 5.11 (.024) 7.23 (.007) .19 (.66)
Note. TESS ⫽Trondheim Early Secure Study; CI ⫽confidence interval.
a
outcome.
b
predictor.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1307
OXTR, PARENTING, AND STUDENT–TEACHER RELATIONSHIPS
the highest and lowest observed values of change in positive
parenting from 4 to 6 years. Thus, the cross-over point was first
fixed to the lowest observed change value using the model con-
straint procedure in Mplus. Then, in a new analysis, the cross-over
point was fixed to the highest observed change value. The results
of these analyses proved that the cross-over-point was clearly
higher than the lowest observed value, C
min
⫽43.40 [33.39;
53.42], p⬍.001, and lower than the highest value,
C
max
⫽⫺21.53 [⫺31.54; ⫺11.51], p⬍.001. The detected G⫻E
interaction proved more consistent with that differential-
susceptibility than diathesis-stress framework.
To test whether a strong versus weak version of differential
susceptibility fit the data best, we compared one model in which
the change in student–teacher relationship did not vary as a func-
tion of change in parenting among G carriers but only among A
homozygotes (i.e., strong differential susceptibility) to a second
model in which the change in student–teacher relationship did vary
as a function of change in parenting in both genotypic groups even
if to a differing degree (i.e., weak differential susceptibility). The
strong differential susceptibility model fit the data well,
2
⫽3.46,
df ⫽3, p⫽.33, as did a weak differential susceptibility model,
2
⫽2.71, df ⫽2, p⫽.26. Because the two models were not
different when compared using Satorra’s method (Satorra, 2000),
⌬
2
⫽.58, df ⫽1, p⫽.44, the strong differential susceptibility
model was preferred for reasons of parsimony. As shown in Figure
1, G carriers did not benefit from increased positive parenting, nor
did they succumb to the potential negative effects of decreased
positive or increased negative parenting. A homozygotes, in con-
trast, benefitted from increased positive parenting and suffered
from declines in positive parenting. Thus, under increased positive
parenting, the children’s prospective relations with their teachers
improved, whereas under declines in positive parenting, their
future relations deteriorated.
Discussion
Given evidence that positive and supportive student–teacher
relationships benefit children, the primary purposes of the research
reported herein were to determine whether parenting, and change
in parenting, predicted the development of this other influential
relationship in interaction with child genotype OXTR rs53576,
given the latter’s repeated association with multiple aspects of
social functioning, and whether any detected G⫻E interaction
proved more consistent with diathesis-stress or differential suscep-
tibility models of Person-⫻-Environment interaction.
The results revealed that (a) although there was no evidence
of genetic moderation of effects of parenting at age 4 on age-4
student–teacher relationships or (b) on change in the student–
teacher relationships from age 4 to 6, (c) such genetic moder-
ation proved evident when change in parenting from age 4 to 6
was used to predict change in student–teacher relationships
from age 6 to 8. Notably, the latter finding proved consistent
with the differential susceptibility rather than diathesis stress
model of Person-⫻-Environment interaction (Belsky et al.,
2007,2013;Belsky & Pluess, 2009,2013;Ellis et al., 2011).
Recall in this regard that in the case of AA homozygotes only,
greater improvement in parenting from age 4 to 6 predicted
increases in the quality of student–teacher relationships from
age 6 to 8, although, at the same time, smaller increases or
greater declines in parenting predicted more deterioration of the
student–teacher relationship.
Among G-carrying children, there were no effects of parenting
at any age on student–teacher relationship, which is why the results
were more consistent with the strong than weak version of differ-
ential susceptibility. Recall that the weak version requires signif-
icant predictor-outcome associations for multiple subgroups, with
these associations simply being stronger in some subgroups than in
others (Belsky et al., 2013;Widaman et al., 2012). G-Carrying
children, then, appear much less developmentally plastic or mal-
leable in response to environmental conditions, at least the envi-
ronmental condition identified in this report, change in parenting.
Whether the same would prove to be the case with regard to other
oxytocin pathway genes, environmental conditions, or develop-
mental outcomes, including those associated with relationships,
and/or at other times of measurement—both earlier and later in
development—remains to be determined.
Although there were grounds for predicting a G⫻E interaction
involving OXTR rs53576 in the current inquiry, it was not antic-
ipated that evidence of such would emerge only when change in
parenting from age 4 to 6 was used to predict change in student–
teacher relationships from age 6 to 8. We are forced to wonder why
a similar G⫻E interaction did not emerge when parenting at age 4
was used to predict change in teacher–student relationships from
age 4 to 6. Because this latter association was significant only for
AA carriers and not for G carriers, although the difference was not
strong enough to generate a significant G⫻E interaction, it would
be a mistake to conclude that the two sets of results were com-
pletely inconsistent. Indeed, the fact that the significant G⫻E
effect involved parenting change whereas the insignificant one
involved only parenting measured at a single point in time raises
the possibility that it may be relational stability—or lack thereof—
that AA carriers are most responsive to.
Also to be considered is that the parenting change that proved
sensitive to genetic moderation occurred across a major develop-
mental transition, from preschool to school. In Norway, students
are scheduled to have one—and the same—teacher in most, if not
all subjects, during Grades 1 to 3. In contrast, the change from day
care (age 4) to 1st grade (age 6) always involves a change of
teacher. Hence, whereas the change from age 6 to 8 in the student–
teacher relationship captures the change within a specific relation-
ship, the change from age 4 to 6 also involves a change of persons
Figure 1. Change in parenting from 4 to 6 years predicting change in
student–teacher relationship from 6 to 8 years in Study 1.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1308 HYGEN, BELSKY, LI, STENSENG, GU
¨ZEY, AND WICHSTRØM
within that relationship. Moreover, in first grade, the teachers had
known the student for only approximately half a year. Together,
these factors may make the change in student–teacher relationship
from day-care to school a less sensitive measure of changes in
relationship quality than within school, from age 6 to 8. Whether
these considerations explain the variation in findings reported here
will need to be determined by future research.
We acknowledge that little is known about the functioning of
OXTR rs53576. However, recent studies using animal models and
brain region-specific manipulations of oxytocin activity are begin-
ning to highlight the complex effects of oxytocin (Maroun &
Wagner, 2016). The many impacts of oxytocin on social behavior
are suggested to partly be due its local effects on amygdala. It is
also reported that AA carriers of rs53576 demonstrate increased
coupling of the hypothalamus and amygdala (Tost et al., 2010).
This is significant in the present context because hypothalamus is
involved in the regulation of the autonomic nervous system (ANS)
and thereby regulating the physiological part of emotional re-
sponses. Moreover, amygdala and hypothalamus are both involved
in emotional processing (Allen, 2009). Additionally, the A allele is
found to confer higher physiological and dispositional stress reac-
tivity (Rodrigues et al., 2009).Taken together, we speculate that
the amygdala-hypothalamus-Oxytocin connection and the stress
reactivity findings may explain some of the present findings; AA
children’s brains might be more reactive to environmental stimuli
such as parenting. However, further experimental research is nec-
essary to achieve greater understanding of the biological mecha-
nisms involved. In fact, given concerns regarding replicability of
all scientific findings, there is a need to determine if those which
emerged from Study 1 can be replicated, the issue addressed in
Study 2.
Study 2
Method
Participants. Participants were recruited during the first 11
month of 1991 in 24 hospitals from 10 data collection locations in
the United States (Charlottesville, VA; Irvine, CA; Lawrence, KS;
Little Rock, AR; Madison, WI; Morganton, NC; Philadelphia, PA;
Pittsburgh, PA; Seattle, WA; and Wellesley, MA). In total, 8,986
women who gave birth in selected 24-hr interval were screened for
eligibility and 1,364 families with healthy newborns completed a
home interview when the infant was 1 month, becoming research
participants. More details about recruitment and selection proce-
dure can be found in NICHD ECCRN (2005).
The analysis sample for Study 2 included all White children on
whom DNA was obtained and for whom the OXTR haplotype
rs53576 was successfully assayed. This subsample included 562
children. DNA was not available if parents or children did not
agree to provide sample or the quality of the sample was compro-
mised or proved too limited to assay. Furthermore, three children
were excluded from the sample because of missing data on vari-
ables central to this inquiry (i.e., parenting and student–teacher
relationship), resulting in a final analysis sample of 559 children-
(n
boys
⫽274). Whereas measurements of parenting were made in
Study 1 at ages 4 and 6 years and of the and student–teacher
relationship at ages 4, 6, and 8 years, related measurements in
Study 2 of parenting were obtained at 54 months and in 1st Grade
(approximately 6 years of age) and, for the teacher–child relation-
ship, at 54 months, 1st grade, and 3rd grade (about 8 years of age).
Measures.
Student–teacher relationship. Children’s 1st and 3rd grade
teachers were asked to rate their perceptions about their relation-
ship with the study child using the Student–Teacher Relationship
Scale, Short form (STRS; Pianta, 2001). This form consisted of 15
items (of the 28 items used in Study 1) scored on a 5-point Likert
scale that ranged from 1 ⫽definitely does not apply to5⫽
definitely applies. The scale measured two dimensions of teacher–
child relationship: teacher–child conflict (7 items; e.g., “The child
easily becomes angry at me”) and teacher–child closeness (8
items, e.g., “I share an affectionate, warm relationship with the
child”). As in Study 1, subscale scores were combined to create a
STRS total score (␣⫽.83, .86 and .89 for 54 months, Grade 1 and
Grade 3, respectively). This was done to reduce number of statis-
tical tests and because there were no distinguishing hypotheses for
the two dimensions of the student–teacher relationship that were
measured.
Parenting. Parenting was measured when children were 54
months of age and, again, in 1st grade. Mother–child interaction
was videotaped in free-play for 15 min; videotapes from all re-
search sites were shipped to a central location and coded by highly
trained and reliable evaluators blind to any and all information
about the family. The response categories ranged from 1 ⫽very
low to7⫽very high on each of the six scales reflecting the
mothers’ behavior, including mothers’ supportive presence, re-
spect for the child’s autonomy, stimulation of cognitive develop-
ment, quality of child assistance, hostility toward the child, and
mothers’ confidence. As in Study 1, we created the positive or
negative parenting ratio by dividing the sum of three positive
parenting scales (i.e., supportive presence, respect for the child’s
autonomy, and quality of child assistance) by the maternal hostility
rating. Therefore, a score of 3 indicated positive and negative
parenting behavior occur to the same extent proportionally,
whereas a score ⬎3 indicated mother–child interaction dominated
by positive parenting.
Genotyping. Genetic data were collected by buccal swabs
when children were 15 years of age. DNA extraction and geno-
typing for the NICHD Study of Early Child Care and Youth
Development (SECCYD) was performed at the Genome Core
Facility in the Huck Institutes for Life Sciences at Penn State
University under the direction of Deborah S. Grove, Director for
Genetic Analysis. Taqman SNP Genotyping Assays were per-
formed using an Allelic Discrimination Assay (Applied Biosys-
tems, Foster City, CA) protocol. Forty nanograms of DNA were
combined in a volume of 5 l with 2X Universal PCR Mix
(Applied Biosystems) and 1/20 the volume of the Taqman SNP
assay in a 384 well plate. A Pre-Read was performed and then PCR
as follows: a 10 min hold at 95 °C, followed by 40 to 45 cycles of
15 secs at 92 °C and then 1 min at 60 °C in a 7900HT PCR System.
After amplification, a Post-Read was performed to analyze. Auto-
matic and manual calls were made. Frequency distributions for the
OXTR rs53576 SNP did not depart significantly from the Hardy-
Weinberg equilibrium, n(GG) ⫽233, n(AG) ⫽260, n(AA) ⫽
66,
2
⫽0.26, p⬎.05. The analysis sample coded GG as “3,” AG
as “2,” and AA as “1.”
Statistical analysis. Analyses in Study 2 followed the exact
same procedures used in Study 1. Specifically, we assessed the
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1309
OXTR, PARENTING, AND STUDENT–TEACHER RELATIONSHIPS
effects of level and change in parenting behavior from 54 months
to 1st grade on level and change in student–teacher relationship
from 54 months to 3rd grade using piecewise growth modeling
within a structural equation framework in Mplus 7.3 (Muthén &
Muthén, 1998–2012). The piecewise growth model yields three
growth-factor-related outcomes, the intercept, which was set at 54
months, the first period of growth in the student–teacher relation-
ship from 54 months to 1st grade, and the second period of growth
in this relationship from 1st grade to 3rd grade. Likewise, as
predictors, we again modeled the level (set at 54 months) and
change (from 54 months to 1st grade) in parenting. To model
growth, the residuals of the STRS and parenting at each time point
were set to zero. As in Study 1, the two periods of growth in the
student–teacher relationship were regressed on the intercept (54
months) of parenting, with the second period of growth in the
student–teacher-relationship regressed on growth in parenting
from 54 months to grade one. The intercept of the student–teacher
relationship and parenting indices were allowed to correlate. To
adjust for expected effects of regression toward the mean, growth
in the student–teacher relationship and in parenting were regressed
on their respective intercepts.
Results
Preliminary descriptive analyses. Table 4 displays the de-
scriptive statistics of and correlations among all variables included
in the primary piecewise growth analysis. Measures of parenting
and the student–teacher relationship were again moderately skewed,
even though parenting measures were skewed to a lower extent.
However, the distribution of changes in parenting and in student–
teacher relationships were again not skewed, S
parenting
⫽⫺.10,
S
STRS12
⫽⫺.24, and S
STRS23
⫽.11. Mean comparisons revealed
that there was a small decline—across the entire sample—in
student–teacher relationship score from 1st grade to 3rd grade as
seen in the mean value of the yearly change score, M⫽⫺0.52,
p⫽.005. No overall change emerged in terms of STRS score from
54 months to 1st grade (M⫽0.36, p⫽.12) and parenting scores
from 54 months to 1st grade (M⫽.01, p⫽.96; unstandardized
change scores reported). Furthermore, parenting at 54 months was
only significantly and positively correlated with student–teacher
relationships at 54 months, not later time points (i.e., 1st and 3rd
grade), despite trends in the same direction. Parenting measured at
1st grade, however, was significantly and positively associated
with student–teacher relationships at 1st grade, but not 3rd grade.
Finally, OXTR genotype was not significantly associated with
student–teacher relationship at any measurement occasion.
Primary prediction analyses. Table 5 presents the results of
the piecewise regression analyses. Recall that these analyses eval-
uated whether rs53576 moderated (1) the effect of level of positive
parenting at 54 months on (a) student–teacher-relationship level at
54 months and on (b) change in the student–teacher relationship
from 54 months to 1st grade and (c) from 1st grade to 3rd grade;
and also whether this polymorphism moderated (2) the effect
change in positive parenting from 54 months to 1st grade on
change in positive student–teacher relationship from 1st grade to
3rd grade. Results revealed that in the case of GG homozygotes, a
higher level of positive parenting at 54 months was associated with
better student–teacher relationship at 54 months (p⫽.02). Even
though this significant effect did not emerge in the case of AA and
AG carriers, the differences in effects between GG and other
genotypes were not significant, that is, there was no (statistically
significant) genetic moderation. Furthermore, none of the other
associations between parenting and student–teacher relationship,
level or change, turned out significant in any of the allelic group,
and none of the associations were moderated by the OXTR
rs53576 genotype.
General Discussion
Recall that results of Study 1 involving a representative com-
munity sample (after back-weighting) of Norwegian children re-
vealed a G⫻E interaction consistent with the (strong) differential-
susceptibility model of Person ⫻Environment interaction. Thus,
for AA homozygotes (only) of the OXTR rs53576 polymorphism,
positive and negative changes in parenting from age 4 to 6 fore-
cast, respectively, positive and negative changes in the student–
teacher relationship from age 6 to age 8. Given widespread con-
temporary concerns about the replicability of many scientific
findings, perhaps most especially those involving measured genes
and involving G⫻E interactions, we sought to replicate the Nor-
wegian findings by taking advantage of somewhat similar data
collected as part of a large scale U.S. study. Clearly, this replica-
tion effort failed, as the G⫻E interaction detected in Study 1 and
involving the same polymorphism and change in parenting failed
to predict change in the student–teacher relationship in Study 2.
Why might that have been the case?
The most obvious reason is that, consistent with previously cited
critiques of G⫻E findings, those discerned in Study 1 simply
Table 4
Descriptions of and Correlations Between Study Variables in the NICHD Study (N ⫽559)
Study variables Mor % SD Skewness 1 2 3 4 5 6 7
1. Parenting–54 months 14.14 5.11 ⫺.82 —
2. Parenting–Grade 1 14.15 5.37 ⫺.71 .33
ⴱⴱⴱ
—
3. Student–teacher relationship–54 months 65.11 7.70 ⫺1.11 .13
ⴱ
.14
ⴱ
—
4. Student–teacher relationship–Grade 1 65.82 7.73 ⫺1.12 .06 .08
ⴱ
.31
ⴱⴱⴱ
—
5. Student–teacher relationship–Grade 3 64.79 8.13 ⫺1.04 .10
†
.04 .35
ⴱⴱⴱ
.46
ⴱⴱⴱ
—
6. OXTR–AA 11.8% .06 .02 ⫺.02 .06 .04 —
7. OXTR–AG 46.5% ⫺.06 .01 ⫺.04 ⫺.08
†
⫺.07
†
⫺.34
ⴱⴱⴱ
—
8. OXTR–GG 41.7% .02 ⫺.01 .06 .04 .05 ⫺.31
ⴱⴱⴱ
⫺.79
ⴱⴱⴱ
Note. NICHD ⫽National Institute of Child Health and Human Development.
†
p⬍.10.
ⴱ
p⬍.05.
ⴱⴱⴱ
p⬍.001.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1310 HYGEN, BELSKY, LI, STENSENG, GU
¨ZEY, AND WICHSTRØM
reflected a false positive finding. Although this very real and
perhaps most likely possibility cannot—and should not—be dis-
missed, it is worth considering differences across the two studies
that might have contributed to the apparent failure to replicate. We
say “apparent” because if the differences to be highlighted actually
played a role in accounting for why Study 1 results did not
reemerge in Study 2, it would imply that failure to replicate would
be an inaccurate characterization of how best to conceptualize the
relation between the results of the two studies.
Recall from the discussion of Study 1 results that Norwegian
children usually remain with the same teacher across Grades 1, 2,
and 3. That is strikingly different from school practices in the
United States, as children there typically change teachers each and
every academic year. Thus, the measure of change in the student–
teacher relationship was based on ratings made by different teach-
ers in the U.S. sample but by the very same teachers in Norway. It
seems quite likely, then, that Norwegian teachers who rated their
third graders knew them much better than did the American third
grade teachers, having spent three times as much time with them.
It also seems possible that, as a result, the quality of student–
teacher relationships might be different across the two countries.
Conceivably, even if uncertainly, such differences across countries
could have played a role in our failure to replicate Study 1 results
in Study 2.
One also needs to consider the fact that school quality and
teacher skills likely vary much more in the United States than in
Norway, because of the funding for schools being determined by
local property taxes, not by the federal or even state governments,
as is the case in Norway. Such a difference stems, no doubt, from
more fundamental differences between the two countries (Andre
& Heien, 2001). Because Norway is a social democratic society in
which the state funds and governs the schools, there is likely much
greater heterogeneity in the quality of teachers and teaching in the
United States than in Norway. Moreover, the OECD PISA study
(Organisation for Economic Co-operation and Development
[OECD], 2016)–The Programme for International Student Assess-
ment: a triennial international survey—indicated that, compared
with Norway, American students were a lot more prone to absen-
teeism and more students viewed their teachers as too strict. These
differences, too, could have played a role in our failure to replicate
Study 1 findings in Study 2.
Related to its social-democratic nature, it also needs to be
appreciated that the social welfare system in Norway is far more
supportive of families than is the American system. Thus Ameri-
can children from the poorest backgrounds have fewer resourc-
es—at home, at child care and at school—than they would if they
lived in Norway. Not inconsistent with this observation is evidence
from the OECD PISA study (OECD, 2012) showing that lower
socioeconomic status in America influenced student performance
and led to a negative disciplinary climate in schools.
From the standpoint of family policy it is also important to
appreciate that in Norway parents are financially compensated for
staying home with their newborn during the entirety of the child’s
first year, whereas there is no federal policy of paid leave in the
United States, even if a few states and some private enterprises do
offer some kind of compensation. To our knowledge, however,
none are as generous as what is available to all families raising an
infant in Norway. Furthermore, all Norwegian children have free
access to free health care until the age of 18 and younger children
Table 5
Effect of Level (At 54 Months) on Level of (At 54 Months) and Change in Positive Parenting From Age 54 Months to 1st Grade and 1st Grade to 3rd Grade and the Effect
of Change in Positive Parenting From 54 Months to 1st Grade on Change in the Student–Teacher Relationship From 1st Grade to 3rd Grade According to Variations in the
rs53576 Genotype (NICHD)
Predictors and outcome
AA (n⫽66) AG (n⫽260) GG (n⫽233) Pairwise comparisons between genotypes
95% CI p-value 95% CI p-value 95% CI p-value
AA vs. AG AA vs. GG AG vs. GG
Wald (p-value) Wald (p-value) Wald (p-value)
Level of student–teacher relationship at 54 months
a
Level of parenting at 54 months
b
.05 [⫺.38, .47] .83 .09 [⫺.08, .26] .28 .20 [.03, .36] .02 .02 (.89) .19 (.66) .48 (.49)
Change from 54 months to grade 1 in student–teacher relationship
a
Level of parenting at 54 months
b
.02 [⫺.11, .16] .73 .02 [⫺.09, .14] .69 ⫺.01 [⫺.12, .10] .83 .00 (.99) .17 (.68) .20 (.65)
Change from grade 1 to grade 3 in student–teacher relationship
a
Level of parenting at 54 months
b
.06 [⫺.22, .33] .68 .02 [⫺.13, .18] .75 ⫺.01 [⫺.17, .16] .92 .01 (.91) .15 (.71) .09 (.76)
Change in parenting from 54 months to grade
one
b
⫺.03 [⫺.30, .24] .82 ⫺.02 [⫺.15, .11] .73 ⫺.05 [⫺.19, .10] .54 .00 (.98) .01 (.93) .02 (.88)
Note. NICHD ⫽National Institute of Child Health and Human Development; CI ⫽confidence interval.
a
outcome.
b
predictor.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1311
OXTR, PARENTING, AND STUDENT–TEACHER RELATIONSHIPS
have regularly scheduled examinations at free health clinics that all
children are expected to attend (and which TESS used to enroll
participants). This means that sampling in the NICHD Study was
probably far less representative of each of the 10 locales from
which families were recruited, including the eight which com-
prised the sample of this report because of their collection of child
DNA, than was the case of TESS.
Beyond the societal differences already outlined that may have
contributed to the replication “failure” under consideration, there
are also methodological differences across studies that merit con-
sideration.
Most worthy of attention is the fact that while TESS used the
same measures of parent–child interaction, of teacher–student
relationships and of child social functioning whenever measure-
ments included in this report were repeated, such assessments
varied across time in the NICHD Study. This meant, of course, that
they also differed at times from those used in TESS, even if the
general constructs being assessed were more similar than different
across inquiries. Like societal differences already considered, then,
these methodological differences meant that Study 2 could not be
considered, even at the measurement level, an exact replication of
Study 1. Once again, all this is not to say that the failure to
replicate should not be taken seriously, only that there would seem
to be enough uncertainty as to its cause to raise questions as to
whether the contrasting findings of Study 1 and 2 truly reflect such
a failure. Moreover, by examining the correlation tables for the
two studies, there are apparent differences. Most significant is that
in TESS, parenting at age 4 is significantly correlated with
student–teacher relationship at age 6 and age 8 (school age), the
same finding emerged for age 6 parenting, namely that it is
significantly correlated with student–teacher relationship at ages 6
and 8. In the NICHD study parenting at age 4 is only significantly
correlated with student–teacher relationship at age 4, whereas age
6 parenting is correlated with age 4 and 6 student–teacher rela-
tionship, not at age 8. No association between parenting at age 6
and student–teacher relationship at age 8 may explain why the
Norwegian study found an association, whereas the American
study did not.
Strengths and Limitations
Whether or not one considers this report as evidence of failure
to replicate, it is important to highlight both strengths and limita-
tions of the work presented herein. With regard to strengths, it is
notable that both Study 1 and 2 included large, community-based
samples—though only Study 1 could be considered representative
of the community from which it was drawn (after backweight-
ing)—independent ratings of exposures (observed parent–child
interaction) and outcome (teacher ratings), adjustment for covari-
ates, and a longitudinal design. Worth noting, in addition, was our
focus on change predicting change using temporally ordered pre-
dictors and outcomes, thereby providing a stronger basis for in-
ferring cause and effect than nonchange-oriented longitudinal
studies.
Despite these evident strengths, the work was not without lim-
itations. Perhaps most important was our focus on a single OXTR
SNP. Thus, future studies should include more SNPs in this gene
to better understand gene-environment interplay involving OXTR.
To be appreciated, however, is that understanding of the function-
ality of other SNPs related to OXTR is presently even more limited
than that of rs53576, which is the most-researched OXTR SNP and
the reason why we focused on the allelic variation that we did.
Another limitation involves the focus on (mostly) mothers, raising
issues of whether the effect of change in paternal behavior would
also be genetically moderated in the manner detected in Study
1—in either a Norwegian or American sample. Finally and most
obviously, this report was limited by its very strength—inclusion
of samples from different countries, thereby making it somewhat
difficult, as we have noted, to draw conclusions about the meaning
of the apparent the “failure” to replicate. The same, of course, is
true of the absence of exactly the same measurements across
studies These differences in particular highlight the need for ad-
ditional research seeking to replicate the G⫻E findings discerned
in Study 1 in other Norwegian communities or even perhaps the
failure to replicate in other American communities.
References
Achenbach, T. M. (1991). Manual for the Child Behavior Checklist/4 –18
and 1991 profile. Burlington, VT: University of Vermont, Department of
Psychiatry.
Allen, J. S. (2009). The lives of the brain. Human evolution and the organ
of mind. Cambridge, MA, and London, England: The Belknap Press of
Harvard University Press. http://dx.doi.org/10.4159/9780674053496
Andre, H. J., & Heien, T. (2001). Four worlds of welfare state attitudes?
A comparison of Germany, Norway, and the United States. European
Sociological Review, 17, 337–356. http://dx.doi.org/10.1093/esr/17.4
.337
Babineau, V., Green, C. G., Jolicoeur-Martineau, A., Bouvette-Turcot,
A. A., Minde, K., Sassi, R.,...theMAVAN project. (2015). Prenatal
depression and 5-HTTLPR interact to predict dysregulation from 3 to 36
months—A differential susceptibility model. Journal of Child Psychol-
ogy and Psychiatry, 56, 21–29. http://dx.doi.org/10.1111/jcpp.12246
Bakermans-Kranenburg, M. J., & van Ijzendoorn, M. H. (2008). Oxytocin
receptor (OXTR) and serotonin transporter (5-HTT) genes associated
with observed parenting. Social Cognitive and Affective Neuroscience, 3,
128–134. http://dx.doi.org/10.1093/scan/nsn004
Bakermans-Kranenburg, M. J., & van Ijzendoorn, M. H. (2011). Differen-
tial susceptibility to rearing environment depending on dopamine-related
genes: New evidence and a meta-analysis. Development and Psychopa-
thology, 23, 39–52. http://dx.doi.org/10.1017/S0954579410000635
Bakermans-Kranenburg, M. J., & van Ijzendoorn, M. H. (2014). A socia-
bility gene? Meta-analysis of oxytocin receptor genotype effects in
humans. Psychiatric Genetics, 24, 45–51. http://dx.doi.org/10.1097/
YPG.0b013e3283643684
Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice
Hall.
Belsky, J. (Ed.), (2005). Differential susceptibility to rearing influence: An
evolutionary hypothesis and some evidence. New York, NY: Guilford
Press.
Belsky, J., Bakermans-Kranenburg, M. J., & van Ijzendoorn, M. H. (2007).
For better and for worse: Differential susceptibility to environmental
influences. Current Directions in Psychological Science, 16, 300–304.
http://dx.doi.org/10.1111/j.1467-8721.2007.00525.x
Belsky, J., & Pluess, M. (2009). Beyond diathesis stress: Differential
susceptibility to environmental influences. Psychological Bulletin, 135,
885–908. http://dx.doi.org/10.1037/a0017376
Belsky, J., & Pluess, M. (2013). Beyond risk, resilience and dysregulation:
Phenotypic plasticity and human development. Development and Psy-
chopathology, 25, 1243–1261.
Belsky, J., Pluess, M., & Widaman, K. F. (2013). Confirmatory and
competitive evaluation of alternative gene-environment interaction hy-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1312 HYGEN, BELSKY, LI, STENSENG, GU
¨ZEY, AND WICHSTRØM
potheses. Journal of Child Psychology and Psychiatry, 54, 1135–1143.
http://dx.doi.org/10.1111/jcpp.12075
Birch, S. H., & Ladd, G. W. (1997). The teacher-child relationship and
children’s early school adjustment. Journal of School Psychology, 35,
61–79. http://dx.doi.org/10.1016/S0022-4405(96)00029-5
Bjørseth, Å., & Wichstrøm, L. (2016). Effectiveness of Parent-Child In-
teraction Therapy (PCIT) in the treatment of young children’s behavior
problems. A randomized controlled study. PLoS ONE, 11, e0159845.
http://dx.doi.org/10.1371/journal.pone.0159845
Booth, C. L., Kelly, J. F., Spieker, S. J., & Zuckerman, T. G. (2003).
Toddlers’ attachment Security to child-Care providers: The Safe and
secure scale. Early Education and Development, 14, 83–100. http://dx
.doi.org/10.1207/s15566935eed1401_6
Bowlby, J. (1973). Attachment and loss- VOL. 2: Separation. New York,
NY: Basic Books.
Bradley, B., Westen, D., Mercer, K. B., Binder, E. B., Jovanovic, T., Crain,
D.,...Heim, C. (2011). Association between childhood maltreatment
and adult emotional dysregulation in a low-income, urban, African
American sample: Moderation by oxytocin receptor gene. Development
and Psychopathology, 23, 439–452. http://dx.doi.org/10.1017/
S0954579411000162
Bresin, K., Sima Finy, M., & Verona, E. (2013). Childhood emotional
environment and self-injurious behaviors: The moderating role of the
BDNF Val66Met polymorphism. Journal of Affective Disorders, 150,
594–600. http://dx.doi.org/10.1016/j.jad.2013.01.050
Brinkmeyer, M. Y., & Eyberg, S. M. (2003). Parent-child interaction
therapy for oppositional children. In A. E. Kazdin & J. R. Weisz (Eds.),
Evidenced-based psychotherapies for children and adolescents (pp.
204–223). New York, NY: Guilford Press.
Brown, G. W., Ban, M., Craig, T. K., Harris, T. O., Herbert, J., & Uher, R.
(2013). Serotonin transporter length polymorphism, childhood maltreat-
ment, and chronic depression: A specific gene-environment interaction.
Depression and Anxiety, 30, 5–13. http://dx.doi.org/10.1002/da.21982
Carter, C. S. (2014). Oxytocin pathways and the evolution of human
behavior. Annual Review of Psychology, 65, 17–39. http://dx.doi.org/10
.1146/annurev-psych-010213-115110
Caspi, A., Hariri, A. R., Holmes, A., Uher, R., & Moffitt, T. E. (2010).
Genetic sensitivity to the environment: The case of the serotonin trans-
porter gene and its implications for studying complex diseases and traits.
The American Journal of Psychiatry, 167, 509–527. http://dx.doi.org/
10.1176/appi.ajp.2010.09101452
Caspi, A., McClay, J., Moffitt, T. E., Mill, J., Martin, J., Craig, I. W.,...
Poulton, R. (2002). Role of genotype in the cycle of violence in mal-
treated children. Science, 297, 851–854. http://dx.doi.org/10.1126/
science.1072290
Caspi, A., Sugden, K., Moffitt, T. E., Taylor, A., Craig, I. W., Harrington,
H.,...Poulton, R. (2003). Influence of life stress on depression:
Moderation by a polymorphism in the 5-HTT gene. Science, 301, 386–
389. http://dx.doi.org/10.1126/science.1083968
Chen, F. S., Kumsta, R., von Dawans, B., Monakhov, M., Ebstein, R. P.,
& Heinrichs, M. (2011). Common oxytocin receptor gene (OXTR)
polymorphism and social support interact to reduce stress in humans.
Proceedings of the National Academy of Sciences of the United States of
America, 108, 19937–19942. http://dx.doi.org/10.1073/pnas.11130
79108
Choate, M. L., Pincus, D. B., Eyberg, S. M., & Barlow, D. H. (2005).
Parent-child interaction therapy for treatment of separation anxiety dis-
order in young children: A pilot study. Cognitive and Behavioral Prac-
tice, 12, 126–135. http://dx.doi.org/10.1016/S1077-7229(05)80047-1
Cicchetti, D., Rogosch, F. A., Hecht, K. F., Crick, N. R., & Hetzel, S.
(2014). Moderation of maltreatment effects on childhood borderline
personality symptoms by gender and oxytocin receptor and FK506
binding protein 5 genes. Development and Psychopathology, 26, 831–
849. http://dx.doi.org/10.1017/S095457941400042X
Crone, M. R., Vogels, A. G. C., Hoekstra, F., Treffers, P. D. A., &
Reijneveld, S. A. (2008). A comparison of four scoring methods based
on the parent-rated Strengths and Difficulties Questionnaire as used in
the Dutch preventive child health care system. BMC Public Health, 8,
106. http://dx.doi.org/10.1186/1471-2458-8-106
De Laet, S., Doumen, S., Vervoort, E., Colpin, H., Van Leeuwen, K.,
Goossens, L., & Verschueren, K. (2014). Transactional links between
teacher-child relationship quality and perceived versus sociometric pop-
ularity: A three-wave longitudinal study. Child Development, 85, 1647–
1662. http://dx.doi.org/10.1111/cdev.12216
DeMulder, E. K., Denham, S., Schmidt, M., & Mitchell, J. (2000). Q-sort
assessment of attachment security during the preschool years: Links
from home to school. Developmental Psychology, 36, 274–282. http://
dx.doi.org/10.1037/0012-1649.36.2.274
Duncan, L. E., & Keller, M. C. (2011). A critical review of the first 10
years of candidate gene-by-environment interaction research in psychi-
atry. The American Journal of Psychiatry, 168, 1041–1049. http://dx.doi
.org/10.1176/appi.ajp.2011.11020191
Ellis, B. J., Boyce, W. T., Belsky, J., Bakermans-Kranenburg, M. J., & van
Ijzendoorn, M. H. (2011). Differential susceptibility to the environment:
An evolutionary—Neurodevelopmental theory. Development and Psy-
chopathology, 23, 7–28. http://dx.doi.org/10.1017/S0954579410000611
Eyberg, S. M., Nelson, M. M., Duke, M., & Boggs, S. R. (Eds.). (2005).
Manual for the parent-child interaction coding system. Gainesville, FL:
University of Florida.
Feldman, R., Monakhov, M., Pratt, M., & Ebstein, R. P. (2016). Oxytocin
pathway genes: Evolutionary ancient system impacting on human affil-
iation, sociality, and psychopathology. Biological Psychiatry, 79, 174–
184. http://dx.doi.org/10.1016/j.biopsych.2015.08.008
Goossens, F. A., & van IJzendoorn, M. H. (1990). Quality of infants’
attachments to professional caregivers: Relation to infant-parent attach-
ment and day-care characteristics. Child Development, 61, 832–837.
http://dx.doi.org/10.2307/1130967
Hamer, D. (2002). Genetics. Rethinking behavior genetics. Science, 298,
71–72. http://dx.doi.org/10.1126/science.1077582
Hamre, B. K., & Pianta, R. C. (2001). Early teacher-child relationships and
the trajectory of children’s school outcomes through eighth grade. Child
Development, 72, 625–638. http://dx.doi.org/10.1111/1467-8624.00301
Henricsson, L., & Rydell, A. (2004). Elementary school children with
behavior problems: Teacher-child relations and self-perception. A pro-
spective study. Merrill-Palmer Quarterly, 50, 111–138. http://dx.doi
.org/10.1353/mpq.2004.0012
Hostinar, C. E., Cicchetti, D., & Rogosch, F. A. (2014). Oxytocin receptor
gene polymorphism, perceived social support, and psychological symp-
toms in maltreated adolescents. Development and Psychopathology, 26,
465–477. http://dx.doi.org/10.1017/S0954579414000066
Howes, C., & Hamilton, C. E. (1992). Children’s relationships with care-
givers: Mothers and child care teachers. Child Development, 63, 859–
866. http://dx.doi.org/10.2307/1131238
Howes, C., & Ritchie, S. (1999). Attachment organizations in children with
difficult life circumstances. Development and Psychopathology, 11,
251–268. http://dx.doi.org/10.1017/S0954579499002047
Hygen, B. W., Belsky, J., Stenseng, F., Lydersen, S., Guzey, I. C., &
Wichstrøm, L. (2015). Child exposure to serious life events, COMT, and
aggression: Testing differential susceptibility theory. Developmental
Psychology, 51, 1098–1104. http://dx.doi.org/10.1037/dev0000020
Hygen, B. W., Guzey, I. C., Belsky, J., Berg-Nielsen, T. S., & Wichstrøm,
L. (2014). Catechol-O-methyltransferase Val158Met genotype moder-
ates the effect of disorganized attachment on social development in
young children. Development and Psychopathology, 26, 947–961. http://
dx.doi.org/10.1017/S0954579414000492
Insel, T. R., & Collins, F. S. (2003). Psychiatry in the genomics era. The
American Journal of Psychiatry, 160, 616–620. http://dx.doi.org/10
.1176/appi.ajp.160.4.616
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1313
OXTR, PARENTING, AND STUDENT–TEACHER RELATIONSHIPS
Jasny, B. R., Chin, G., Chong, L., & Vignieri, S. (2011). Data replication
& reproducibility. Again, and again, and again...Introduction. Science,
334, 1225. http://dx.doi.org/10.1126/science.334.6060.1225
Jerome, E. M., Hamre, B. K., & Pianta, R. C. (2009). Teacher-child
relationships from kindergarten to sixth grade: Early childhood predic-
tors of teacher-perceived conflict and closeness. Social Development, 18,
915–945. http://dx.doi.org/10.1111/j.1467-9507.2008.00508.x
Karg, K., Burmeister, M., Shedden, K., & Sen, S. (2011). The serotonin
transporter promoter variant (5-HTTLPR), stress, and depression meta-
analysis revisited: Evidence of genetic moderation. Archives of General
Psychiatry, 68, 444–454. http://dx.doi.org/10.1001/archgenpsychiatry
.2010.189
Kaufman, J., Gelernter, J., Kaffman, A., Caspi, A., & Moffitt, T. (2010).
Arguable assumptions, debatable conclusions. Biological Psychiatry, 67,
e19–e20. http://dx.doi.org/10.1016/j.biopsych.2009.07.041
Klahr, A. M., Klump, K., & Burt, S. A. (2015). A constructive replication
of the association between the oxytocin receptor genotype and parenting.
Journal of Family Psychology, 29, 91–99. http://dx.doi.org/10.1037/
fam0000034
Krueger, F., Parasuraman, R., Iyengar, V., Thornburg, M., Weel, J., Lin,
M.,...Lipsky, R. H. (2012). Oxytocin receptor genetic variation
promotes human trust behavior. Frontiers in Human Neuroscience, 6, 4.
http://dx.doi.org/10.3389/fnhum.2012.00004
Kryski, K. R., Smith, H. J., Sheikh, H. I., Singh, S. M., & Hayden, E. P.
(2014). Evidence for evocative gene-environment correlation between
child oxytocin receptor (OXTR) genotype and caregiver behavior. Per-
sonality and Individual Differences, 64, 107–110. http://dx.doi.org/10
.1016/j.paid.2014.02.028
Ladd, G. W., Birch, S. H., & Buhs, E. S. (1999). Children’s social and
scholastic lives in kindergarten: Related spheres of influence? Child
Development, 70, 1373–1400. http://dx.doi.org/10.1111/1467-8624
.00101
Lee, H.-J., Macbeth, A. H., Pagani, J. H., & Young, W. S., III. (2009).
Oxytocin: The great facilitator of life. Progress in Neurobiology, 88,
127–151.
Lenze, S. N., Pautsch, J., & Luby, J. (2011). Parent-child interaction
therapy emotion development: A novel treatment for depression in
preschool children. Depression and Anxiety, 28, 153–159. http://dx.doi
.org/10.1002/da.20770
Li, J., Zhao, Y., Li, R., Broster, L. S., Zhou, C., & Yang, S. (2015).
Association of oxytocin receptor Gene (OXTR) rs53576 polymorphism
with sociality: A Meta-analysis. PLoS ONE, 10, e0131820. http://dx.doi
.org/10.1371/journal.pone.0131820
Maroun, M., & Wagner, S. (2016). Oxytocin and memory of emotional
stimuli: Some dance to remember, some dance to forget. Biological
Psychiatry, 79, 203–212. http://dx.doi.org/10.1016/j.biopsych.2015.07
.016
McInnis, O. A., McQuaid, R. J., Matheson, K., & Anisman, H. (2015). The
moderating role of an oxytocin receptor gene polymorphism in the
relation between unsupportive social interactions and coping profiles:
Implications for depression. Frontiers in Psychology, 6, 1133. http://dx
.doi.org/10.3389/fpsyg.2015.01133
Meyer-Lindenberg, A., Domes, G., Kirsch, P., & Heinrichs, M. (2011).
Oxytocin and vasopressin in the human brain: Social neuropeptides for
translational medicine. Nature Reviews Neuroscience, 12, 524–538.
http://dx.doi.org/10.1038/nrn3044
Moore, S. R., & Thoemmes, F. (2016). What is the biological reality of
gene-environment interaction estimates? An assessment of bias in de-
velopmental models. Journal of Child Psychology and Psychiatry, 57,
1258–1267. http://dx.doi.org/10.1111/jcpp.12579
Murray, C., & Greenberg, M. T. (2001). Relationships with teachers and
bonds with school: Social emotional adjustment correlates for children
with and without disabilities. Psychology in the Schools, 38, 25–41.
http://dx.doi.org/10.1002/1520-6807(200101)
Murray, C., & Zvoch, K. (2011). Teacher-student relationships among
behaviorally at-risk African American youth from low-income back-
grounds: Student perceptions, teacher perceptions, and Socioemotional
adjustment correlates. Journal of Emotional and Behavioral Disorders,
19, 41–54. http://dx.doi.org/10.1177/1063426609353607
Muthén, L. K., & Muthén, B. O. (1998–2012). Mplus user’s guide (7th
ed.). Los Angeles, CA: Author.
NICHD Early Child Care Research Network. (Ed.). (2005). Child care and
child development: Results from the NICHD study of early child care
and youth development. New York, NY: Guilford Press.
Nixon, R. D., Sweeney, L., Erickson, D. B., & Touyz, S. W. (2003).
Parent-child interaction therapy: A comparison of standard and abbre-
viated treatments for oppositional defiant preschoolers. Journal of Con-
sulting and Clinical Psychology, 71, 251–260. http://dx.doi.org/10.1037/
0022-006X.71.2.251
O’Connor, E., & McCartney, K. (2006). Testing associations between
young children’s relationships with mothers and teachers. Journal of
Educational Psychology, 98, 87–98. http://dx.doi.org/10.1037/0022-
0663.98.1.87
Organisation for Economic Co-operation and Development (OECD).
(2012). Programme for international student assessment (PISA). Results
from PISA 2012. United States. Retrieved from http://www.oecd
.org/pisa/keyfindings/PISA-2012-results-US.pdf
Organisation for Economic Co-operation and Development (OECD).
(2016). PISA 2015 results (Volume II): Policies and practices for suc-
cessful schools. Paris: PISA, OECD Publishing.
Patterson, G. R., DeBaryshe, B. D., & Ramsey, E. (1989). A developmental
perspective on antisocial behavior. American Psychologist, 44, 329–
335. http://dx.doi.org/10.1037/0003-066X.44.2.329
Pianta, R. C. (2001). STRS: Student–Teacher Relationship Scale: Profes-
sional manual. Lutz, FL: Psychological Assessment Resources.
Plomin, R., DeFries, J. C., McClearn, G. E., & McGuffin, P. (2008).
Behavioral genetics. New York, NY: Worth Publishers.
Reitz, E., Dekovic, M., Meijer, A. M., & Engels, R. (2006). Longitudinal
relations among parenting, best friends, and early adolescent problem
behavior: Testing bidirectional effects. The Journal of Early Adoles-
cence, 26, 272–295. http://dx.doi.org/10.1177/0272431606288591
Risch, N., Herrell, R., Lehner, T., Liang, K. Y., Eaves, L., Hoh, J.,...
Merikangas, K. R. (2009). Interaction between the serotonin transporter
gene 5-HTTLPR, stressful life events, and risk of depression: Meta-
analysis. Journal of the American Medical Association, 301, 2462–
24671. Incorrect Wording in: Interaction between the serotonin trans-
porter gene (5-HTTLPR), stressful life events, and risk of depression: A
meta-analysis. Journal of the American Medical Association, 302, 492.
http://dx.doi.org/10.1001/jama.302.5.492
Rodrigues, S. M., Saslow, L. R., Garcia, N., John, O. P., & Keltner, D.
(2009). Oxytocin receptor genetic variation relates to empathy and stress
reactivity in humans. Proceedings of the National Academy of Sciences
of the United States of America, 106, 21437–21441. http://dx.doi.org/10
.1073/pnas.0909579106
Roorda, D. L., Koomen, H. M. Y., Spilt, J. L., & Oort, F. J. (2011). The
influence of affective teacher-student relationships on students’ school
engagement and achievement: A meta-analytic approach. Review of
Educational Research, 81, 493–529.
Rudasill, K. M., Reio, T. G., Jr., Stipanovic, N., & Taylor, J. E. (2010). A
longitudinal study of student-teacher relationship quality, difficult tem-
perament, and risky behavior from childhood to early adolescence.
Journal of School Psychology, 48, 389–412. http://dx.doi.org/10.1016/
j.jsp.2010.05.001
Rutter, M. (2010). Gene-environment interplay. Depression and Anxiety,
27, 1–4. http://dx.doi.org/10.1002/da.20641
Ryan, M. J. (2011). Replication in field biology: The case of the frog-
eating bat. Science, 334, 1229–1230. http://dx.doi.org/10.1126/science
.1214532
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1314 HYGEN, BELSKY, LI, STENSENG, GU
¨ZEY, AND WICHSTRØM
Rydell, A. M., Bohlin, G., & Thorell, L. B. (2005). Representations of
attachment to parents and shyness as predictors of children’s relation-
ships with teachers and peer competence in preschool. Attachment &
Human Development, 7, 187–204. http://dx.doi.org/10.1080/146167
30500134282
Salvatore, J. E., & Dick, D. M. (2015). Gene-environment interplay: Where
we are, where we are going. Journal of Marriage and Family, 77,
344–350. http://dx.doi.org/10.1111/jomf.12164
Satorra, A. (2000). Scaled and adjusted restricted tests in multi-sample
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 Aca-
demic Publishers. http://dx.doi.org/10.1007/978-1-4615-4603-0_17
Schafer, J. L., & Graham, J. W. (2002). Missing data: Our view of the state
of the art. Psychological Methods, 7, 147–177. http://dx.doi.org/10
.1037/1082-989X.7.2.147
Silver, R. B., Measelle, J. R., Armstrong, J. M., & Essex, M. J. (2005).
Trajectories of classroom externalizing behavior: Contributions of child
characteristics, family characteristics, and the teacher-child relationship
during the school transition. Journal of School Psychology, 43, 39–60.
http://dx.doi.org/10.1016/j.jsp.2004.11.003
Skalická, V., Stenseng, F., & Wichstrom, L. (2015). Reciprocal relations
between student-teacher conflict, children’s social skills and externaliz-
ing behavior: A three-wave longitudinal study from preschool to third
grade. International Journal of Behavioral Development, 39, 413–425.
http://dx.doi.org/10.1177/0165025415584187
Thompson, S. M., Hammen, C., Starr, L. R., & Najman, J. M. (2014).
Oxytocin receptor gene polymorphism (rs53576) moderates the inter-
generational transmission of depression. Psychoneuroendocrinology, 43,
11–19. http://dx.doi.org/10.1016/j.psyneuen.2014.01.012
Tost, H., Kolachana, B., Hakimi, S., Lemaitre, H., Verchinski, B. A.,
Mattay, V. S.,...Meyer-Lindenberg, A. (2010). A common allele in the
oxytocin receptor gene (OXTR) impacts prosocial temperament and
human hypothalamic-limbic structure and function. Proceedings of the
National Academy of Sciences of the United States of America, 107,
13936–13941. http://dx.doi.org/10.1073/pnas.1003296107
Uher, R., & McGuffin, P. (2008). The moderation by the serotonin trans-
porter gene of environmental adversity in the aetiology of mental illness:
Review and methodological analysis. Molecular Psychiatry, 13, 131–
146. http://dx.doi.org/10.1038/sj.mp.4002067
Uher, R., & McGuffin, P. (2010). The moderation by the serotonin trans-
porter gene of environmental adversity in the etiology of depression:
2009 update. Molecular Psychiatry, 15, 18–22. http://dx.doi.org/10
.1038/mp.2009.123
Webster-Stratton, C., & Hammond, M. (1999). Marital conflict manage-
ment skills, parenting style, and early-onset conduct problems: Processes
and pathways. Journal of Child Psychology and Psychiatry, 40, 917–
927. http://dx.doi.org/10.1111/1469-7610.00509
Webster-Stratton, C., Reid, M. J., & Hammond, M. (2001). Preventing
conduct problems, promoting social competence: A parent and teacher
training partnership in head start. Journal of Clinical Child Psychology,
30, 283–302. http://dx.doi.org/10.1207/S15374424JCCP3003_2
Widaman, K. F., Helm, J. L., Castro-Schilo, L., Pluess, M., Stallings,
M. C., & Belsky, J. (2012). Distinguishing ordinal and disordinal inter-
actions. Psychological Methods, 17, 615–622. http://dx.doi.org/10.1037/
a0030003
Wyatt Kaminski, J. W., Valle, L. A., Filene, J. H., & Boyle, C. L. (2008).
A meta-analytic review of components associated with parent training
program effectiveness. Journal of Abnormal Child Psychology, 36,
567–589. http://dx.doi.org/10.1007/s10802-007-9201-9
Wyrick, A. J., & Rudasill, K. M. (2009). Parent involvement as a predictor
of teacher-child relationship quality in third grade. Early Education and
Development, 20, 845–864. http://dx.doi.org/10.1080/1040928080
2582803
Zuckerman, M. (1999). Vulnerability to psychopathology: A biosocial
model. Washington, DC: American Psychological Association. http://dx
.doi.org/10.1037/10316-000
Received April 29, 2016
Revision received January 31, 2017
Accepted March 8, 2017 䡲
E-Mail Notification of Your Latest Issue Online!
Would you like to know when the next issue of your favorite APA journal will be available
online? This service is now available to you. Sign up at https://my.apa.org/portal/alerts/ and you will
be notified by e-mail when issues of interest to you become available!
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1315
OXTR, PARENTING, AND STUDENT–TEACHER RELATIONSHIPS
Content uploaded by Beate Wold Hygen
Author content
All content in this area was uploaded by Beate Wold Hygen on May 08, 2017
Content may be subject to copyright.
