A Longitudinal Cross-Lagged Panel Investigation of the Association between Parenting Practices and Children’s Externalizing Behavior
Karen Fischer
Research Master Psychology University of Amsterdam
Date: 10.09.2016
Studentnumber: 10756701 Email: fischer.karen@gmx.net
Supervisors:
Prof. Dr. Geertjan Overbeek Dr. Helle Larsen
Second Assessor: Prof. Dr. Reinout Wiers
Abstract
A crucial factor implied in the development of children’s externalizing problems are dysfunctional parenting practices. Yet, the child may also function as an active agent in evoking aversive parenting behaviors. A cross-lagged panel analysis was conducted, with data available from a three-wave longitudinal study on 387 at-risk families with children aged 4-8, to examine the hypothesized bi-directional associations between parenting prctices and children’s externalizing behavior over time. In line with prior research, it was furthermore investigated whether certain children are genetically more susceptible to both aversive and nurturant parental influences. Results from cross-lagged panel analyses in R showed that there were no bi-directional associations between parent and child behavior over time, and that child genotype did not moderate these associations. These results demonstrate that gene-environment interactions are difficult to replicate and call for further research on the reciprocal relationship between parent and child behavior.
A longitudinal cross-lagged panel investigation of the association between parenting practices and children’s externalizing behavior
Early onset of disruptive behavior is a robust predictor of impairing externalizing psychopathology in later life (von Stumm et al., 2011). Conduct problems during early childhood predict a range of maladaptions in later life, such as academic problems, impaired social relationships, cognitive development (e.g. Smith et al., 2014; Campbell et al., 2006). Not only do externalizing behaviors pose a problem for the development of the individual, but they also constitute a major societal problem. In most cases, severity of externalizing behaviors persists and exacerbates over time if left untreated (e.g. Mesman, Bongers, & Koot, 2001; Vaughn, Salas-wright, Delisi, & Maynard, 2013), leading to destructive later life outcomes, such as criminal activity, unemployment, and poor social relationships (Moffitt et al., 2011).
Problems in emotion and behavior regulation are present in a majority of individuals with disruptive behavior problems (Moffitt et al., 2011). Children may experience difficulties focusing and shifting attention as a result of overarousal, which in turn diminishes self-regulatory abilities (Eisenberg et al., 2005). A diminished sense of self-control is closely linked to aversive life outcomes such as poor health behaviors, psychiatric disorders, and early mortality (e.g. Kern & Friedman, 2008, Moffitt et al., 2011). Especially an early onset of disruptive behavior in childhood has been shown to represent a marked risk factor to develop conduct problems at clinical levels in later life. Therefore, it comes to no surprise that in up to 60 % of adult clinical diagnoses
et al., 2003). In light of the high economic and societal burden that is associated with respective dysfunctional life and behavior outcomes in later adulthood, it is evident that identifying its main causes and intervening at an early age is indispensible.
One of the crucial factors implied in the development of children’s externalizing problems are dysfunctional parenting practices. A plethora of research and meta-analytic evidence documents that children’s externalizing behavior and parenting practices are significantly related (e.g. Rothbaum & Weisz, 1994). From the research literature, generally 3 models can be deduced about how parenting and children’s externalizing behavior are associated: a) children’s externalizing behavior is a product of dysfunctional parenting, b) externalizing behavior causes dysfunctional parenting, and c) there is a reciprocal relationship between children’s externalizing behavior and dysfunctional parenting.
Research on the first model has shown that dysfunctional parenting practices play a key role in eliciting externalizing and disruptive behavior problems in children (e.g., Miner & Clarke-Stewart, 2008). Specifically, many studies showed that lack of reward-based parenting, parental punishment, and inconsistent discipline function as longitudinal predictors of externalizing behaviors (e.g., Bor, Sanders, Markie-Dadds, 2002; Karreman, Van Tuijl, Van Aken, & Dekovic, 2006). For example, a review by Campbell et al. (2000) revealed that quality of caregiving when children were 2 years old predicted severity of conduct problems 6 years later by caregivers and teacher-reports (Smith et al., 2014). As Smith et al. (2014) describe, children transfer the dysfunctional feedback received from their parents to environments outside the home, where they develop aversive interactions with other authorative figures such as teachers or peers. Not
surprisingly then, it has been shown that coercive, harsh, and overcontrolling parenting predicts a range of adjustment and conduct problems in elementary school (e.g.,
Campbell et al., 2000; Shaw, Owens, Giovannelli, & Winslow, 2001).
In the second model, children are regarded as active agents that shape their own development, by eliciting certain parenting behaviors (Smith et al. 2014). Studies show that difficult, noncompliant children evoke more dissapproving and even aversive physical feedback from their caregivers (e.g. Braungart-Rieker, Garwood, & Stifter, 1997; Hartup & Van Lieshout, 1995). For example in the study by Braungart, Garwood, and Stifter (1997), mother and child were observed in four guidance and control tasks to assess the degree of child compliance in a variety of situations. The results revealed that mothers of more difficult children (higher in negative reactivity and passive
noncompliance) implemented more controlling parenting practices and less guidance in the interaction with their child. Braungart et al. (1997) therefore postulated that high negatively reactive children may evoke a notion in their mothers that harsh parenting practices are the last resort option to regain control and elicit compliance in their children.
The third model comprises longitudinal studies that illustrate a bi-directional relationship of dysfunctional parenting behavior and children’s externalizing behavior. For example, the longitudinal study by De Haan, Prinzie, & Dekovic (2012) followed children from 9-17 years of age to investigate the relationship between overreactive and warm parenting and aggressive/rule-breaking behaviors. The results of cross-lagged analyses in their study demonstrated a bi-directional relationship between parental overreactivity and children’s aggression/rule breaking, independent of the gender of the
child or parent. These findings suppport the notion that both parent and child evoke aversive behaviors in their counterparts respectively, and suggest that this reciprocity functions to sustain a cycle of dysfunctional parent-child interactions.
A plethora of longitudinal studies corroborate this reciprocity (e.g. Keijsers,
Loeber, Branje, Meeus, 2011; Roche, Ghazarian, Little, & Leventhal, 2010) and crucially illustrate the more serious sequelae in later childhood and adolscence. Parenting thereby acts as an environmental factor that children not only get influenced by, but also actively shape. Patterson (2002) proposed a transactional model, in which child noncompliance and aggression evokes conflict with parents who dispose over ineffective and emotional strategies to deal with the misconduct. Such ineffective reactions entail withdrawal and giving in (e.g. Patterson & Cobb, 1971), or entail coercive, harsh, punitive parenting (e.g., Odgers et al., 2008). Crucially, these parenting practices encourage more aversive behavior in children, creating a reinforcing and self-maintaining cycle of child
misconduct and parent coersion (Patterson, 2002).
Researchers have aimed to investigate what potential individual differences may underlie the relationship between parenting behavior and children’s externalizing
behavior. The empirically well-supported perspective of children’s differential
susceptibility (Boyce & Ellis, 2005; Belsky & Pluess, 2009) holds that the same children
which are most affected by an adverse environment, such as dyfsunctional parenting, may also reap the greatest benefits from a supportive environment, such as with positive parenting. This idea is of great interest to parenting research, as the same children that are at risk of developing the most unfavorably in a dysfunctional family environment may profit the most from improved parenting (e.g. through intervention).
Certain endogenous susceptibility factors have shown to underlie this “for better and for worse” differential susceptibility phenomenon. Leading scholars in the field argue that differential susceptibility is gene-based, and a solid body of meta-analytic evidence supports this. Specifically, polymorphisms in the serotonin transporter (e.g., Karg, Burmeister, Shedden & Sen, 2011), dopamine receptor genes (e.g.,
Bakermans-Kranenburg & Van IJzendoorn, 2011), and polymorphisms that degrade the latter (e.g. Byrd & Manuck, 2014) have consistently emerged as vulnerability genes that render children more sensitive to both positive and negative environmental influences (e.g. Bakermans-Kranenburg & van IJzendoorn, 2015). Polymorphisms related to dopamine functioning have shown to modulate children’s reward- (Schultz et al., 2010; Bakermans-Kranenburg & Van Ijzendoorn, 2011) and punishment-based learning (Matthys et al. 2013; 2012). Therefore, dopamine pathways have been proposed as crucial in explaining children’s susceptibility to reward- or punishment-based parenting (Davies et al., 2015; Posner & Rothbart, 2009). In line with the differential susceptibility perspective, children disposing over certain dopamine-related polymorphisms have shown to be more reactive to parent’s punishment cues, but also more responsive to reward cues (Belsky & Pluess, 2009; 2013). Especially in at-risk populations, parenting is characterized by high
punishment and low reward parenting behavior, rendering especially genetically
susceptible children prone to the adverse effects of dysfunctional parenting. This explains why children with a dopaminergic susceptibility have shown to develop more
externalizing behavior in a dysfunctional parenting environment than children without such a susceptibility (e.g. Chhangur, Weeland, Overbeek, Matthys & Orobio de Castro, 2012). The present study will examine relevant polymorphisms related to dopamine
functioning (DRD4, DRD2, DAT1, MAOA, COMT), which have been associated with children’s heightened susceptibility to both positive and negative environmental
influences (Belsky & Pluess, 2009; 2013).
It should be noted that, when viewing differential susceptibility in light of the bi-directionality of parent-child interactions, differential susceptible children may also elicit stronger reactions from their parents, by being more emotionally reactive. For example, children disposing over more dopamine related genetic polymorphisms may exhibit more externalizing behavior at baseline and thereby elicit more punishing and harsh discipline from their parents (e.g. Klahr et al., 2013; Braungart et al., 1997). Furthermore,
genetically susceptible children have shown to exhibit greater emotional reactivity to parenting, which may in turn elicit both negative and positive parenting behaviors to a greater extent than less emotionally reactive children (e.g. Weeland, Overbeek, Orobio de Castro, & Matthys, 2015; Matthys, Vanderschuren, & Schutter, 2013).
Despite the rapid growing body of evidence, research on gene-environment interactions in children’s externalizing problem behavior is only in its infancy. This can be attributed to a large extent to the difficulty of identifying the responsible genes or combinations of genetic pathways of complex behavioral outcomes (Dick, 2011). The existing gene-environment studies provide preliminary results at best, because a majority of those studies relied on an overly simplistic candidate-gene approach. There is
increasing criticism towards social scientists that the investigation of a single candidate gene is outdated as it disregards genetic complexity (Dick et al., 2011). The functional contribution of dopamine genes in shaping parent-child G×E interactions is probably polygenetic in nature (Dick et al., 2015). Therefore, a superior approach is to identify
multiple genetic polymorphisms in a dopaminergic-pathway rather than single candidate-genes, in order to illuminate which (combination of) dopaminergic pathway genes modulate children’s reward- (Schultz, 2010; Bakermans-Kranenburg & Van Ijzendoorn, 2011) and punishment-based learning (Matthys et al., 2013; Matthys et al., 2012). By investigating the genetic roots of how children respond and act to parenting stategies, light is shed onto how parenting-interventions can be adapted to the individual child to derive maximum profit in reducing or preventing externalizing problem behavior. The Present Study
The present study made use of the three-wave longitudinal data from the ORCHIDS study (Chhangur et al., 2012) on 387 at-risk families with children 4-8 years old, examining the relationship between parenting practices and children’s externalizing behavior. An advantage of the present study was that we used an at-risk sample, which inherently disposes over a higher proportion of dysfunctional parent and child behavior. This increases power and yields a less skewed variance (Van IJzendoorn & Bakermans-Kranenburg, 2015). Furthermore, instead of examining a single candidate gene, a
polygenic index was created, composed of several susceptibility allelic variants. Thereby a comprehensive composite score of several relevant polymorphisms can be yielded to provide a more complex measure of susceptibility than a single candidate gene. Lastly, in the present study we did not only investigate the association between negative parenting practices, but also that of positive parenting practices and children’s externalizing behavior. In this manner, a stringent test of longitudinal-bidirectional relationships for both positive and negative dimensions of parenting can was conducted and provided a true test of differential susceptibility.
The main aim of the present study was two-fold. First, using a cross-lagged longitudinal panel design, we investigated the longitudinal bi-directional associations between parenting practices and children’s externalizing behavior. This was done in two analyses. In line with the two dimensions of parenting, nurturant and coercive parenting (Patterson, 1986), we investigated the relationship of both negative and positive parenting to children’s externalizing behavior. In the first analysis we investigated the linkage between negative parenting behavior and children’s externalizing behavior. It was
hypothesized that harsh, punitive, and inconsistent disciplining parenting predicts greater children’s externalizing behavior. In line with the bi-directional perspective, we
furthermore hypothesized that children’s externalizing behavior predicts greater negative parenting. In the second analyses, we examined the relationship between positive
parenting behavior and children’s externalizing behavior. We hypothesized that greater rewarding and warm parenting predicts less externalizing behavior among children. Analogous, we hypothesized that less externalizing behavior predicts more positive parenting. The second aim of the study was to examine whether the cross-lagged, bi-directional relationships between parenting practices and children’s externalizing
behavior would be stronger for children with a higher score on a dopaminergic polygenic susceptibility index. Here, we hypothesized, in line with the differential susceptibility hypothesis, that for children with high genetic susceptibility, the cross-lagged
associations between parenting behavior and children’s externalizing behavior would be significantly stronger than for children with low genetic susceptibility.
Method Participants and Procedure
The present study made use of three-wave longitudinal data from the ORCHIDS study (Chhangur et al., 2012), in which 378 parents (Mage = 38.09, SD = 4.84; 91 % mothers) participated with their 4-8 year old children (Mage = 6.21, SD = 1.33; 55.30 % boys). Participating children were screened for being at risk of externalizing behavior. At-risk status was determined with a cut-off score at the 75th percentile on the Eyberg Child Behavior Inventory (ECBI, Eyberg & Pincus, 1999), which parents filled out prior to study. Half of the parents were enrolled in an intervention condition (n= 197) and the other half in a control condition (n=190). Over 84 % percent of the parents were born in the Netherlands. 50.5 % percent of the mothers had high education, while 27.5 % and 21.2 % percent had medium and low education, respectively (for fathers: 45.6 % high, 26.2 % medium, 25.6 % low education). 87 % percent of parents were either married or living together, 8.8 % were single, and 4.1 % classified as other. 26.4 % of parents were not working, from which 44.4 % were a stay-at home parent, 23.3 % were unemployed, and 32.3 % classified as other. 39.15 % percent of the families were Christian, 5.45 % Islamic, 21.2 % other religion, and 34.2 % were non-religious. In the ORCHIDS study, children’s externalizing behavior and parenting practices were assessed at three time points (T1, T2, T3), with measurement intervals of 6 and 4 months each. There was a 93 % retention rate at T3.
Measures
Questionnaire Measures
Parenting Practices (PP) Both negative and positive parenting practices were assessed with the Parenting Practices Inventory (PPI; Webster-Stratton, 2001). The PPI consists of 64 items ordered in 7 subscales, but in the present study we only used the
subscales of ‘Harsh and Inconsequent Discipline’ (16 items) and ‘Physical Punishment’ (6 items) to construct a measure of negative parenting practices, and the subscales ‘Positive Verbal Discipline’ (9 items) and ‘Praise and Incentives’ (11 items) to construct a measure of positive parenting practices. The other subscales of the PPI did not measure exlusively positive or negative parenting practices and were therefore not included as measures. All items were scored on a 7-point Likert scale, which assessed parents‘ perception of how likely they used a certain parenting practices, ranging from 1= very unlikely to 7= very likely. Cronbach’s alpha for the negative parenting practices measure was 0.79, 0.83, and 0.80 at T1, T2, and T3, respectively. Cronbach’s alpha for the positive parenting practices measure was 0.74, 0.77, and 0.80 at T1, T2, and T3.
Children’s externalizing behavior (CEB) Children’s externalizing behavior was assessed by parent reports on the Eyberg Child Behavior Inventory (ECBI, Eyberg & Pincus, 1999), a measure designed to assess conduct problems in children aged 7 to 16 years. The present study utilized the ‘Intensity Scale’ of the ECBI, which consists of 36 items and demonstrated a Cronbach’s alpha of 0.85, 0.86, and 0.88 at T1, T2, and T3, respectively. All items are scored on a 7-point Likert scale, which assesses how often a certain child behavior occurs, ranging from 1= never to 7=always.
Genetic susceptibility Measure
Genetic susceptibility was determined at wave 1 of the study with saliva samples using Oragene DNA collection kits (DNA Genotek; Ontario, Canada). The dopamine polymorphisms of interest for the present study were DRD4 (7R allele), DRD2 (A1 allele), DAT1 (10R allele), MAOA (low-activity allele), COMT (val allele), as these
polymorphisms have been related to children’s differential susceptibility to both positive and negative environmental influences (Belsky & Pluess, 2009; 2013).
DRD2 and COMT The SNP’s of the DRD2 rs1800497 and COMT rs4680 polymorphisms were analyzed with TaqMan® chemistry (Cat. # 4351379, Applied Biosystems). Analysis was done with 1ul of saliva sample in the 7500 System SDS software on the ABI-7500 Real-Time PCR instrument. 1 microliter of PCR product was combined with 0.3 µl LIZ-500 size standard (Applied Biosystems) and 11,7 µl
formamide (Applied Biosystems) and fragment analyzed with the AB 3730 instrument (50 cm capillaries). GeneMarker software was used for analysis, with DRD2 genotypes (n = 247 A2/A2, n = 122 A2/A1, n = 14 A1/A1) in Hardy-Weinberg equilibrium (HWE)
χ2 (1, n = 383) = .05, p = .82 (n = 4 no genotyping). Similarly for the COMT genotypes,
following details applied: n=90met/met, n=185met/val,n=108val/val)werein HWE,χ2 (1,N=383)=.04,p=.84 (n = 4 no genotyping).
DRD4. Primers 5’- GCGACTACGTGGTCTACTCG -3’ (FAM-labelled) and 5’- AGGACCCTCATGGCCTTG -3’ (reverse primer) were used to amplify the relevant area of the DRD4. Between 10 to 100 ng genomic DNA templates were yield in the standard PCR reactions, with forward and reverse primers of 10pmol implemented. PCR was conducted with 7.5% DMSO with 5x buffer (supply: enzyme and 1.25U of LongAmp Taq DNA Polymerase (NEB), 30 µl volume) and cycling conditions in respective order:
1) initial denaturation (10 min at 95oC), 2) 27 cycles of 30 sec 95oC, 30 sec 60oC, 60 sec
no7R/7R, n = 8 7R/7R) were set in HWE, χ2 (1, N = 375) = 2.11, p = .15 (n = 12 no genotyping).
DAT 1. Primers 5’- TGTGGTGTAGGGAACGGCCTGAG -3’(FAM-labelled) and 5’- CTTCCTGGAGGTCACGGCTCAAGG -3’ (reverse primer) were used to amplify the relevant area of the DAT 1. Between 10 to 100 ng genomic DNA templates were yield in the standard PCR reactions, with forward and reverse primers of 10pmol implemented. PCR was conducted with 3.3% DMSO with 1.25U of LongAmp Taq DNA Polymerase (NEB) (30 ìl volume) with cycling conditions of 1) denaturation (5 min at
95oC), 2) 29 cycles of 30 sec 95oC, 30 sec 68oC, 60 sec 65oC, 3) and an extension of 5
min 65oC. Genotypes (n = 31 no10R/no10R, n = 148no10R/10R,n=20310R/10R) were
set in HWE,χ2 (1,N=382)=.03,p=.86(n=5no genotyping).
MAOA. Primers 5’-GGATAACAATTTCACACAGG-3’ (FAM-labelled) and 5’-ggataacaatttcacacaggACAGCCTGACCGTGGAGAAG-3’ (reverse primer) were used to amplify the relevant area of the MAOA. Between 10 to 100 ng genomic DNA
templates were yield in the standard PCR reactions, with forward primer (1 pmol), reverse and MR primers (10 pmol). PCR was conducted with 5 % DMSO with 1.25U of LongAmp Taq DNA Polymerase (NEB) (30 ìl volume) with following cycling conditions
in respective order: 1) initial denaturation (5 min at 94oC), 2) 38 cycles of 30 sec 94oC,
30 sec 55oC, 30 sec 72oC, and 3) final extension (4 min 72oC). Genotypes (n = 97
low/low, n = 77 low/high, n = 196 high/high) were set in HWE, χ2 (1, N = 370) = 112.61, p = .001 (n = 17 no genotyping).
Polygenic Index Score Children were classified as either ‘low’ susceptible or ‘high’ susceptible to externalizing behavior depending on their polygenic score. The polygenic score represents a sumscore that was created by assigning a point for each polymorphism that disposed over at least one of the susceptibility alleles. If the sum-score ranged from 0-2, children were classified as ‘low’ on the polygenic index, signifiying low genetic susceptibility. A score of 3-5 classified as ‘high’ genetic susceptibility, signifying high genetic susceptibility (Belsky & Beaver, 2011).
Statistical Procedures
In the present cross-lagged panel-study, the longitudinal bi-directional
associations between both negative and positive parenting and children’s externalizing behavior were investigated (cross-lagged paths), and their stability over time (stability paths). The present study used structural equation modeling to investigate the cross-lagged assocations between parenting practices and children’s externalizing behavior. Two structural equation models were built, one for an examination of the bidirectional prospective links between positive parenting practices and children’s externalizing behavior (analysis 1), and one to investigate the bidirectional prospective links between negative parenting practices and children’s externalizing behavior (analysis 2).
Measurement Model
In a first step, a measurement model was created for each analysis separately to represent the relationships of the latent variables to their indicators. Figure 1a. (see Appendix) represents the measurement model for negative parenting and Figure 1b. (see Appendix) represents the measurement model for positive parenting across the three measurement waves. Each latent variable disposes over three indicators. For the latent
variable of children’s exernalizing behavior (CEB), three indicator variables were formed by parceling the ECBI scale into three parcels. This was done by means of
unidimensional parceling, where items are randomly assigned into one of the three sets of parcels (Kishton & Widaman, 1994). For the latent variable of negative parenting
practices (analysis 1), three indicator variables were formed. These were created by parceling the HID subscale into two parcels and by implementing the full PP subscale as one indicator. The reliabilities of the first HID parcel (six items) at T1, T2, and T3 were 0.61, 0.67 and 0.62. The second parcel (nine items) revealed reliabilities of 0.72, 0.78, and 0.76 at T1, T2, and T3. Correlations between the two parcels were 0.44, 0.50, and 0.42 at T1, T2, and T3, respectively. These correlations are significant at the alpha level 0.01. For the latent variable of positive parenting practices (analysis 2), three indicator
variables were formed, by parceling the PVD subscale into two parcels and implementing the full PI subscale as a third indicator. The reliabilities for the first PVD parcel (five items) were 0.64, 0.64, 0.68 and for the second parcel (four items) were 0.77, 0.80, 0.83 at T1, T2, T3, respectively. Correlations between the two parcels were 0.26, 0.11, and 0.27 at T1, T2, and T3, respectively. These correlations were significant at the alpha level 0.01 (T1, and T3) and 0.05 level (T2). Parcels were formed according to domain representative parceling (Kishton & Widaman, 1994): the HID subscale was split into two parcels of “harsh discipline” and “inconsequent discipline” and the PVD subscale into two parcels of “praise” and “correcting and discussing behavior”. The robust correlations between the parcels for each negative parenting practices, positive parenting practices, and children’s externalizing behavior signify that a common psychological variable underlies the separate created dimensions and demonstrate that parceling was successful (Kishton &
Widaman, 1994).
Structural Model A structural model was created to model the hypothesized cross-lagged paths between parenting practices and children’s externalizing behavior from one time point to the next, over three waves of longitudinal data. Furthermore, the structural model specified the stability paths between the time points of parenting
practices and children’s externalizing behavior each, to determine the constructs’ stability over time. The structural model for analysis 1 represents the cross-lagged and stability paths between negative parenting practices and child externalizing behavior, and for analysis 2 between positive parenting practices and child externalizing behavior. The structural models are depicted in Figure 2a. (negative parenting) and Figure 2b. (positive parenting). Additionally, factors for ‘age’ and ‘gender’ were included in the structural model as covariates. Measurement errors were correlated over time across the 3 time points for the same indicator, in line with the assumption that there may be covariation among measurement errors of repeated measures (Pitts et al., 1996). Missing data was handled by means of the full-information maximum likelihood (FIML) method. The advantage of FIML over traditional techniques, such as listwise- or pairwise deletion and mean-imputation methods, is that it there is less bias, greater efficiency, and no decrease in sample size (Little, Schnabel, & Baumert, 2000).
A multigroup analysis was conducted, in order to investigate whether there were differences in cross-lagged and stability effects of parenting practices and children’s externalizing behavior across the two groups of genetic susceptibility, ‘high genetic susceptibility’ and ‘low genetic susceptibility’. First, both groups were fitted
parameters in each group, to establish the ‘baseline model’. In order to test whether there are differences in 1) the longitudinal bi-directional associations between parenting practices and children’s externalizing behavior and 2) the stability of parenting practices and children’s externalizing behavior across groups, the stability and cross-lagged effects were constrained to be equal across groups, representing the ‘constrained model’. The constrained model was then compared to the baseline model with a chi-square difference test, in order to test whether there are significant differences in the longitudinal
bi-directional associations between parenting practices and children’s externalizing behavior across the two genetic susceptibility groups.
Results Descriptive Statistics
Table 2 (see Appendix) displays the mean and standard deviations for negative and positive parenting practices, and children’s externalizing behavior. Mean scores for negative parenting practices were marginally non-clinical and in the clinical range for positive parenting practices (Webster-Stratton, 2001). These scores reflect the nature of our at-risk sample, with moderately low scores on positive parenting and moderately high scores on negative parenting. The elevated scores on children’s externalizing behavior also confirm the at-risk characteristic for children’s externalizing behavior in the present sample (Eyberg & Pincus, 1999). Table 2 furthermore displays the correlations between positive, negative parenting and children’s externalizing behavior. Nearly all correlations were significant. Table 2 clearly shows that overall, there was a significant
cross-sectional and longitudinal bivariate association between negative and positive parenting practices and children’s externalizing behavior.
Measurement model-data fit assessment
The means and standard deviations of the indicators are displayed in Table 1 (see Appendix). Measurement models for both negative and positive parenting practices showed adequate model-data fit (negative parenting: CFI= 0.97, TLI= 0.96, RMSEA= 0.05; positive parenting: CFI= 0.98, TLI= 0.97, RMSEA= 0.04).
Structural model-data fit assessment
The fit of the structural models for negative and positive parenting practices were assessed. The present study comprises a sample size of nearly 400, whereby a distortion to a significant chi-square statistic is expected (Hayduk, et al., 2007; Burkholder & Harlow-Cross, 2003). In cases like these, a state-of-the-art approach is to use several fit indices as fit criteria (Hooper et al., 2012). Therefore, the following model-fit indices were assessed: Comparative Fit Index (CFI: Bentler, 1990), the Tucker Lewis Index (TLI: Bentler, 1990), and Root Mean Squared Error of Approximation (RMSEA: Steiger & Lind, 1980). Values of 0.95 and over for the CFI and TLI, and values under 0.07 for the RMSEA signify good model-data fit. For negative parenting practices, the structural model revealed good fit to the data, with a CFI of 0.96, TLI of 0.95, and a value of 0.05 on the RMSEA. For positive parenting practices, results also revealed good model-data fit (CFI= 0.97, TLI= 0.96, RMSEA= 0.04). These fit indices demonstrate adequate development of the structural model for the models of negative and positive parenting.
Cross-lagged effects between parenting and children’s externalizing behavior Figure 2a. and 2b. (see Appendix) depict the structural equation models for negative and positive parenting practices, respectively for the total sample. The
cross-lagged paths across the 3 time points represent the longitudinal bi-directional
relationships between parenting practices and children’s externalizing behavior with standardized regression coeffients. For both negative and positive parenting, all coefficients of cross-lagged regression paths were non-significant (p’s > 0.05 level). However, all stability paths for both negative and positive parenting were significant (p’s < 0.0005), showing that both parents’ perception of their own parenting practices and of their child’s externalizing behavior were highly stable over time. Interestingly, these stability effects even increased slightly over time, which means that parent’s perception of their own parenting practices and of their child’s externalizing behavior became even more stable over time. The respective results of the cross-lagged and stability paths mean that, when controlling for these high levels of stability, neither negative nor positive parenting practices predicted subsequent children’s externalizing behavior, and vice versa that children’s externalizing behavior does not subsequently evoke more negative or less positive parenting practices.
Age and gender as covariates Age and gender were implemented as covariates in the structural models of negative and positive parenting practices. Overall, age and gender were not associated with parenting practices and children’s externalizing behavior, except for one instance where positive parenting practices at T1 were significantly negatively associated with age.
Bi-directional associations by genetic group
For the model of negative parenting practices, the comparison test showed no significant differences between the constrained and baseline model (X2diff= 1.022, df= 8, p= 0.998). For positive parenting practices, there were also no differences between the
constrained and baseline model (X2diff= 6.643, df= 8, p= 0.576). This means that there were no differences in the longitudinal bi-directional associations between parenting practices and children’s externalizing behavior, nor the stability paths between the ‘high’ and ‘low’ genetic susceptible group. Even when controlling for differences in stability paths, no differences between the cross-lagged paths emerged (negative parenting practices model: X2diff= 0.759, df= 4, p= 0.943; positive parenting practices model: X2diff = 4.085, df= 4, p= 0.395). In other words, children’s dopamine-based polygenetic score did not affect the longitudinal relationships between negative nor positive parenting practices and children’s externalizing behavior. Specifically, Table 4 (see Appendix) shows the standardized parameter estimates of the cross-lagged and stability paths for both genetic subgroups. None of the cross-lagged paths were significant, neither in the ‘high’ nor the ‘low’ genetic susceptibility group, but in both groups all stability paths were significant and strong. Table 3 (see Appendix) shows the mean scores for indicators of negative parenting practices, positive parenting practices and children’s externalizing behavior by genetic group.
Discussion
Contrary to prior research in the field, the present study revealed that there were no longitudinal associations between parenting practices and childrens externalizing behavior over time. Furthermore, there were no differences in these relationships between the low and high genetic susceptibility group, which also stands in contrast to prior research. Yet, the results did show that parenting behavior and children’s externalizing behavior were stable over time and that there were cross-sectional associations between parenting practices and children’s externalizing behavior.
Other studies, however, also did not find cross-lagged associations from children’s externalizing behavior to negative parenting reactions, thus not supporting a transactional relationship between negative parental reactions and child externalizing behavior (Calkins et al. 2002;Mackler et al., 2015). Interestingly, in both of these studies significant bi-directional associations emerged between parental stress and children’s externalizing behavior. The authors propose that once parental stress exacerbateas, negative parenting behaviors may emerge at later time points as a secondary product of parental stress (Mackler et al., 2015). An explanation for our null-findings on the cross-lagged associations between negative parenting and children’s externalizing behavior may then be that these effects go undetected if the time span of the longitudinal study is too short. Indeed, the total time span of the present study comprised 10 months, which is considerably shorter than many longitudinal studies in the field that often span over several years (e.g. Mackler et al., 2015; Eisenberg et al., 2005; Mesman et al. 2001). It should be noted that in the present study, levels of negative parenting were already elevated (at-risk). We propose that an addition of later measurement intervals may have allowed detection of negative parenting levels in the clinical range. Clinical levels of dysfunctional parenting may in turn elicit dysfunctional parent-child exchanges more readily or to a stronger extent than at-risk parenting practices.
Similarly, regarding positive parenting practices, other large-scale longitudinal studies (e.g. Rueter & Conger, 1998; n= 5100) support the lack of bi-directional
associations between positive parenting practices and children’s externalizing behavior. These findings corroborate research that shows that positive child behaviors result from -rather than elicit- positive parenting practices (Conger et al., 1986). Roeter and Conger
(1998) explain that positive behavioral reciprocalities may require longer time periods than negative reciprocalities to form. Thereby, we may expect that positive parent-child transactional associations emerge even later than those between negative parenting practices and children’s externalizing behavior. Crucially then, it may be that especially
positive parenting and children’s externalizing behavior longitudinal associations go
undetected if the longitudinal study time span is too short.
Another explanation for why we did not find cross-lagged effects between parenting and child behavior, may be that bi-directional relations between parenting and children’s externalizing behavior do not emerge due to the type of implemented measures used to assess parenting behavior: The present study used the Parenting Practices
Inventory (Webster-Stratton, 2001), which assesses parent behaviors to hypothetical child disruptive scenarios. Indeed, other studies, in which like measures were implemented, postulated that hypothetical scenarios do not represent actual observed parenting
behaviors and may thereby not yield effects (Mackler et al., 2015). As such, parents may underrate certain parenting behaviors because they cannot relate to these scenarios occuring in their own family context. Furthermore, hypothetical scenarios may lower the threshold for occurrence of social desireability bias, as parents may experience less cognitive dissonance when underreporting hypothetical negative- or low positive parenting behaviors as opposed to behaviors that actually occurred.
The results revealed that children’s externalizing behavior and both negative and positive parenting remained significantly stable at marginally clinical levels over the 3 measurement waves. Interestingly, stability paths gained in strength over time, which suggests that the stability of children’s externalizing behavior and parenting practices
increased over time. In support of the former, externalizing problems have consistently emerged as highly stable longitudinal constructs, even in the first 4 years of life (e.g. Campbell & Ewing, 1990; Lavigne et. al., 1998) and shown to become worse over time (e.g. Mesman et al., 2001, Vaughn et al., 2013). In support of stability of parenting practices over time, prior research shows that both negative and lower positive parenting practices remain significantly stable and grow stronger over time (e.g. Miner & Clarke-Stewart, 2008; Mackler et al., 2015). This effect may be explained by parents decreasing tolerance to children’s externalizing behavior over time, which perpetuates an increasing use of negative parenting reactions and less positive parenting practices (e.g. Mackler et al., 2015). Large-scale longitudinal studies (e.g., Rueter & Conger, 1998) corroborate these contentions, by demonstrating the robust stability of levels of negative and positive parental practices over a span of two years in a sample of roughly 5100 parent-child dyads. The stability of both negative and positive parenting practices over time suggest that parents’ perception of their own parenting practices are robust to change, whether for better or for worse.
Opposite to our expectations, the current study also did not show any longitudinal bidirectional associations and stability paths between the high and low genetic
susceptibility group. This is in contrast to research reviewed earlier that supports that children disposing over a certain dopaminergic susceptibility are affected to a greater extent by both negative and positive environmental influences (e.g.
Bakermans-Kranenburg & Van IJzendoorn, 2011; Schultz et al., 2010, Davies et al., 2015; Posner & Rothbart, 2009; Belsky & Pluess, 2009; 2013). Our findings do not support this
genetic moderation results may be that longitudinal studies are in general less effective than randomized controlled trials in controlling and assessing the environment
(Bakermans-Kranenburg & Van IJzendoorn, 2015). This contributes to lower power in detecting a genetic moderation effect, because there is greater preciseness in measuring the genetic component (low measurement error) than the environment (high measurement error). Indeed, the resulting unequal measurement error has been postulated as one of the main contributors to Plomin’s paradox, which says that even though there is an
omnipresence of gene-environment interactions, these effects are hard to detect and are seldomly replicated (Wachs & Plomin, 1991). Crucially, it has been postulated that proper measurement of the environment, as with randomized experimental trials, is an even more critical criteria than adequate sample size in finding gXe effects (Rutter, 2006; Wong et al. 2003). Thus, there is an increasing call to move away from longitudinal and correlation research to randomized controlled experiments to test the differential
susceptibility hypothesis (Van IJzendoorn et al. 2011). Furthermore, studies
implementing observational or interview measures have demonstrated greater genetic moderation effects than studies using self-reports (Karg et al., 2011). Considering that the present study was both a longitudinal study with self-reports, it may be suggested that these methodological characteristics may have interfered with the detection of a gXe effect.
A surprising finding was the overall non-significant effects of age and gender on children’s externalizing behavior and on both negative and low positive parenting
practices, as it is commonly found that externalizing behavior becomes more severe with increasing age (e.g. Campbell et al. 2000, Mesman et al., 2001, Vaughn et al., 2013). It
was furthermore surprising that there was no effect of gender on children’s externalizing behavior or parenting behavior, as it is consistently found that boys exhibit more
externalizing behaviors than girls from age 4 on (Menting et al., 2013). Notably, these gender differences are commonly observed in samples of the general population
(Rosenfield, Phillips, & White, 2006), whereas these gender differences tend to vanish in at-risk or clinical populations (e.g. Marston, Russell, Obsuth, & Watson, 2012; Timmons-Mitchell et al., 1997). This may explain why no gender differences were found in our at-risk sample.
Certain limitations of the current study need to be addressed. Firstly, with 91 % of the parents in this study being mothers, the results are not representative of transactions between paternal parenting and child behavior. Secondly, in light of self-reports as our sole measure in the study, it needs to be taken into account that self-report measures may introduce potential biases such as social desireability, which may skew both the parent and child ratings. Additionally, the study would profit from additional rater reports (e.g. teacher reports) to capture triggers of dysfunctional behavior outside of the home, such as social problems at school, which have crucially shown to be key players in perpetuating externalizing problems (e.g. Cole, Martin, Powers, & Truglio, 1996). Furthermore, the use of multiple, independent informants is highly recommended to avoid method variance effects that produce biased parameter estimates (e.g. Bank, Dishion, Skinner, &
Patterson, 1990). Lastly, even though creating a polygenic index allows the assessment of several susceptibility allelic variants at the same time and is more effective than single candidate-gene testing, it carries the disadvantage of representing a categorical sum score (Belsky & Beaver, 2011). New software programs (e.g. ‘PLINK’, Purcell et al., 2007) are
available which allow to overcome this problem, by the calculation of a continuous polygenic score of several hundred genotyped markers for each individual (Purcell et al., 2007).
Strengths of the study were that we worked with an at-risk sample with a higher proportion of dysfunctional parent and child behavior. This boosts power and produces less skewed variance (Van Ijzendoorn & Bakermans-Kranenburg, 2015). What makes the study different from most studies on parenting practices and children’s externalizing behavior, is the focus on both positive and negative parenting practices. In this manner, the present study conducted a stringent test of longitudinal-bidirectional relationships for both positive and negative dimensions of parenting and was able to test the differential susceptibility hypothesis. Other, methodological strengths of the study include a relatively large sample size (378 parent-child dyads), with a low overall attrition rate (retention rate of 93 %). Lastly, even though assessing parent-child behaviors with the self-report method can be criticized from one perspective, self-reports are necessary in order to assess parent’s perceptions of their child’s behavior, which have crucially shown to predict child problem behavior (e.g., McMahon & Meins, 2012). Furthermore, they are indispensible when assessing behaviors, which do not occur -or cannot be elicited- in experimental paradigms (Mackler et al., 2015).
The present study has several important clinical implications. The study showed that children’s externalizing behavior remains highly stable and that this robustness even increases slightly over time. This finding illustrates that interventions for children are crucial in order to prevent externalizing behavior from exacerbating. Interventions should be appointed as early as possible to prevent aversive outcomes and self-reinforcing
behavior problem cycles. Similarly, both positive and negative parenting behavior showed to be stable, and even increasing stability, over time. This signifies that parenting behaviors, whether low in the positive or high in the negative parenting dimension, are robust to change such as through intervention. This supports why parenting interventions that are administered continuously over longer time frames are so successful in breaking down aversive- and building up functional parenting practices. Since both negative and positive parenting occurred at clinical levels (high negative and low positive parenting), this furthermore supports the effectiveness of interventions that work not only on reducing negative parenting, but crucially also boosting positive parenting. One of the most important insights in the present study is that children’s
externalizing behavior and parenting practices can co-exist, remaining stable over time, without showing significant reciprocal associations over time. These findings are in contrast to prior research which supports a bi-directional perspective on dysfunctional parent-child interactions. In light of the methodological limitations of this study, these finding should be interpreted with caution. More research is necessary to investigate the reinforcing nature of child-and-parent behaviors over time against the backdrop of genetic susceptibility in order to optimize existing parenting-interventions.
References
Bank, L., Dishion, T , Skinner, M., & Patterson, G. R. (1990). Method variance in structural equation modeling: Living with ' 'glop.'' In G. R. Patterson (Ed.), Depression and aggression infamily interactions (pp. 247-279). Hillsdale, NJ: Erlbaum.
Bakermans-Kranenburg, M. J., & Van IJzendoorn, M. H. (2011). Differential
susceptibility to rearing environment depending on dopamine-related genes: New evidence and a meta-analysis. Development and Psychopathology, 23, 39–52.
Bakermans-Kranenburg, M. J., & Van IJzendoorn, M. H. (2015). The hidden efficacy of interventions: Gene × environment experiments from a differential susceptibility perspective. Annual Review of Psychology, 66, 381–409.
Belsky, J., & Pluess, M. (2009). Beyond diathesis stress: Differential susceptibility to environmental influences. Psychological Bulletin, 135, 885–908.
Belsky, J., & Beaver, K. M. (2011). Cumulative-genetic plasticity, parenting and
adolescent self-regulation. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 52, 619–26.
Belsky, J., & Pluess, M. (2013). Beyond risk, resilience, and dysregulation: Phenotypic plasticity and human development. Development and Psychopathology, 25, 1243– 1261.
Bentler, P.M. (1990), "Comparative Fit Indexes in Structural Models," Psychological Bulletin, 107, 238-46.
Bor, W., Sanders, M. R., & Markie-Dadds, C. (2002). The effects of the triple P-positive parenting program on preschool children with co-occurring disruptive behavior and attentional/hyperactive difficulties. Journal of Abnormal Child Psychology, 30, 571– 587.
Boyce, W. T., & Ellis, B. J. (2005). Biological sensitivity to context: I. An evolutionary- developmental theory of the origins and functions of stress reactivity. Development
and Psychopathology, 17, 271–301.
Braungart-Rieker, J., & Garwood, M. M., Stifter, C. A. (1997). Compliance and
noncompliance: The roles of maternal control and child temperament. Journal of Applied Developmental Psychology, 18, 411– 428.
Burkholder, G. J., & Harlow, L. L. (2003). An Illustration of a Longitudinal Cross- Lagged Design for Larger Structural Equation Models, Structural Equation Modeling, 10, 465–486.
Byrd, A. L., & Manuck, S. B. (2014). MAOA, childhood maltreatment, and antisocial Behavior: Meta-analysis of a gene-environment interaction. Biological Psychiatry, 75, 9–17.
Calkins, S. D. (2002). Does aversive behavior during toddlerhood matter? The effects of difficult temperament on maternal perceptions and behavior. Infant Mental Health
Journal, 23, 381– 402.
Campbell, S. B., & Ewing, L. J. (1990). Follow-up of hard-to- manage preschoolers : Adjustment at age 9 and predictors of continuing symptoms. Journal of Child
Psychology and Psychiatry, 31, 871–889.
Campbell, S. B., Spieker, S., Burchinal, M., Poe, M. D., Belsky, J., Booth-LaForce, C.,
Weinraub, M. (2006). Trajectories of aggression from toddlerhood to age 9 predict academic and social functioning through age 12. Journal of Child Psychology and
Psychiatry and Allied Disciplines, 47, 791–800.
Campbell, S. B., Shaw, D. S., & Gilliom, M. (2000). Early externalizing behavior problems: Toddlers and preschoolers at risk for later maladjustment. Development and Psychopathology, 12, 467–488.
Chhangur, R. R., Weeland, J., Overbeek, G., Matthys, W., & Orobio de Castro, B. (2012). ORCHIDS: An observational randomized controlled trial on childhood differential susceptibility. BMC Public Health, 12, 917–923.
Cole, D. A., Martin, J. M., Powers, B., & Truglio, R. (1996). Modeling causal relations between academic and social competence and depression: A multitrait-
multimethod longitudinal study of children. Journal of Abnormal Psychology, 105, 258–270.
Conger, R. D., Brainerd, D. W., Birch, L. L., Friedberg, P. J., & Navarro, L. A. (1986). Assessing the quality of family observations: A comparative analysis. Journal of Marriage and the Family, 48, 361-373.
Davies, P., Cicchetti, D., & Hentges, R. F. (2015). Maternal unresponsiveness and child disruptive problems: the interplay of uninhibited temperament and dopamine
Dick, D. M., Agrawal, A., Keller, M. C., Adkins, A., Aliev, F., Monroe, S., & Sher, K. J. (2015). Candidate gene-environment interaction research: Reflections and
recommendations. Perspectives on Psychological Science, 10, 37–59.
Dick, D. M., Latendresse, S. J., & Riley, B. (2011). Incorporating genetics into your studies: a guide for social scientists. Frontiers in Psychiatry, 2, 1-10.
Eisenberg, N., Zhou, Q., Spinrad, T. L., Valiente, C., Fabes, R. A., & Liew, J. (2005). Relations Among Positive Parenting, Children’s Effortful Control, and Externalizing Problems: A Three-Wave Longitudinal Study. Child Development, 76, 1055-1071.
Eyberg, S. M., & Pincus, D. (1999). Eyberg child behavior inventory and sutter-eyberg
student behavior inventory-revised: Professional manual. Odessa, FL: Psychological
Assessment Resources.
Hartup, W. W., van Lieshout, C. F. M. (1995). Personality development in social contex. Annual Review of Psychology, 45, 655–687.
Hayduk, L., Cummings, G. G., Boadu, K., Pazderka-Robinson, H., and Boulianne, S. (2007). "Testing! Testing! One, Two Three – Testing the theory in structural equation models!" Personality and Individual Differences, 42, 841-50.
Hooper, D., Coughlan, J., Mullen, M. (2008). Structural Equation Modelling: Guidelines for Determining Model Fit. Electronic Journal of Business Research Methods, 6, 53- 60.
aggressive and rule-breaking behaviors and parental support and dysfunctional discipline. Development and Psychopathology, 24, 301-315.
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, 444-54.
Karreman, A., Van Tuijl, C., Van Aken, M. A., & Dekovic, M. (2006). Parenting and self-regulation in preschoolers: A meta-analysis. Infant and Child Development, 15, 561–579.
Keijsers, L., Loeber, R., Branje, S., & Meeus, W. (2011). Bidirectional links and concurrent development of parent-child relationships and boys’ offending behavior. Journal of Abnormal Psychology, 120, 878-889.
Kern, M. L., & Friedman, H. S. (2008). Do conscientious individuals live longer? A quantitative review. Health Psychology, 27, 505–512.
Kim-Cohen, J., Caspi, A., Moffitt, T. E., Harrington, H., Milne, B. J., Poulton, R. (2003).
Prior juvenile diagnoses in adults with mental disorder: developmental follow-back of a prospective-longitudinal cohort. Archives of General Psychiatry, 60, 709-17.
Kishton, J. M., Widaman, K. F. (1994). Unidimensional versus domain representative parceling of questionnaire items: an empirical example. Educational and
Psychological Measurement, 54, 757-765.
Evocative gene–environment correlation in the mother–child relationship: A twin study of interpersonal processes, Development and Psychopathology, 25, 105–118.
Lavigne, J. V., Arend, R., Rosenbaum, D., Binns, H. J., Kaufer-Cristoel, K., & Gibbons, R. D. (1998). Psychiatric disorders with onset in the preschool years: I. Stability of diagnoses. Journal of the American Academy of Child and Adolescent Psychiatry, 37, 1246–1254.
Little, T. D., Schnabel, K. U., & Baumert, J. (2000). Longitudinal and multi-group modeling with missing data. Chicago: Smallwaters Corporation.
McMahon, C. A., Meins, E. (2012). Mind-mindedness, parenting stress, and emotional availability in mothers of preschoolers. Early Childhood Research Quarterly, 27, 245–252.
Mackler, J. S., Kelleher, R., Shanahan, L., Calkins, S. D., Keane, S. P., O’Brien, M. (2015). Parenting Stress, Parental Reactions, and Externalizing Behavior From Ages 4 to 10. Journal of Marriage and Family, 77, 388–406.
Marston, E. G., Russell, M. A., Obsuth, I., & Watson, G. K. (2012). Dealing with double jeopardy: Mental health disorders among girls in the juvenile justice system. In S. Miller, L.
Matthys, W., Vanderschuren, L. J. M. J., & Schutter, D. J. L. G. (2013). The neurobiology of oppositional defiant disorder and conduct disorder: Altered
functioning in three mental domains. Development and Psychopathology, 25, 193– 207.
Matthys, W., Vanderschuren, L. J. M. J., Schutter, D. J. L. G., & Lochman, J. E. (2012). Impaired Neurocognitive Functions Affect Social Learning Processes in
Oppositional Defiant Disorder and Conduct Disorder: Implications for
Interventions. Clinical Child and Family Psychology Review, 15(3), 234–246.
Menting, A. T. A., Orobio de Castro, B., & Matthys, W. (2013). Effectiveness of The Incredible Years parent training to modify disruptive and prosocial child behavior: A meta-analytic review. Clinical Psychology Review, 33, 901–913.
Mesman, J., Bongers, I. L., & Koot, H. M. (2001). Preschool Developmental Pathways to Preadolescent Internalizing and Externalizing Problems. Journal of Child
Psychology and Psychiatry, 42, 679–689.
Miner, J. L., & Clarke-Stewart, K. A. (2008). Trajectories of externalizing behavior from age 2 to age 9: relations with gender, temperament, ethnicity, parenting, and rater. Developmental Psychology, 44, 771–786.
Moffitt, T. E., Arseneault, L., Belsky, D., Dickson, N., Hancox, R. J., Harrington, HL., Houts, R., Poulton, R., Roberts, B. W., Ross, S., Sears, M. R., Thomson, W. M., Caspi, A., & Heckman, J. J. (2011). A gradient of childhood self-control predicts health, wealth, and public safety. PNAS, 108, 2693-2698.
Odgers, C. L., Moffitt, T. E., Broadbent, J. M, Dickson, N., Hancox, R. J., Harrington, H., & Patterson, G. R. (2002). The early development of coercive family process. In: Reid, JB.;
Patterson, G. R., & Snyder, J. (2002). Antisocial behavior in children and adolescents: A developmental analysis and model for intervention. Washington, DC: American Psychological Association; 25-44.
Patterson, G. R., & Cobb, J. A. (1971). A dyadic analysis of “aggressive” behaviors. In: Hill, JP., editor. Minnesota symposia on child psychology. Minneapolis, MN: University of Minnesota Press, 72-129.
Patterson, G. R. (1986). Performance models for antisocial boys. American Psychologist, 41, 432-444.
Pitts, S. C., West, S. G., & Tein, J. (1996). Longitudinal measurement models in evaluation research: Examining stability and change. Evaluation and Program Planning, 19, 333–350.
Posner, M. I., & Rothbart, M. K. (2009). Toward a physical basis of attention and self- regulation. Physics of Life Reviews, 6, 103-120.
PLINK: A tool set for whole- genome association and population-based linkage analyses. The American Journal of Human Genetics, 81, 559-575.
Roche, K. M., Ghazarian, S. R., Little, T. D., & Leventhal, T. (2010). Understanding links between punitive parenting and adolescent adjustment: the relevance of context and reciprocal associations. Journal of Research on Adolescence, 21, 448- 460.
Rosseel, Y. (2012). Lavaan: An R Package for Structural Equation Modeling. Journal of
Rosenfield, S., Phillips, J., & White, H. (2006). Gender, race, and the self in mental health and crime. Social Problems, 53, 161–185.
Rothbaum, F., & Weisz, J. R. (1994). Parental caregiving and child externalizing behavior in nonclinical samples: A meta-analysis. Psychological Bulletin, 116, 55– 74.
Rueter, M. A., Conger, D. (1998). Reciprocal Influences Between Parenting and Adolescent Problem-Solving Behavior, Developmental Psychology, 34, 1470- 1482.
Rutter, M. (2006). Genes and Behavior. Nature-Nurture Interplay Explained. Oxford, UK: Blackwell .
Schultz, W. (2010). Dopamine signals for reward value and risk: Basic and recent data.
Behavioral and Brain Functions, 6, 24.
Shaw, D. S., Owens, E. B., Giovannelli, J., Winslow, E. B. (2001). Infant and toddler pathways leading to early externalizing disorders. Journal of the American Academy of Child & Adolescent Psychiatry. 40, 36–43.
Smith, J. D., Dishion, T. J., Shaw, D. S., Wilson, M. N., Winter, C., & Patterson, G. R. (2014). Coercive Family Process and Early-Onset Conduct Problems From Age 2 to School Entry. Developmental Psychopathology, 26, 917-932.
factors. Paper presented at the annual meeting of the Psychometric Society, Iowa City, IA.
Timmons-Mitchell, J., Brown, C., Schulz, S. C., Webster, S. E., Underwood, L. A., & Semple, W. E. (1997). Comparing the mental health needs of female and male incarcerated juvenile delinquents. Behavioral Sciences & the Law, 15, 195–202.
Van IJzendoorn, M. H., & Bakermans-Kranenburg, M. J. (2015). Genetic differential susceptibility on trial: Meta-analytic support from randomized controlled experiments. Development and Psychopathology, 27, 151–162.
Van Ijzendoorn, M. H., Bakermans-Kranenburg, M. J., Belsky J, Beach, S., Brody, G., Dodge, K. A., Greenberg, M., Posner, M., Scott, S. (2011). Gene-by- environment experiments: a new approach to finding the missing heritability. Nature Review Genetics, 12, 881-882.
Vaughn, M. G., Salas-wright, C. P., Delisi, M., & Maynard, B. R. (2013). Violence and externalizing behavior among youth in the United States: Is There a severe 5%? Youth Violence and Juvenile Justice, 12, 3–21.
Von Stumm, S., Deary, I. J., Kivimäki, M., Jokela, M., Clark, H., & Batty, G. D. (2011). Childhood behavior problems and health at midlife: 35-year follow-up of a Scottish birth cohort. Journal of Child Psychology and Psychiatry, 52, 992–1001.
Wachs, T. D., & Plomin, R. (1991). Conceptualization and Measurement of Organism- Environment Interaction. Washington, DC: American Psychological Association.
Webster-Stratton, C. (2001). Parenting practices interview. Unpublished Assessment
Instrument. Retrieved from
http://www.son.washington.edu/centers/parenting-clinic/forms.asp
Weeland, J., Overbeek, G., de Castro, B. O., & Matthys, W. (2015). Underlying mechanisms of gene–environment interactions in externalizing behavior: A
systematic review and search for theoretical mechanisms. Clinical Child and Family Psychology Review, 18, 413–442.
Wong, M. Y., Day, N. E., Luan, J. A., Chan, K. P., & Wareham, N. J. (2003). The detection of gene-environment interaction for continuous traits: Should we deal with measurement error by bigger studies or better measurement? Interntional Journal of Epidemiology, 32, 51–57.
Appendix
