Journal Article Summary

Geneticvulerabilitytoconflictandparent-adolescentrelationships.pdf

Home >Psychology homework help >Journal Article Summary

Developmental Psychology

Predictive Links Between Genetic Vulnerability to Depression and Trajectories of Warmth and Conflict in the Mother–Adolescent and Father–Adolescent Relationships Charlie Brouillard, Mara Brendgen, Frank Vitaro, Ginette Dionne, and Michel Boivin

Online First Publication, May 16, 2019. http://dx.doi.org/10.1037/dev0000751

CITATION

Brouillard, C., Brendgen, M., Vitaro, F., Dionne, G., & Boivin, M. (2019, May 16). Predictive Links Between Genetic Vulnerability to Depression and Trajectories of Warmth and Conflict in the Mother–Adolescent and Father–Adolescent Relationships. Developmental Psychology. Advance online publication. http://dx.doi.org/10.1037/dev0000751

Predictive Links Between Genetic Vulnerability to Depression and Trajectories of Warmth and Conflict in the Mother–Adolescent

and Father–Adolescent Relationships

Charlie Brouillard University of Quebec at Montreal

Mara Brendgen University of Quebec at Montreal and Sainte-Justine Hospital

Research Centre, Montreal, Canada

Frank Vitaro Sainte-Justine Hospital Research Centre, Montreal, Canada,

and University of Montreal

Ginette Dionne and Michel Boivin Laval University

The present study used a genetically informed design of twins raised in the same family (375 monozy- gotic and 290 dizygotic twins; 50.2% girls) to examine the association between adolescents’ genetic risk for depressive symptoms and the course of the parent– child relationship quality throughout adolescence. Depressive symptoms and the quality of the parent–adolescent relationships were measured through adolescents’ self-reports from ages 13 to 17. Group-based trajectory modeling revealed that most adolescents experienced high-quality relationships with both of their parents, characterized by high levels of warmth and low levels of conflict, and marked by gradual changes over adolescence. However, 3% of adolescents showed a trajectory of high and increasing conflict with their mothers and 16% of adolescents showed a trajectory of low warmth with their fathers, which decreased until mid-adolescence before increasing thereafter. Moreover, in line with an evocative gene– environment correlation process, a higher genetic vulnerability to depressive symptoms increased the likelihood of following a more problematic relationship trajectory with parents. This rGE was mediated by adolescents’ actual depres- sive behavior symptoms. Results also suggest that adolescents’ depression symptoms may affect girls’ and boys’ relationship with their parents in a similar way, with specific sex-patterns revolving more around the sex of the parent.

Keywords: depressive symptoms, genetic risk, mother–adolescent and father–adolescent relationship, trajectories, twins

During adolescence, the parent– child relationship is character- ized by a gradual decrease of support and warmth, as well as an increase of hostility and conflict (Furman & Buhrmester, 1992; Smetana, Campione-Barr, & Metzger, 2006). This shift is partly

explained by the individuation process that takes place during that developmental period, which allows the adolescent to be- come more autonomous from parents (Blos, 1967). The mostly authoritarian nature of the parent– child relationship during childhood is also redefined and renegotiated throughout ado- lescence, gradually transforming into a more mutual and recip- rocal one (Smollar & Youniss, 1989). Although these changes may entail occasional disagreements or conflicts, between 5% and 15% of parent–adolescent relationships undergo consider- able turmoil and emotional challenges during this period (Smetana et al., 2006). To better understand what characterizes these troublesome parent–adolescent relationships, the role of adolescent depression ought to be explored. Indeed, considering the interconnection between depressive symptoms and interper- sonal relationship quality (Coyne, 1976; Hammen, 2006), as well as the dramatic increase of depressive symptoms from childhood through adolescence with 20% of 7th graders show- ing depressive cognitions and behaviors (Saluja et al., 2004), adolescent depression may be at play in the multifaceted real- ities of those families. Using a genetically informed design of twins, we aimed to address this issue by examining the associ-

Charlie Brouillard, Department of Psychology, University of Quebec at Montreal; Mara Brendgen, Department of Psychology, University of Que- bec at Montreal, and Sainte-Justine Hospital Research Centre, Montreal, Canada; Frank Vitaro, Sainte-Justine Hospital Research Centre, and School of Psycho-Education, University of Montreal; Ginette Dionne and Michel Boivin, Department of Psychology, Laval University.

Funding for this study was provided by the Canadian Institutes of Health Research (MOP 97882). We thank Jocelyn Malo and Marie-Elyse Bertrand for coordinating the data collection and Hélène Paradis for data manage- ment and preparation. We also thank the twins and their families for participating in this study.

Correspondence concerning this article should be addressed to Mara Brendgen, Department of Psychology, University of Quebec at Montreal, C.P. 8888 Succursale Centre-Ville, Montreal, Quebec, Canada H3C 3P8. E-mail: [email protected]

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

ation between youngsters’ genetic risk for depressive symptoms and the evolution of the parent– child relationship quality in adolescence.

The Evolution of the Parent–Child Relationship Quality in Adolescence

Previous research used latent growth curve analysis (LGCA) to portray the normative growth trajectory of the parent– child rela- tionship during adolescence. Overall, these studies support the view of a normative gradual decline of warmth as well as an increase of negativity within the parent– child relationship over adolescence (Laursen, Delay, & Adams, 2010; Shanahan, McHale, Crouter, & Osgood, 2007). However, these studies also suggest that there are significant individual differences in the course of the parent–adolescent relationship quality. To better capture these individual differences, finite mixture modeling—either in the form of group-based trajectory modeling (GBTM) or growth mixture modeling (GMM)—is deemed more appropriate, as it identifies subgroups that follow distinct developmental trajectories (Nagin & Odgers, 2010). To date, only two studies used finite mixture models to examine the parent– child relationship quality during adolescence (Kim, Thompson, Walsh, & Schepp, 2015; Seiffge- Krenke, Overbeek, & Vermulst, 2010). Seiffge-Krenke and col- leagues found three distinct developmental trajectories of adoles- cents’ relationship quality with their mother and with their father from age 14 to age 17 years: (a) normative (i.e., high and grad- ually declining support-closeness and low and stable negativity: 60.53% of mother–adolescent relationships, 73.25% of father– adolescent relationships), (b) increasingly negative (i.e., low and declining support-closeness and increasing negativity: 29.82% of mother–adolescent relationships, 12.28% of father– adolescent relationships), and (c) decreasingly negative/distant (i.e., low and declining support-closeness and decreasing neg- ativity: 9.65% of mother–adolescent relationships, 5.26% of father–adolescent relationships). Kim and colleagues also iden- tified three trajectories of conflict in the parent–adolescent relationship (high-decreasing trajectory: 13.64% of parent– adolescent relationships; low-increasing trajectory: 9.09% of parent–adolescent relationships; low-stable trajectory: 77.27% of parent–adolescent relationships), as well as one trajectory for support, characterized by a slight decline in support across time for all parent–adolescent relationships.

Although both studies highlight individual differences in the evolution of the parent–adolescent relationship quality, they also present some limitations. First, potential sex differences within parent–adolescent dyads were not formally tested. Although some research showed no sex differences in the quality of the parent– adolescent relationship (McGue, Elkins, Walden, & Iacono, 2005), other findings suggest that girls may have a higher relationship quality with their mother than boys (Branje, Hale, Frijns, & Meeus, 2010; Furman & Buhrmester, 1992), whereas boys per- ceive the relationship with their father as being closer than girls (Furman & Buhrmester, 1992; Starrels, 1994). Thus, the sex- composition of the parent–adolescent dyad should be considered to clarify these associations. Second, the Seiffge-Krenke et al. study did not examine individual characteristics that may explain why youths differ in their trajectories of the parent–adolescent relation- ship quality and Kim et al.’s study only considered parental de-

pression as a predictor of the different trajectories. Although many factors may influence changes in the relationship quality with parents during adolescence, youths’ behaviors— especially depres- sive behavior symptoms—may play a particularly important role.

Depressive Symptoms as a Child-Driven Effect on Parent–Adolescent Relationship Quality

Past research examining the developmental links between the parent–adolescent relationship quality and adolescents’ behaviors, including depressive symptoms, has often conceptualized the for- mer as a predictor of the latter (for a review, see Laursen & Collins, 2009). However, several scholars maintain that the poten- tial influence of the child’s characteristics on parental behavior, and thus the parent– child relationship quality, is equally important (e.g., Kuczynski, 2002; Laursen & Collins, 2009). The Interper- sonal Theory of Depression (Coyne’, 1976) as well as the Gener- ation Hypothesis of Depression (Hammen, 2006) explain how youths’ own depressive symptoms may act as a child-driven effect on the quality of the parent–adolescent relationship. According to these theories, depressed individuals may display aversive behav- iors such as irritability, apathy, insecurity and negativity, which in turn can elicit interpersonal rejection from others (Coyne, 1976) and thus lead to the development of relational stress (Hammen, 2006). Empirical evidence for a predictive association between depressive symptoms, interpersonal rejection, and stress has been found both for clinical depression and depressive symptomatology in children, adolescents, and adults (Eberhart, Auerbach, Bigda- Peyton, & Abela, 2011; Flynn & Rudolph, 2011; Gibb & Hanley, 2010; Liu & Alloy, 2010; Segrin & Dillard’s, 1992). A study by Branje and colleagues (Branje et al., 2010) suggests that this predictive effect of depression symptoms also applies to the parent–adolescent relationship: The results showed that adoles- cents’ elevated depressive symptoms at one point in time were related to poorer quality (i.e., less support and more conflict) of the relationship with the mother and with the father up to two years later.

The results of the Branje et al. study (2010) are in line with Coyne’s (1976) and Hammen’s (2006) theories, suggesting the development of a less supporting and more hostile relationship with parents in reaction to depressed adolescents’ aversive behav- iors. However, that study did not examine to what extent youths’ depressive behavior symptoms predict different trajectories of the relationship with their parents over the course of adolescence. Moreover, the link between adolescents’ depressive symptoms and the quality of the parent–adolescent relationship might differ de- pending on the adolescent’s and/or the parent’s sex. First, with regard to the sex of the adolescent, males may be more at risk than females of eliciting hostile reactions from others when displaying depressive behaviors (Hammen & Peters, 1977; Joiner, Alfano, & Metalsky, 1992). Whether these sex differences also apply to the parent–adolescent relationship is unclear, however. Indeed, studies that assessed the concurrent association between adolescents’ de- pressive symptoms and parent–adolescent relationship quality re- ported similar findings for boys and girls (Eberhart, Shih, Ham- men, & Brennan, 2006; Sheeber, Davis, Leve, Hops, & Tildesley, 2007). Second, mothers and fathers may react differently to their offspring’s depressive symptoms. Thus, a longitudinal study of depressed adolescents and their parents from ages 12 to 19 years

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

2 BROUILLARD, BRENDGEN, VITARO, DIONNE, AND BOIVIN

found that mothers tend to react by increasing their implication toward their child, whereas fathers respond more passively, dis- tancing themselves from the relationship (Sheeber & Sorensen, 1998).

Overall, the previously mentioned studies suggest the presence of a depression-rejection effect within the parent– child relation- ship context. Like most research examining personal predictors of parent– child relationship quality, these studies are based on a correlational design using one child per family. Such designs, however, cannot provide a completely valid test of a causal link. This limitation is true even for transactional longitudinal studies, which are typically considered the most stringent for testing the directionality of association between two variables (Boivin, Petit- clerc, Feng, & Barker, 2010). One alternative is the use of a genetically informed design, such as a behavioral genetic study based on twins. As argued by Moffitt (2005), by disentangling genetic from nongenetic sources of interindividual variance, be- havioral genetic studies can provide a more comprehensive test of transactional processes between individual, potentially heritable characteristics and environmental experiences than correlational studies using singleton samples. Moffitt (2005) maintains that, although such designs cannot provide conclusive proof of causa- tion, they offer notable advantages for testing developmental hy- potheses concerning child-driven effects that may derive from gene-environment correlation (rGE).

Gene–Environment Correlation as an Indicator of Child-Driven Effects

The term gene-environment correlations (rGE) typically de- scribes a situation where individuals’ genotype is associated with the kind of environment they experience (Scarr & McCartney, 1983). Genetically informative studies such as those based on twin designs suggest that adolescents’ depressive symptoms is to a significant extent explained by genetic factors, with reported esti- mates ranging between 23% and 45% of explained variance (Hicks, DiRago, Iacono, & McGue, 2009; Lau & Eley, 2008). rGE would be indicated if individuals with a stronger genetic risk for depressive behavior symptoms also experience a more problematic trajectory of relationship quality with parents during adolescence. rGEs can arise through different processes. Of specific interest for the association between depressive symptoms and the parent– adolescent relationship are passive and evocative rGEs (Jaffee & Price, 2007). A passive rGE occurs when parents provide their child with a specific environment related to their own genetic predispositions, in addition to passing their genes on to their offspring. In the context of the parent– child relationship, it is possible that parents with a genetic vulnerability to depression, in addition to transmitting this predisposition to their offspring, ex- pose them to a more conflicted or less warm family environment. In contrast, an evocative rGE arises when an individual’s geneti- cally influenced characteristics elicit or evoke specific reactions from their environment. In the context of the parent– child rela- tionship, adolescents’ genetically influenced depression-related behavior may elicit negative reactions from parents, thus creating a more conflict-ridden and less warm parent– child relationship.

In their meta-analysis of 56 twin- and adoption studies, Klahr and Burt (2014) found that—in addition to parents’ own geneti- cally influenced characteristics— children’s genetically influenced

characteristics also play a major role in predicting positive and negative parent– child interactions, highlighting the role of evoc- ative rGE. At least one study also reported specific evidence that youths’ depression symptoms, which are significantly influenced by genetic factors (Hicks et al., 2009; Lau & Eley, 2008), are associated with more family chaos and parents’ negative feelings and behaviors toward their offspring via evocative rGE (Wilkin- son, Trzaskowski, Haworth, & Eley, 2013). However, no study so far has examined the specific link between the expression of adolescents’ genetic vulnerability to depressive symptoms and the evolution of the positive and negative features of relationship quality with their mother and their father over the course of adolescence.

The Present Study

The first objective of the present study was to identify sub- groups of adolescents who follow distinct trajectories of relation- ship quality with their mother and with their father from early to late-adolescence, that is, from Grade 7 (Time 1) to Grade 11 (Time 4). Based on Seiffge-Krenke et al.’s study (2010), we expected that three trajectories referring to warmth and three trajectories refer- ring to conflict would be identified for each relationship and each relationship aspect: (a) a stable relationship trajectory of warmth and conflict, (b) an improving relationship trajectory (increasing warmth/decreasing conflict), and (c) a deteriorating relationship trajectory (decreasing warmth/increasing conflict). Distinguishing between the relationship with the mother and with the father allowed us to examine whether these trajectories differ according to the sex of the parent. The second objective was to explore whether the sex of the adolescents would predict their following of a specific relationship quality trajectory with each of their parents. However, because some studies suggest a higher relationship qual- ity within same-sex parent– child dyads (Branje et al., 2010; Fur- man & Buhrmester, 1992; Starrels, 1994), whereas others found no sex-specific patterns (McGue et al., 2005), no specific predictions could be made. The third objective was to investigate whether adolescents’ genetic vulnerability for depressive symptoms—via actual depressive symptoms as a mediating variable—predicts their belonging to a particular trajectory of warmth and/or conflict with the mother and the father, while controlling for parents’ depressive symptoms and family stress. Following Coyne’s inter- personal theory of depression (1976) and Hammen’s (2006) stress generation hypothesis of depression, as well as Wilkinson et al.’s (2013) findings of evocative rGE linking adolescents’ depressive symptoms with a more negative family environment, we expected that a stronger genetic disposition for depressive symptoms would be indirectly associated with a parent– child relationship trajectory of poorer quality, via adolescents’ actual depressive behavior symptoms as a mediating mechanism. We expected to find these results even when controlling for parents’ depressive symptoms and family stress. Parental depression and family stress likely reflect not only environmental influences but also, at least in part, parents’ genetic vulnerability to depression and related mental health problems. When controlling for these variables, a predictive effect of adolescents’ genetic vulnerability for depressive symp- toms (mediated by adolescents’ actual depressive behavior symp- toms) on the parent–adolescent relationship may thus be more readily interpretable as evocative (rather than passive) rGE. We

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

3TRAJECTORIES OF PARENT-ADOLESCENT RELATIONSHIP QUALITY

also expected that this rGE might be stronger for boys than girls, as depressive behaviors displayed by boys have been found to elicit more negative responses from others (Hammen & Peters, 1977; Joiner et al., 1992). We also expected this rGE to be stronger for the relationship trajectory with the father than with the mother, as fathers’ responses to depressive symptoms have been found to be generally less adapted to the child’s needs than mothers’ re- sponses (Sheeber & Sorensen, 1998).

Method

Participants

The present study utilized a genetically informative design based on a sample of 375 monozygotic (MZ) and 290 dizygotic (DZ) twins from same-sex twin pairs (i.e., a total of 665 individ- uals; 50.2% girls) assessed in Grade 7 (M ! 13.05, SD ! .29), Grade 8 (M ! 14.09, SD ! .30), Grade 9 (M ! 15.08, SD ! .26), and in Grade 11 (M ! 17.07, SD ! .31). They were part of a population-based sample of 467 MZ and same-sex dizygotic DZ twin pairs from the greater Montreal area, who were recruited at birth between November 1995 and July 1998. Zygosity was as- sessed by genetic marker analysis of eight to 10 highly polymor- phous genetic markers and twins were diagnosed as MZ when concordant for every genetic marker. When genetic material was insufficient, zygosity was determined based on physical resem- blance questionnaires at 18 months and again at age 9 (Goldsmith, 1991; Spitz et al., 1996). The comparison of zygosity based on genotyping with zygosity based on physical resemblance in a subsample of 237 pairs revealed a 94% correspondence rate, which is extremely similar to rates obtained in other studies (Magnusson et al., 2013; Spitz et al., 1996). The sample was followed longi- tudinally in the first years of their life, then in preschool and elementary school, as well as in high school. Regarding socioeco- nomic characteristics, 84% percent of the families were of Euro- pean descent, 3% were of African descent, 2% were of Asian descent, and 2% were Native North Americans. The remaining families (9%) did not provide ethnicity information.

The demographic characteristics of the twin families were com- parable with those of a sample of single births representative of the urban centers in the province of Quebec (Santé Québec, Jetté, Desrosiers, & Tremblay, 1998). The same percentage (95%) of parents in both samples lived together at the time of birth of their child(ren); 44% of the twins compared with 45% of the singletons were the firstborn children in the family; 66% of the mothers and 60% of the twins’ fathers were between 25 and 34 years old, compared with 66% of mothers and 63% of fathers for the single- tons; 17% of the mothers and 14% of the twins’ fathers had not finished high school, compared with 12% and 14% of mothers and fathers, respectively, for the singletons; the same proportion of mothers (28%) and fathers (27%) in both samples held a university degree; 83% of the twin parents and 79% of singleton parents were employed; 10% of the twin families and 9% of the singleton families received social welfare or unemployment insurance; and, finally, 30% of the twin families and 29% of the singleton families had an annual total income of less than Can$30,000, 44% and 42% had an annual total income between Can$30,000 and Can$59,999, and 27% and 29% had an annual total income of more than Can$60,000.

To be included in the analyses of the present study, participants had to have valid data on at least one of the four time points of the dependent variable (i.e., relationship quality with mother or father) from Grades 7 to 11. With an average yearly attrition rate of approximately 2% over the course of the longitudinal study, these criteria resulted in the aforementioned total study sample of 665 individuals. Multilevel regressions showed that participants in- cluded in the final study sample did not differ from those lost due to attrition regarding family status, parents’ age at the twins’ birth and parents’ level of education. However, family income was higher among participants included in the present study.

Measures

Relationship quality. In Grades 7, 8, 9, and 11, adolescents’ perceptions of the relationship quality with their mother and their father were assessed with 10 items from the Network of Relation- ships Inventory (NRI; Furman & Buhrmester, 1985, 1992). Six items referred to positive relationship attributes related to warmth (e.g., “How much does this person like or love you?” “How much does this person treat you like you’re good at many things?” “How much does this person help you figure out or fix problems?”) and four items referred to negative relationship attributes related to conflict (e.g., “How often do you and this person get mad at or get into fights with each other?” “How often do you and this person disagree and quarrel with each other?”). The participant had to score each item on a 5-point Likert-type scale ranging from 1 little or never to 5 most of the time. Individual items scores were averaged to compute scale scores, separately for warmth and for conflict in the relationship with each parent in Grades 7, 8, 9, and 11, respectively. Table 1 presents the internal consistencies, means, and standard deviations of warmth and conflict in the mother–adolescent and the father–adolescent relationships for the overall sample as well as means and standard deviations according to sex and grade.

Depressive symptoms. In Grades 7, 8, 9, and 11, adolescents completed the short version (10 items) of the Children’s Depres- sion Inventory (CDI; Kovacs, 1992). The excellent validity and fidelity of the CDI have been demonstrated many times (Smucker, Craighead, Craighead, & Green, 1986). The CDI assesses a variety of depressive symptoms such as disturbed mood, hedonic capacity, vegetative functions, self-evaluation, and interpersonal behaviors. For each item, participants had to choose the response option (ranging from 0 to 2) that represents them most accurately. For example: “Since the last two weeks . . . I am sad once in a while (0), I am sad many times (1), I am sad all the time” (2) or “Since the last two weeks . . . I succeed in almost everything I do (0), I fail at lots of things (1), I fail at everything” (2). Four of the 10 items were reversed and the mean of individual item scores was calcu- lated to create individual scale scores of depressive symptoms in each grade: Grade 7 (M ! .22, SD ! .24, min ! 0, max ! 1.3, " ! .75); Grade 8 (M ! .25, SD ! .27, min ! 0, max ! 1.8, " ! .79); Grade 9 (M ! .27, SD ! .28, min ! 0, max ! 1.6, " ! .79); and Grade 11 (M ! .29, SD ! .30, min ! 0, max ! 1.6, " ! .80). Depressive symptoms between adjacent time points were rela- tively stable, varying between r ! .41 and r ! .64. Scale scores were thus averaged across the four time points to obtain a measure of adolescents’ tendency for depression symptoms over time (M ! .26, SD ! .22, min ! 0, max ! 1.4).

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

4 BROUILLARD, BRENDGEN, VITARO, DIONNE, AND BOIVIN

Genetic risk for depressive symptoms. When data are col- lected on twin pairs, it is possible to estimate the relative influence of genetic factors on observed traits or behaviors such as depres- sion symptoms. Specifically, genetic factors are implicated if identical or monozygotic (MZ) twins, who share 100% of their genetic makeup, are more similar to each other than are same-sex fraternal or dizygotic (DZ) twins, who share on average only 50% of their genes (Petrill, 2003). This pattern was also observed in the present sample, with an intrapair correlation of at least .41 for MZ twin pairs versus .24 for DZ twin pairs. Using information about the pair’s genetic relatedness (as indicated by zygosity) and their levels of depressive symptoms, it is also possible to calculate an ordinal score of each individual’s genetic risk for depressive symptoms (Andrieu & Goldstein, 1998). This method has been used in several studies to test gene-environment interactions and correlations with an epidemiological twin design (e.g., Brendgen et al., 2013; Guimond et al., 2014; Jaffee et al., 2005; Wichers et al., 2009). Specifically, one twin from each twin pair was selected as the target twin and the second twin as the co-twin. Each twin pair was represented in the data set twice, with each twin of a pair serving once as the target twin and another time as the cotwin. For each individual, an ordinal score of genetic risk for depressive symptoms was then computed as a function of (a) the pair’s genetic relatedness and (b) the presence or absence of depressive symptoms in the cotwin. To this end, the global depressive symp- toms score was dichotomized using the 75th percentile as the cut-off, which corresponds to a score of 0.4 on the average of the 4 short versions of the CDI completed at each time point. This cut-off score allowed us to identify the most depressed adolescents in this normative sample, while assuring a sufficient sample size for statistical analysis. A similar cut-off was also used in other studies on depressive symptoms (Brendgen, Vitaro, Turgeon, & Poulin, 2002; Brendgen et al., 2013). For each individual, the information about the presence or absence of depressive symptoms in the cotwin was then combined with information on the pair’s genetic relatedness into an index of genetic risk for depressive symptoms, ranging from 1 (very low) to 4 (very high). Specifically, when an individual was part of an MZ pair (who share 100% of

their genetic material) and when depressive symptoms were pres- ent in the cotwin, the individual’s genetic risk for depressive symptoms was considered to be very high (13.46% of the sample). An individual’s genetic risk for depressive symptoms was some- what lower, albeit still relatively high, when he or she was part of a DZ pair (who on average share 50% of their genes) and when depressive symptoms were present in the cotwin (9.68%). An individual’s genetic risk for depressive symptoms was relatively low when he or she was part of a DZ pair and when depressive symptoms were absent in the cotwin (33.93%). An individual’s genetic risk for depressive symptoms was very low when he or she was part of an MZ pair and when depressive symptoms were absent in the cotwin (42.63%).

Family stress. Based on a measure used in previous research (e.g., Poirier et al., 2016), a composite family stress index was created using parent reports on: (a) family status (twins living with both biological parents or not) in Grade 6, (b) marital satisfaction in Grade 6 reported by the mother and by the father, based on eight items from the Dyadic Adjustment Scale (Spanier, 1976; " ! .86 for mothers, " ! .87 for fathers), (c) mother’s and father’s level of education (assessed when the twins were 5, 18, 30, and 60 months), and (d) family revenue (also assessed when the twins were 5, 18, 30, and 60 months). A score of 0 was attributed to family status if the child was living with both natural parents and a score of 1 was attributed to all other cases. Mother’s marital dissatisfaction was scored 1 for values in the lowest quartile of the marital satisfaction scale and 0 for all other values. Father’s marital dissatisfaction was also scored 1 for values in the lowest quartile of the marital satisfaction scale and 0 for all other values. A score of 1 was attributed to the mother’s level of education when she did not have a high school diploma, and a score of 0 was attributed to all other cases. Similarly, a score of 1 was attributed to the father’s level of education when he did not have a high school diploma and a score of 0 was attributed to all other cases. A score of 1 was attributed to family revenue if the family annual revenue was below $30,000 more than 50% of the time between the twins were aged 5 and 60 months, and a score of 0 was attributed to all other

Table 1 Descriptive Statistics of Warmth and Conflict in the Parent–Adolescent Relationship

Scale " M (SD)

whole sample M (SD)

girls M (SD)

boys

Warmth mother Grade 7 .81 3.84 (0.79) 3.88 (0.82) 3.78 (0.75) Warmth father Grade 7 .83 3.58 (0.85) 3.59 (0.86) 3.57 (0.84) Conflict mother Grade 7 .83 1.95 (0.71) 1.99 (0.72) 1.90 (0.69) Conflict father Grade 7 .83 1.87 (0.72) 1.87 (0.73) 1.86 (0.72) Warmth mother Grade 8 .86 3.50 (0.87) 3.60 (0.86) 3.39 (0.88) Warmth father Grade 8 .88 3.23 (0.96) 3.28 (1.00) 3.18 (0.92) Conflict mother Grade 8 .84 2.06 (0.73) 2.12 (0.72) 1.99 (0.64) Conflict father Grade 8 .85 1.95 (0.75) 2.04 (0.83) 1.84 (0.71) Warmth mother Grade 9 .84 3.71 (0.82) 3.73 (0.83) 3.68 (0.81) Warmth father Grade 9 .87 3.38 (0.94) 3.32 (0.96) 3.46 (0.91) Conflict mother Grade 9 .87 2.11 (0.80) 2.20 (0.80) 2.01 (0.80) Conflict father Grade 9 .85 2.02 (0.79) 2.10 (0.85) 1.91 (0.71) Warmth mother Grade 11 .86 3.71 (0.83) 3.74 (0.86) 3.68 (0.80) Warmth father Grade 11 .88 3.37 (0.94) 3.29 (0.98) 3.44 (0.88) Conflict mother Grade 11 .91 2.17 (0.84) 2.30 (0.90) 2.03 (0.76) Conflict father Grade 11 .87 2.04 (0.82) 2.12 (0.86) 1.95 (0.78)

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

5TRAJECTORIES OF PARENT-ADOLESCENT RELATIONSHIP QUALITY

cases. A total family stressors index was computed by summing the individual stressors (M ! .94, SD ! .99, min ! 0, max ! 6).

Parental depression. Mother’s and father’s depression was assessed when the twins were in Grade 6. A score of 1 was given when the parent confirmed having experienced a depressive dis- order in the past 9 years, and a score of 0 was attributed to the parent when he or she did not experience a depressive episode. Mother’s and father’s depression were included separately in the analysis (mother depression: M ! .42, SD ! .49, min ! 0, max ! 1; father depression: M ! .28, SD ! .45, min ! 0, max ! 1).

Procedure

All instruments were administered in paper-and-pencil format in either English or French, depending on the language spoken by the families. The French version of the CDI was drawn from the validated French questionnaire (Saint-Laurent, 1990). For all other instruments, following the procedure suggested by Vallerand (1989), instruments that were administered in French but were originally written in English were translated into French and then translated back to English. Bilingual judges verified the semantic similarity between the back-translated items and the original items. Parents were contacted by letter and active written consent from parents and adolescents was obtained. Data collection took place in the spring during home interviews and took approximately one hour. Instruments and procedures were in accordance with APA ethical standards and were approved by the Institutional Review Board of the Ste. Justine Hospital Research Centre (project title: “Peer abuse and adolescent health”; ethics certificate number 3039).

Analyses

Preliminary analyses. Prior to the main analysis, bivariate correlations between all study variables were examined (see Table 2). As shown, depressive symptoms were positively corre- lated with conflict with the mother and with the father from Grades 7 to 11, and negatively correlated with warmth from the mother and from the father from Grades 7 to 11. Genetic risk for depres- sive symptoms was also correlated with the majority of parent– adolescent relationship attributes in the expected direction. Re- garding control variables, sex, family stress, and parental depression were each correlated with attributes of relationship quality with the mother and the father at several time points. In turn, the attributes of the relationship quality with the mother and the father were correlated with each other most of the time in the expected direction.

Main analyses. Distinct trajectories of warmth and conflict within the mother–adolescent and father–adolescent relationships, respectively, were identified with the Mplus Version 8 software (Muthén & Muthén, 1998 –2017), using Group-Based Trajectory Modeling (Nagin & Odgers, 2010). In contrast to standard Growth Mixture Modeling, within-group parameter variances are not al- lowed to vary in Group-based Trajectory Modeling. This approach was chosen because our goal was to identify a finite number of groups to approximate the unknown distribution of trajectories within the population, rather than assuming that the population distribution of trajectories is composed of truly distinct subpopu- lations (Nagin & Odgers, 2010). Several series of models were

fitted to the data, starting with a one-group trajectory model up to a five-group trajectory model. In accordance with the Group-Based Trajectory Modeling approach (Nagin & Odgers, 2010), all within- group variances were fixed to 0. The best fitting model was established based on the Bayesian Information Criteria (BIC), the Entropy, the Lo-Mendell-Rubin likelihood ratio test (LMR-LRT), and the average posterior assignment probabilities that assess the accuracy of an individual’s assignment to a specific trajectory. The BIC is a fit index where lower values indicate a more parsimonious model, whereas Entropy is a measure of classification accu- racy with values closer to 1 designating greater precision. The LMR-LRT establishes the ideal number of trajectories, with a p value below .05 indicating that the k trajectory model is a better fit to the data compared with the k # 1 trajectory model. Finally, average posterior probabilities greater than .70 to .80 show that the modeled trajectories assemble individuals with similar longitudi- nal profiles and discriminate between individuals with dissimilar longitudinal profiles. All trajectory models were initially estimated including linear as well as quadratic trends. Trends that did not reach statistical significance were subsequently removed and mod- els were rerun to achieve maximum parsimony. To account for missing data (4.40% of data points) and for data interdependency attributable to twinning, Full Information Maximum Likelihood- Robust (FMLR) was used to fit the data, as well as the COMPLEX option for adjusting standard error estimates.

To test the expected predictive associations between our inde- pendent variables and the parent– child relationship trajectories, we followed the three-step approach described by Asparouhov and Muthén (2014; see also Vermunt, 2010). Specifically, after having determined in Step 1 the number of distinct trajectories that best fit the data based on the observed variable indicators, the best fitting model’s posterior assignment probabilities were used to calculate each individual’s most likely trajectory group membership and classification uncertainty rate in Step 2. These variables were calculated for the positive and the negative relationship quality with the mother and the father, respectively. Next (Step 3), the group-based trajectory models were rerun, separately for the pos- itive and the negative relationship quality with the mother and the father, with the most likely trajectory group membership used as a latent class indicator variable with fixed logits based on the un- certainty rates obtained in the previous step. This approach ac- counts for classification uncertainty when examining predictors of trajectory group membership. In that final step, we included the independent variables to predict individuals’ odds of following their most likely trajectory. The model comprised the following predictor variables: sex, family stress, mother’s and father’s de- pression scores, adolescents’ genetic vulnerability for depressive symptom as well as a two-way interaction between sex and ado- lescents’ genetic vulnerability for depressive symptoms. We also included adolescents’ actual depressive symptoms and a two-way interaction between sex and adolescents’ actual depressive symp- toms, and these variables were, in turn, regressed on the previously mentioned predictor variables. Finally, we examined (a) the total effect of adolescents’ genetic risk for depressive symptoms and the odds of following a specific relationship trajectory, and (b) the indirect effect between adolescents’ genetic risk for depressive symptoms and the odds of following a specific relationship trajec- tory, with adolescents’ actual depressive symptoms as the inter- vening variable. Significance of the total and indirect effects was

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

6 BROUILLARD, BRENDGEN, VITARO, DIONNE, AND BOIVIN

T ab

le 2

B iv

ar ia

te C

or re

la ti

on s

of th

e St

ud y

V ar

ia bl

V ar

ia bl

e 1

2 3

4 5

6 7

8 9

10 11

12 13

14 15

16 17

18 19

20 21

1. S

ex —

2. M

ot he

r de

pr es

si on

a #

.0 4

— 3.

F at

he r

de pr

es si

on a

.0 3

.1 7!

— 4.

F am

il y

st re

ss a

.0 2

.3 3

.3 5!

— 5.

D ep

re ss

io n

G ra

de 7–

11 #

.2 5!

.0 6!

# .0

1 .0

2† —

6. G

en et

ic ri

sk G

ra de

7– 11

# .1

7! .2

6! #

.0 3

.1 0!

.4 3!

— 7.

W ar

m th

m ot

he r

G ra

de 7

# .0

6 #

.0 3

# .0

6 #

.2 2!

# .1

6! —

8. W

ar m

th fa

th er

G ra

de 7

# .0

1 #

.1 5†

# .0

5 #

.1 7!

# .2

4! #

.1 6!

.7 3!

— 9.

C on

fl ic

t m

ot he

r G

ra de

7 #

.0 7

.1 2†

# .0

6 .0

5 .3

1! .1

8! #

.2 3!

# .1

5! —

10 .

C on

fl ic

t fa

th er

G ra

de 7

# .0

1 .1

1† .0

5 .0

1 .2

7! .0

9† #

.0 9!

# .2

2! .6

3! —

11 .

W ar

m th

m ot

he r

G ra

de 8

# .1

2! #

.1 2

.0 9

# .0

1 #

.2 7!

# .2

1! .5

4! .3

7! #

.1 9!

# .1

3! —

12 .

W ar

m th

fa th

er G

ra de

8 #

.0 5

# .1

9! #

.0 3

# .0

7 #

.3 1!

# .2

3! .3

9! .5

5! #

.1 2!

# .2

4! .7

6! —

13 .

C on

fl ic

t m

ot he

r G

ra de

8 #

.0 9!

.1 0

# .0

6 .0

3 .2

8! .1

8! #

.1 7

# .0

3 .5

9! .3

5! #

.2 4!

# .1

3! —

14 .

C on

fl ic

t fa

th er

G ra

de 8

# .1

3! .2

3! .0

8 #

.0 2

.3 1!

.1 6!

# .0

8 #

.1 8!

.3 0!

.6 0!

# .1

0! #

.2 9!

.4 7!

— 15

. W

ar m

th m

ot he

r G

ra de

9 #

.0 3

# .2

0! .1

2 #

.0 1

# .3

1! #

.2 0!

.4 7!

.3 3!

# .2

4! #

.1 1!

.6 1!

.4 9!

# .2

4! #

.1 9!

— 16

. W

ar m

th fa

th er

G ra

de 9

# .0

8† #

.3 0!

# .0

6 #

.1 3!

# .3

7! #

.2 8!

.3 1!

.4 8!

# .1

1! #

.2 2!

.4 2!

.6 6!

# .1

2! #

.3 5!

.6 5!

— 17

. C

on fl

ic t

m ot

he r

G ra

de 9

# .1

2! .1

1 #

.1 0

# .0

03 .3

9! .2

1! #

.2 0!

# .1

0! .5

7! .3

1! #

.2 6!

# .1

7! .6

4! .2

5! #

.4 0!

# .2

1! —

18 .

C on

fl ic

t fa

th er

G ra

de 9

# .1

2! .2

4! .1

6 #

.0 3

.3 9!

.1 9!

# .0

8† #

.1 9!

.4 7!

# .1

5! #

.3 3!

.2 6!

.6 1!

# .1

8! #

.4 5!

.4 3!

— 19

. W

ar m

th m

ot he

r G

ra de

11 #

.0 3

# .0

2 .0

5 #

.0 4

# .2

8! #

.1 2!

.4 2!

.3 4!

# .2

2! #

.0 8

.4 7!

.4 0!

# .1

5! #

.0 9†

.6 1!

.3 0!

# .3

0! #

.1 5!

— 20

. W

ar m

th fa

th er

G ra

de 11

.0 8

# .1

9! #

.0 03

# .1

3! #

.3 3!

# .1

9! .2

8! .5

0! #

.0 6

# .1

1! .3

3! .5

4! #

.0 3

# .2

0! .4

2! .6

4! #

.1 0†

# .2

7! .6

3! —

21 .

C on

fl ic

t m

ot he

r G

ra de

11 #

.1 6!

.0 4

# .0

2 .0

7 .3

6! .1

7! #

.1 7!

# .1

0† .4

3! .1

7! #

.1 6!

# .1

1! .4

7! .1

8! #

.2 6!

# .1

3! .6

0! .2

6! #

.4 0!

# .2

0! —

22 .

C on

fl ic

t fa

th er

G ra

de 11

# .1

0! .3

0! .0

7 #

.0 1

.3 2!

.1 2!

# .0

5 #

.1 6!

.1 7!

.3 4!

# .0

8 #

.2 1!

.2 3!

.4 2!

# .1

5! #

.2 3!

.2 5!

.5 3!

# .1

3! #

.4 0!

.5 1!

N ot

e. B

iv ar

ia te

co rr

el at

io ns

be tw

ee n

al l

st ud

y va

ri ab

le s

ar e

sh ow

n be

lo w

th e

di ag

on al

. S

ex is

co de

d su

ch th

at 0

in di

ca te

s gi

rl s

an d

1 in

di ca

te s

bo ys

. a

B iv

ar ia

te as

so ci

at io

ns be

tw ee

n fa

m il

y- le

ve l

va ri

ab le

s an

d in

di vi

du al

tw in

-l ev

el va

ri ab

le s

w er

e ex

am in

ed w

it h

a m

ul ti

le ve

l re

gr es

si on

to ac

co un

t fo

r da

ta in

te rd

ep en

de nc

y at

tr ib

ut ab

le to

tw in

ni ng

. †

p !

10 .

! p

! .0

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

7TRAJECTORIES OF PARENT-ADOLESCENT RELATIONSHIP QUALITY

evaluated using their bias-corrected 95% confidence intervals based on 5,000 bootstrap resamples. For each model, the least problematic trajectory identified in the first step (i.e., trajectories indicating most warmth and least conflict) was used as the refer- ence trajectory. Because none of the sex interaction terms was significant, models were rerun without interaction terms and these final models are presented for parsimony.

Results

Trajectories of Warmth and Conflict in the Mother–Adolescent and Father–Adolescent Relationship

Figure 1 presents the best fitting two-group trajectory model and the fit indices for the one- to five-group trajectory models of warmth in the mother–adolescent relationship from Grades 7 to 11. Although the BIC decreased with increasing number of groups, the three-group trajectory model did not provide superior fit compared with the two-group model based on the LMR-LRT. Moreover, the two-group trajectory model had an entropy value closest to 1 of all tested models and showed excellent average posterior assignment probabilities (ranging from .91 to .92). Examination of the two- group model revealed one group (38.95% or 259 participants) who experienced a moderate level of warmth in Grade 7, which slightly decreased in Grades 8 and 9, before increasing again in Grade 11 (Medium-Warmth Decreasing-Increasing group; Intercept ! 3.20, p $ .001; Linear trend ! #0.30, p $ .001; Quadratic trend: 0.07, p $ .001). The second group (61.05% or 406 participants) expe- rienced a high level of warmth in Grade 7, which slightly de- creased in Grades 8 and 9 before increasing again in Grade 11

(High-Warmth Decreasing-Increasing group; Intercept ! 4.16, p $ .001; Linear trend ! #0.09, p ! .02; Quadratic trend: 0.02, p ! .01).

Figure 2 presents the best fitting three-group model and the fit indices for the one- to five-group models of warmth in the father– adolescent relationship from Grades 7 to 11. Here, the BIC stopped decreasing after the four-group model. However, LMR-LRT indi- cated that the 4-group model did not provide superior fit compared with the three-group model. The three-group model also had a slightly superior entropy value than the four-group model, and it showed better average posterior assignment probabilities, ranging from .84 to .88. Examination of that model revealed one group (15.68% or 101 participants) who experienced lower warmth in Grade 7, which decreased in Grades 8 and 9, before increasing in Grade 11 (Low-Warmth Decreasing-Increasing group; Intercept ! 2.42, p $ .001; Linear trend ! #0.40, p $ .001; Quadratic trend ! 0.09, p $ .01). A second group (32.14% or 207 partici- pants) experienced high and stable warmth from Grade 7 to 11 (High-Warmth Stable group; Intercept ! 4.17, p $ .001). The third group (52.17% or 336 participants) initially experienced a moderate level of warmth in Grade 7, which decreased in Grades 8 and 9 before increasing in Grade 11 (Medium-Warmth Decreasing-Increasing group; Intercept ! 3.43, p $ .001; Linear trend ! #.31, p $ .001; Quadratic trend ! .07, p $ .001).

Figure 3 presents the best fitting three-group model and the fit indices for the one- to five-group models of conflict in the mother– adolescent relationship from Grades 7 to 11. Although the BIC decreased with increasing number of groups, LMR-LRT indicated that the four-group model did not provide superior fit compared with the three-group model. The latter model also had a higher Entropy value than the four-group and five-group models, and it

1-Group Model

2-Group

Model

3-Group

Model

4-Group

Model

5-Group

Model

AIC 5285.023 4807.967 4721.412 4676.844 4655.729 BIC 5304.022 4843.966 4775.409 4748.84 4745.725

Entropy -- 0.707 0.607 0.668 0.636 LMR-LRT p-value -- 0 0.1705 0 0.4889

0 0.5

1 1.5

2 2.5

3 3.5

4 4.5

Grade 7 Grade 8 Grade 9 Grade 11

L ev

el o

f M ot

he r -

W ar

m th

Medium-Warmth Decreasing- Increasing (38.95%)

High-Warmth Decreasing- Increasing (61.05%)

Figure 1. Predicted longitudinal profiles of warmth in the mother–adolescent relationship from Grade 7 to Grade 11 (i.e., age 13 to age 17) based on the best fitting (i.e., the two-group) model.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

8 BROUILLARD, BRENDGEN, VITARO, DIONNE, AND BOIVIN

showed excellent average posterior assignment probabilities, rang- ing from .88 to .96. Examination of the three-group model revealed one group (3.33% or 22 participants) who experienced a high and stable level of conflict from Grade 7 to Grade 11 (High-Conflict Stable group; Intercept ! 4.04, p $ .001). The second group (73.68% or 490 participants) experienced a low and stable level of conflict from Grade 7 to 11 (Low-Conflict Stable group; Inter- cept ! 1.77, p $ .001). The last group (23.01% or 153 partici- pants) experienced a moderate level of conflict in Grade 7, which

gradually increased up to Grade 11 (Medium-Conflict Increasing group; Intercept ! 2.41, p $ .001; Linear trend ! .16, p $ .001).

Figure 4 presents the best fitting two-group model and the fit indices for the one- to five-group trajectory models of conflict in the father–adolescent relationship from Grades 7 to 11. Although the BIC decreased with increasing number of groups, models with more than two groups did not provide superior fit according to the LMR-LRT. The two-group model also had a higher Entropy value than the three-group model, and it showed excellent average pos-

1-Group Model

2-Group Model

3-Group Model

4-Group Model

5-Group Model

AIC 5303.981 4833.558 4687.759 4649.93 4638.151 BIC 5321.852 4869.299 4732.436 4721.413 4727.505 Entropy -- 0.681 0.691 0.66 0.704 LMR-LRT p-value -- 0 0.0029 0.2587 0.0154

0 0.5

1 1.5

2 2.5

3 3.5

4 4.5

Grade 7 Grade 8 Grade 9 Grade 11

L ev

el o

f F at

he r-

W ar

m th

Low-Warmth Decreasing- Increasing (15.68%)

High-Warmth Stable (32.14%)

Medium Warmth Decreasing- Increasing (52.17%)

Figure 2. Predicted longitudinal profiles of warmth in the father–adolescent relationship from Grade 7 to Grade 11 (i.e., age 13 to age 17) based on the best fitting (i.e., the three-group) model.

1-Group Model

2-Group Model

3-Group Model

4-Group Model

5-Group Model

AIC 4952.8 4388.157 4212.266 4123.274 4067.194 BIC 4970.799 4424.155 4243.765 4195.271 4157.189 Entropy -- 0.852 0.835 0.794 0.699 LMR-LRT p-value -- 0.0001 0.0002 0.0695 0.0126

0.5

1.5

2.5

3.5

4.5

Grade 7 Grade 8 Grade 9 Grade 11

L ev

el o

f M ot

he r-

C on

fli ct

High-Conflict Stable (3.33%)

Low-Conflict Stable (73.68%)

Medium-Conflict Increasing (23.01%)

Figure 3. Predicted longitudinal profiles of conflict in the mother–adolescent relationship from Grade 7 to Grade 11 (i.e., age 13 to age 17) based on the best fitting (i.e., the three-group) model.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

9TRAJECTORIES OF PARENT-ADOLESCENT RELATIONSHIP QUALITY

terior assignment probabilities, ranging from .87 to .96. Examina- tion of the two-group model revealed one group (21.63% or 138 participants) who initially experienced a moderate level of conflict in Grade 7, which significantly increased in Grades 8 and 9 before decreasing slightly in Grade 11 (Medium-Conflict Increasing- Decreasing group; Intercept ! 2.62, p $ .001; Linear trend ! .31, p ! .01; Quadratic trend ! #.06, p ! .04). The second group (78.37% or 500 participants) experienced a low level of conflict, which gradually increased from Grade 7 to Grade 11 (Low- Conflict Increasing group; Intercept ! 1.66, p $ .001; Linear trend ! .03, p ! .002).

Prediction of Parent–Adolescent Relationship Quality Trajectories

With regard to warmth in the mother–adolescent relationship (Table 3 top left), girls were more likely than boys to follow the Medium-Warmth Decreasing-Increasing trajectory relative to the High-Warmth Decreasing-Increasing trajectory (odds ! 2.42, p ! .001). Adolescents with higher depressive symptoms were also more at risk of following the Medium-Warmth Decreasing- Increasing trajectory relative to the High-Warmth Decreasing- Increasing trajectory (odds ! 2.16, p $ .001). The mediation test revealed a significant total effect of genetic vulnerability for de- pressive symptoms on the odds of following the Medium-Warmth Decreasing-Increasing trajectory relative to the High-Warmth Decreasing-Increasing trajectory (total effect b ! .44; 95% boot- strap CI [19, .68]; odds ! 1.50). The indirect effect was also significant, suggesting that a higher genetic vulnerability for de- pressive symptoms increased the odds of following the Medium- Warmth Decreasing-Increasing trajectory via actual depressive symptoms as an intervening variable (indirect effect b ! .31; 95% bootstrap CI [.19, .46]).

With regard to warmth in the father–adolescent relationship (Table 3 top right), a higher level of family stress was associated with increased odds of following the Low-Warmth Decreasing-

Increasing compared with the High-Warmth Stable trajectory (odds ! 1.73, p $ .01). Adolescents with higher depressive symptoms were more likely to follow either the Low-Warmth Decreasing-Increasing trajectory (odds ! 3.94, p $ .001) or the Medium-Warmth Decreasing-Increasing trajectory (odds ! 2.67, p $ .001) compared with the High-Warmth Stable trajectory. Moreover, there were significant total effects of genetic vulnera- bility for depressive symptoms on the odds of following either the Low-Warmth Decreasing-Increasing trajectory (total effect b ! .97; 95% bootstrap CI [.52, 1.47]; odds ! 2.12) or the Medium- Warmth Decreasing-Increasing trajectory (total effect b ! .70; 95% bootstrap CI [.32, 1.20]; odds ! 1.57) compared with the High Warmth Stable trajectory group. The indirect effects of genetic vulnerability for depressive symptoms on these odds—via actual depressive symptoms as an intervening variable—were also significant (indirect effect b ! .57; 95% bootstrap CI [.35, .86] for the Low-Warmth Decreasing-Increasing trajectory and indirect effect b ! .41; 95% bootstrap CI [.21, .69] for the Medium- Warmth Decreasing-Increasing trajectory).

Regarding conflict in the mother–adolescent relationship (Table 3 bottom left), adolescents with higher depressive symptoms were more likely to follow either the High-Conflict Stable trajectory (odds ! 3.37, p $ .001) or the Medium-Conflict Increasing trajectory (odds ! 2.82, p $ .001) compared with the Low- Conflict Stable trajectory. The mediation test revealed a total effect of genetic vulnerability for depressive symptoms on the odds of following the High-Conflict Stable trajectory compared with the Low-Conflict Stable trajectory (total effect b ! .91; 95% bootstrap CI ! [.04, 7.25]; odds ! 1.51). Moreover, genetic vulnerability for depressive symptoms had significant indirect effects—via actual depressive symptoms as an intervening variable— on the odds of following the High-Conflict Stable (indirect effect b ! .50; 95% bootstrap CI [.09, .95]) or the Medium-Conflict Increasing trajec- tories (indirect effect b ! .43; 95% bootstrap CI [.21, .69]) compared with the Low-Conflict Stable trajectory.

1-Group Model

2-Group Model

3-Group Model

4-Group Model

5-Group Model

AIC 4590.166 4147.865 4017.822 3919.879 3875.67 BIC 4608.043 4179.15 4071.453 3991.387 3965.055 Entropy -- 0.788 0.772 0.789 0.79 LMR-LRT p-value -- 0.0006 0.7213 0.0397 0.4661

0.5

1.5

2.5

3.5

Grade 7 Grade 8 Grade 9 Grade 11

L ev

el o

f F at

he r-

C on

fli ct

Medium-Conflict Increasing- Decreasing (21.63%)

Low-Conflict Increasing (78.37%)

Figure 4. Predicted longitudinal profiles of conflict in the father–adolescent relationship from Grade 7 to Grade 11 (i.e., age 13 to age 17) based on the best fitting (i.e., the two-group) model.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

10 BROUILLARD, BRENDGEN, VITARO, DIONNE, AND BOIVIN

Finally, regarding conflict in the father–adolescent relationship (Table 3 bottom right), a higher level of mother’s depression increased the odds of following the Medium-Conflict Increasing- Decreasing trajectory compared with the Low-Conflict Increasing trajectory (odds ! 1.43, p ! .02). Adolescents with higher de- pressive symptoms were also more likely to follow the Medium- Conflict Increasing-Decreasing trajectory compared with the Low- Conflict Increasing trajectory (odds ! 2.35, p $ .001). Moreover, the mediation test showed a significant total effect of a higher genetic vulnerability for depressive symptoms on the odds of following the Medium-Conflict Increasing trajectory compared with the Low-Conflict Increasing trajectory (total effect b ! .34; 95% bootstrap CI ! [.07, .61]; odds ! 1.54). The indirect effect of genetic vulnerability for depressive symptoms on these odds— via actual depressive symptoms—was also significant (indirect effect b ! .35, 95% bootstrap CI [.21, .54]).

Discussion

The present study aimed to identify subgroups of adolescents who follow distinct developmental trajectories of warmth and conflict within the mother–adolescent and the father–adolescents relationships and to test whether the odds of following a specific trajectory vary depending on adolescents’ sex or genetic predis- position to depressive symptoms. In line with previous studies using latent growth curve analysis (Laursen et al., 2010; Shanahan et al., 2007), most adolescents in the present study experienced high quality relationships with both of their parents, characterized

by high levels of warmth and low levels of conflict, and marked by gradual changes over the course of adolescence. Consistent with Smetana’s et al. (2006) literature review, a sizable minority of adolescents reported more problematic levels of warmth and con- flict with their parents. Specifically, 16% of adolescents reported low levels of warmth in the relationship with their father, which further decreased (but finally increased) over the course of ado- lescence, and around 3% reported a high and stable level of conflict with the mother. Interestingly, no participant reported low levels of warmth from the mother or a high level of conflict with the father during adolescence. It thus seems adolescents may be more likely to experience a conflict-laden relationship with their mother than with their father, whereas a low level of warmth seems to be more likely to occur in the relationship with the father than with the mother. These findings may be explained by considering the particular nature of the mother–adolescent and the father– adolescent relationship. Mothers generally shoulder more direct responsibility in their child’s upbringing even during adolescence, for example, in terms of organizing daily care and activities, using disciplinary measures and supervising schoolwork (Phares, Fields, & Kamboukos, 2009). Because conflicts with parents during ad- olescence mainly revolve around household chores, schoolwork and adolescents’ quest for increased autonomy (Laursen, 1995), it can be speculated that the greater involvement of mothers in these areas might lead to a higher potential for conflict. These conflicts around household chores may increase with age as adolescents spend increasing time outside the family home, which may explain

Table 3 Predictors of Parent–Adolescent Warmth and Conflict Trajectories

Predictor

Mother: Medium-warmth decreasing– increasing versus high-warmth

decreasing–increasing Father: Low-warmth decreasing–

increasing versus high-warmth stable Father: Medium-warmth decreasing– increasing versus high-warmth stable

Estimate (SE) p OR 95% CI

Sex .89 (.27) .001 2.42 [.35, 1.42] .40 (.41) .34 1.49 [.66, 3.35] .32 (.30) .27 1.39 [.77, 2.48] Family stress .20 (.13) .12 1.22 [#.05, .45] .55 (.19) .01 1.73 [1.19, 2.53] .08 (.18) .64 1.09 [.77, 1.54] Mother depression .14 (.13) .28 1.15 [#.11, .39] .14 (.20) .49 1.15 [.78, 1.69] .14 (.15) .36 1.15 [.85, 1.55] Father depression #.20 (.13) .11 .82 [#.45, .05] #.11 (.19) .57 .90 [.62, 1.31] #.22 (.17) .21 .81 [.58, 1.13] Genetic risk .13 (.14) .36 1.13 [#.14, .39] .41 (.21) .06 1.50 [.99, 2.27] .30 (.19) .11 1.35 [.93, 1.95] Depression .77 (.14) .00 2.16 [.49, 1.05] 1.37 (.27) .00 3.94 [2.34, 6.64] .98 (.25) .00 2.67 [1.63, 4.37] Indirect effect b ! .31, 95% bootstrap CI [.19, .46] b ! .57, 95% bootstrap CI [.35, .86] b ! .41, 95% bootstrap CI [.21, .69] Total effect b ! .44, 95% bootstrap CI [.19, .68] b ! .97, 95% bootstrap CI [.52, 1.47] b ! .70, 95% bootstrap CI [.32, 1.20]

Mother: High-conflict stable versus low-conflict stable

Mother: Medium-conflict increasing versus low-conflict stable

Father: Medium-conflict increasing- decreasing versus low-conflict

increasing

Estimate (SE) p OR 95% CI

Sex .48 (.79) .54 1.69 [.35, 7.58] #1.37 (.80) .09 .25 [.05, 1.21] #.34 (.30) .26 .71 [.39, 1.29] Family stress .52 (.35) .14 1.69 [.85, 3.37] #.05 (.32) .88 .95 [.51, 1.78] .02 (.16) .91 1.02 [.75, 1.38] Mother depression #.08 (.38) .83 .92 [.44, 1.93] .41 (.30) .17 1.51 [.84, 2.72] .36 (.16) .02 1.43 [1.05, 1.94] Father depression var ! 0 var ! 0 .22 (.13) .11 1.24 [.95, 1.61] Genetic risk .41 (.53) .44 1.51 [.53, 4.26] #.12 (.31) .70 .89 [.49, 1.62] #.01 (.15) .93 .10 [.74, 1.33] Depression 1.22 (.37) .001 3.37 [1.63, 6.95] 1.04 (.22) .00 2.82 [1.82, 4.38] .86 (.18) .00 2.35 [1.64, 3.37] Indirect effect b ! .50, 95% bootstrap CI [.09, .95] b ! .43, 95% bootstrap CI [.21, .69] b ! .35, 95% bootstrap CI [.21, .54] Total effect b ! .91, 95% bootstrap CI [.04, 7.25] b ! .31, 95% bootstrap CI [#.49, .95] b ! .34, 95% bootstrap CI [.07, .61]

Note. Sex is coded such that 0 indicates girls and 1 indicates boys. SE ! standard error; CI ! confidence interval; OR ! odds ratio; var ! 0 ! Father depression needed to be removed from the models because none of the fathers scored for depression in the Mother high-conflict trajectory group.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

11TRAJECTORIES OF PARENT-ADOLESCENT RELATIONSHIP QUALITY

why—at least for some youths— conflict with the mother showed a linear increase with age. In contrast, although fathers typically maintain their involvement in organizing family leisure activities throughout childhood and adolescence, they gradually spend less time with their adolescent offspring (Lewis & Lamb, 2003). Al- though this may potentially lead to a lower level of perceived warmth, it may also limit opportunities for potential conflict with the father. Together, these results highlight the importance of distinguishing between the relationship with the mother and the father to gain a better understanding of how the parent– child relationship evolves during adolescence.

It is noteworthy that the lowest level of warmth and highest level of conflict in the more problematic relationship trajectories (except conflict with the mother) seemed to occur in mid- adolescence in Grade 9 (i.e., at around age 15 years). This is in line with a previous study showing that adolescents perceived less involvement and positive regard from—and more conflict with— their parents at age 14 than at age 11 years (McGue et al., 2005). Our results indicate, however, that youths’ relationships with their parents generally improve again toward the end of adolescence. Interestingly, empirical evidence suggests that a similar curvilinear trend, peaking at around age 15, can be observed with respect to many adolescents’ depression levels (Costello, Swendsen, Rose, & Dierker, 2008; Dekker et al., 2007). As an explanation, Dekker and colleagues (2007) propose that older adolescents may be less exposed to conflict with or detachment from parents due to youths’ increased quest for autonomy, factors that are predominant in mid-adolescence. The findings of the present study suggest that adolescents’ depression symptoms may also, in turn, affect the quality of the parent–adolescent relationship. Indeed, our analyses show that adolescents’ expression of a genetic vulnerability for depressive symptoms significantly contributes to predicting the developmental course of the relationships with their mothers and fathers. Mediated by adolescents’ actual depressive behavior symptoms, a higher genetic risk for depression was associated with a risk of following a Medium-Warmth Decreasing-Increasing tra- jectory with the mother, and a Low-Warmth or a Medium-Warmth Decreasing-Increasing trajectories with the father, compared with more favorable trajectories. A higher genetic risk for depression was also indirectly associated—via adolescents’ actual depressive behavior symptoms—with an increased risk of following a High- Conflict stable trajectory or a Medium-Conflict Increasing trajec- tory with the mother, and a Medium-Conflict Increasing- Decreasing trajectory with the father, compared with more favorable trajectories. These results concord with previous find- ings showing that adolescents’ depressive symptoms predict a decrease in parent–adolescent relationship quality (Branje et al., 2010). Our results are also in line with Coyne’s (1976) interper- sonal theory of depression and Hammen’s (2006) stress generation hypothesis of depression, as they underscore the role of depresso- genic behaviors in the generation of relational stress. Because of the genetically informed design of the present study, this associ- ation might be interpreted as an evocative rGE process. By dis- playing depression-related behaviors, such as self-doubt, irritabil- ity, and negativity, genetically at-risk adolescents may prompt parents to gradually weaken their emotional availability and warmth. Parents’ rejecting responses might, in turn, reinforce adolescents’ insecurities and exacerbate negative interactions within the family. However, the fact that no low-warmth trajectory

was found for the mother–adolescent relationship (but only in the father–adolescent-relationship) suggests that mothers stay engaged and supportive even when children display aversive behaviors. Fathers have been found to respond more passively to children’s depressive symptoms (Sheeber & Sorensen, 1998). They might thus be more prone to experiencing an emotional erosion process when feeling incapable to meet their child’s needs, leading to lower availability to and support of the child. Conversely, no trajectories of problematic levels of conflict with the father were found in the present sample, even for adolescents with a high genetic vulnerability for depressive symptoms. Mothers’ tendency to react more actively to their child’s difficulties (Sheeber & Sorensen, 1998), and their greater involvement in children’s up- bringing (Phares et al., 2009) may explain why adolescents’ depression-related behaviors (e.g., irritability, withdrawal) may trigger conflict particularly in the relationships with their mothers.

Our findings further indicate that adolescents’ depression symp- toms may affect girls’ and boys’ relationship with their parents in the same way. This lack of sex moderation concords with studies reporting a similar (concurrent) link between the parent– adolescent relationship quality and depressive symptoms for boys and girls (Eberhart et al., 2006; Sheeber et al., 2007). Moreover, although the bivariate correlations suggested that girls experienced higher level of conflicts with both parents than boys throughout adolescence, these sex differences disappeared when genetic risk for depressive symptoms and actual depressive behavior were included as predictors of the parent– child conflict trajectories. The initially observed sex differences in parent–adolescent relationship quality thus seemed to be, for the most part, explained by sex differences in depressive symptoms during adolescence. Indeed, adolescent girls also reported significantly higher levels of depres- sion symptoms than boys in the present study. This finding is in line with research showing that the prevalence of clinical depres- sion is about twice as high for adolescent girls, varying between 4% and 23% between the ages of 15 and 18 years, compared with 1% and 11% for boys at the same age (Hankin et al., 1998). The only sex difference observed in the present study was that of girls being less likely to follow the warmer relationship trajectory with their mothers compared with boys. This finding may be explained, at least in part, by the development of girls’ social relations during adolescence. Although friendships become increasingly important for both boys and girls over the years (Collins & Steinberg, 2006), the specific nature of adolescent girls’ friendships could explain their lower perception of closeness with their mothers. As de- scribed by Rose and Rudolph (2006) in their review of sex differ- ences in peer relationship processes, girls’ friendships are charac- terized by higher levels of self-disclosure, engagement, empathy and cooperation than boys’ friendships. Girls’ greater tendency to engage in very close friendship interactions may thus lead them to increasingly turn to friends than to their mothers for emotional support, resulting in a relatively lower perception of warmth in the relationship with the mother during adolescence.

Strengths, Limitations, and Conclusion

The present study has several strengths. Specifically, the use of a genetically informed design based on twins allowed a better test of child-driven effects on the parent–adolescent relationship than is afforded in studies based on single births (Moffitt, 2005).

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

12 BROUILLARD, BRENDGEN, VITARO, DIONNE, AND BOIVIN

Moreover, the separate examination of mother–adolescent and father–adolescent relationships in the present study provided a nuanced portrait according to the sex of the parent.

Our study also presents some limitations. First, the use of adolescents’ reports to assess the parent–adolescent relationship quality may have inflated the reported associations. Previous stud- ies have indeed shown that adolescents often overestimate the level of conflict and minimize the level of support in the relation- ship with their parents (Ehrmantrout, Allen, Leve, Davis, & Shee- ber, 2011). Conversely, however, parents often tend to avoid revealing more problematic aspects of the relations with their children (Aquilino, 1999). It has also been shown that adolescents’ perceptions are a better predictor of their further adjustment (Shoval et al., 2013), favoring the use of adolescents’ reports. Nonetheless, combining parents’ and adolescents’ reports in future studies may best capture the reality of families’ relationships. Second, the distribution of depressive symptoms in our longitudi- nal, nonrisk population-based sample was considerably skewed. Indeed, although some adolescents indicated elevated levels of depressive symptomatology, most reported no depressive symp- toms. The low prevalence of depressive symptoms may have limited the variability of the depressive symptoms measure, po- tentially resulting in an underestimation of the true association with the quality of the parent–adolescent relationship. Future stud- ies should thus include a greater number of youths from high-risk backgrounds and with clinical levels of depression to examine whether the present findings generalize beyond relatively well- adjusted populations. Third, the genetically informed design was based on a rough approximation of overall genetic risk for depres- sive symptoms using an ordinal scale, which is not comparable to information of individuals’ genetic risk based on a specific geno- type. However, this method has been used in previous studies to show evidence of gene-environment interplay (Jaffee et al., 2005; Wichers et al., 2009). In addition, the inclusion of parental depres- sion as a control variable lends further support to the notion that the observed rGEs may be interpreted in terms of evocative (i.e., child driven), rather than passive (i.e., parent-driven) processes. Still, the present findings should be replicated in future studies with larger samples using alternative operationalizations of genetic risk (e.g., either using biometric modeling or genotypic data).

Despite these limitations, the present study offers new insights into individual and normative changes within the parent– adolescent relationship throughout adolescence and the role of child-driven effects in the quality of those relationships. Aware- ness of the interactional processes at play in depression could sensitize parents to stay supportive of their child and thus possibly avoid a deterioration of the parent– child relationship. Adolescents could also benefit from a clearer understanding of the impact of depression-related behavior on others and be more informed on how to seek support and resolve conflicts efficiently. This may enable adolescents at risk for depression symptoms to take an active ownership in their mental health and to become better equipped on their way to adulthood.

References

Andrieu, N., & Goldstein, A. M. (1998). Epidemiologic and genetic ap- proaches in the study of gene-environment interaction: An overview of

available methods. Epidemiologic Reviews, 20, 137–147. http://dx.doi .org/10.1093/oxfordjournals.epirev.a017976

Aquilino, W. S. (1999). Two views of one relationship: Comparing par- ents’ and young adult children’s reports of the quality of intergenera- tional relations. Journal of Marriage and the Family, 61, 858 – 870. http://dx.doi.org/10.2307/354008

Asparouhov, T., & Muthén, B. (2014). Auxiliary variables in mixture modeling: Three-step approaches using M plus. Structural Equation Modeling, 21, 329 –341. http://dx.doi.org/10.1080/10705511.2014 .915181

Blos, P. (1967). The second individuation process of adolescence. The Psychoanalytic Study of the Child, 22, 162–186. http://dx.doi.org/10 .1080/00797308.1967.11822595

Boivin, M., Petitclerc, A., Feng, B., & Barker, E. D. (2010). The devel- opmental trajectories of peer victimization in middle to late childhood and the changing nature of their behavioral correlates. Merrill-Palmer Quarterly, 56, 231–260. http://dx.doi.org/10.1353/mpq.0.0050

Branje, S. J., Hale, W. W., III, Frijns, T., & Meeus, W. H. (2010). Longitudinal associations between perceived parent-child relationship quality and depressive symptoms in adolescence. Journal of Abnormal Child Psychology, 38, 751–763. http://dx.doi.org/10.1007/s10802-010- 9401-6

Brendgen, M., Vitaro, F., Bukowski, W. M., Dionne, G., Tremblay, R. E., & Boivin, M. (2013). Can friends protect genetically vulnerable children from depression? Development and Psychopathology, 25, 277–289. http://dx.doi.org/10.1017/S0954579412001058

Brendgen, M., Vitaro, F., Turgeon, L., & Poulin, F. (2002). Assessing aggressive and depressed children’s social relations with classmates and friends: A matter of perspective. Journal of Abnormal Child Psychology, 30, 609 – 624. http://dx.doi.org/10.1023/A:1020863730902

Collins, W. A., & Steinberg, L. (2006). Adolescent development in inter- personal context. In N. Eisenberg, W. Damon, & L. M. Richard (Eds.), Handbook of child psychology (6th ed., Vol. 3, pp. 1003–1067). Hobo- ken, NJ: Wiley.

Costello, D. M., Swendsen, J., Rose, J. S., & Dierker, L. C. (2008). Risk and protective factors associated with trajectories of depressed mood from adolescence to early adulthood. Journal of Consulting and Clinical Psychology, 76, 173–183.

Coyne, J. C. (1976). Toward an interactional description of depression. Psychiatry: Journal for the Study of Interpersonal Processes, 39, 28 – 40. http://dx.doi.org/10.1080/00332747.1976.11023874

Dekker, M. C., Ferdinand, R. F., van Lang, N. D., Bongers, I. L., van der Ende, J., & Verhulst, F. C. (2007). Developmental trajectories of de- pressive symptoms from early childhood to late adolescence: Gender differences and adult outcome. Journal of Child Psychology and Psy- chiatry, 48, 657– 666. http://dx.doi.org/10.1111/j.1469-7610.2007 .01742.x

Eberhart, N. K., Auerbach, R. P., Bigda-Peyton, J., & Abela, J. R. (2011). Maladaptive schemas and depression: Tests of stress generation and diathesis-stress models. Journal of Social and Clinical Psychology, 30, 75–104. http://dx.doi.org/10.1521/jscp.2011.30.1.75

Eberhart, N. K., Shih, J. H., Hammen, C. L., & Brennan, P. A. (2006). Understanding the sex difference in vulnerability to adolescent depres- sion: An examination of child and parent characteristics. Journal of Abnormal Child Psychology, 34, 493–508. http://dx.doi.org/10.1007/ s10802-006-9020-4

Ehrmantrout, N., Allen, N. B., Leve, C., Davis, B., & Sheeber, L. (2011). Adolescent recognition of parental affect: Influence of depressive symp- toms. Journal of Abnormal Psychology, 120, 628 – 634. http://dx.doi.org/ 10.1037/a0022500

Flynn, M., & Rudolph, K. D. (2011). Stress generation and adolescent depression: Contribution of interpersonal stress responses. Journal of Abnormal Child Psychology, 39, 1187–1198. http://dx.doi.org/10.1007/ s10802-011-9527-1

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

13TRAJECTORIES OF PARENT-ADOLESCENT RELATIONSHIP QUALITY

Furman, W., & Buhrmester, D. (1985). Children’s perceptions of the personal relationships in their social networks. Developmental Psychol- ogy, 21, 1016 –1024. http://dx.doi.org/10.1037/0012-1649.21.6.1016

Furman, W., & Buhrmester, D. (1992). Age and sex differences in per- ceptions of networks of personal relationships. Child Development, 63, 103–115. http://dx.doi.org/10.2307/1130905

Gibb, B. E., & Hanley, A. J. (2010). Depression and interpersonal stress generation in children: Prospective impact on relational versus overt victimization. International Journal of Cognitive Therapy, 3, 358 –367. http://dx.doi.org/10.1521/ijct.2010.3.4.358

Goldsmith, H. (1991). A zygosity questionnaire for young twins: A re- search note. Behavior Genetics, 21, 257–269. http://dx.doi.org/10.1007/ BF01065819

Guimond, F. A., Brendgen, M., Vitaro, F., Forget-Dubois, N., Dionne, G., Tremblay, R. E., & Boivin, M. (2014). Gene-environment interplay in the link of friends’ and nonfriends’ behaviors with children’s social reticence in a competitive situation. Developmental Psychology, 50, 956 –967. http://dx.doi.org/10.1037/a0034354

Hammen, C. (2006). Stress generation in depression: Reflections on ori- gins, research, and future directions. Journal of Clinical Psychology, 62, 1065–1082. http://dx.doi.org/10.1002/jclp.20293

Hammen, C. L., & Peters, S. D. (1977). Differential responses to male and female depressive reactions. Journal of Consulting and Clinical Psy- chology, 45, 994 –1001. http://dx.doi.org/10.1037/0022-006X.45.6.994

Hankin, B., Abramson, L., Moffitt, T., Silva, P., McGee, R., & Angell, K. (1998). Development of depression from preadolescence to young adult- hood: Emerging gender differences in a 10-year longitudinal study. Journal of Abnormal Psychology, 107, 128 –140. http://dx.doi.org/10 .1037/0021843X.107.1.128

Hicks, B. M., DiRago, A. C., Iacono, W. G., & McGue, M. (2009). Gene-environment interplay in internalizing disorders: Consistent find- ings across six environmental risk factors. Journal of Child Psychology and Psychiatry, 50, 1309 –1317. http://dx.doi.org/10.1111/j.1469-7610 .2009.02100.x

Jaffee, S. R., Caspi, A., Moffitt, T. E., Dodge, K. A., Rutter, M., Taylor, A., & Tully, L. A. (2005). Nature % nurture: Genetic vulnerabilities interact with physical maltreatment to promote conduct problems. Develop- ment and Psychopathology, 17, 67– 84. http://dx.doi.org/10.1017/ S0954579405050042

Jaffee, S. R., & Price, T. S. (2007). Gene-environment correlations: A review of the evidence and implications for prevention of mental illness. Molecular Psychiatry, 12, 432– 442. http://dx.doi.org/10.1038/sj.mp .4001950

Joiner, T. E., Jr., Alfano, M. S., & Metalsky, G. I. (1992). When depression breeds contempt: Reassurance seeking, self-esteem, and rejection of depressed college students by their roommates. Journal of Abnormal Psychology, 101, 165–173. http://dx.doi.org/10.1037/0021-843X.101.1 .165

Kim, J., Thompson, E. A., Walsh, E. M., & Schepp, K. G. (2015). Trajectories of parent-adolescent relationship quality among at-risk youth: Parental depression and adolescent developmental outcomes. Archives of Psychiatric Nursing, 29, 434 – 440. http://dx.doi.org/10 .1016/j.apnu.2015.07.001

Klahr, A. M., & Burt, S. A. (2014). Elucidating the etiology of individual differences in parenting: A meta-analysis of behavioral genetic research. Psychological Bulletin, 140, 544 –586. http://dx.doi.org/10.1037/ a0034205

Kovacs, M. (1992). Children’s Depression Inventory (CDI) manual. North Tonawanda, NY: Multi Health Systems.

Kuczynski, L. (Ed.). (2002). Handbook of dynamics in parent-child rela- tions. Thousand Oaks, CA: Sage.

Lau, J. Y. F., & Eley, T. C. (2008). New behavioral genetic approaches to

depression in childhood and adolescence. In J. R. Z. Abela & B. L. Hankin (Eds.), Handbook of depression in children and adolescents (pp. 124 –148). New York, NY: Guilford Press.

Laursen, B. (1995). Conflict and social interaction in adolescent relation- ships. Journal of Research on Adolescence, 5, 55–70. http://dx.doi.org/ 10.1207/s15327795jra0501_3

Laursen, B., & Collins, W. A. (2009). Parent– child relationships during adolescence. In R. M. Lerner & L. Steinberg (Eds.), Handbook of adolescent psychology (pp. 3– 42). Hoboken, NJ: Wiley. http://dx.doi .org/10.1002/9780470479193.adlpsy002002

Laursen, B., DeLay, D., & Adams, R. E. (2010). Trajectories of perceived support in mother-adolescent relationships: The poor (quality) get poor- er. Developmental Psychology, 46, 1792–1798. http://dx.doi.org/10 .1037/a0020679

Lewis, C., & Lamb, M. E. (2003). Fathers’ influences on children’s development: The evidence from two-parent families. European Journal of Psychology of Education, 18, 211–228. http://dx.doi.org/10.1007/ BF03173485

Liu, R. T., & Alloy, L. B. (2010). Stress generation in depression: A systematic review of the empirical literature and recommendations for future study. Clinical Psychology Review, 30, 582–593. http://dx.doi .org/10.1016/j.cpr.2010.04.010

Magnusson, P. K. E., Almqvist, C., Rahman, I., Ganna, A., Viktorin, A., Walum, H., . . . Lichtenstein, P. (2013). The Swedish Twin Registry: Establishment of a biobank and other recent developments. Twin Re- search and Human Genetics, 16, 317–329. http://dx.doi.org/10.1017/thg .2012.104

McGue, M., Elkins, I., Walden, B., & Iacono, W. G. (2005). Perceptions of the parent-adolescent relationship: A longitudinal investigation. Devel- opmental Psychology, 41, 971–984. http://dx.doi.org/10.1037/0012- 1649.41.6.971

Moffitt, T. E. (2005). The new look of behavioral genetics in developmen- tal psychopathology: Gene-environment interplay in antisocial behav- iors. Psychological Bulletin, 131, 533–554. http://dx.doi.org/10.1037/ 0033-2909.131.4.533

Muthén, L. K., & Muthén, B. O. (1998 –2017). Mplus user’s guide (Ver- sion 8). Los Angeles, CA: Author.

Nagin, D. S., & Odgers, C. L. (2010). Group-based trajectory modeling in clinical research. Annual Review of Clinical Psychology, 6, 109 –138. http://dx.doi.org/10.1146/annurev.clinpsy.121208.131413

Petrill, S. A. (2003). Examining social behavior and relationships using genetically sensitive designs: An introduction. Marriage & Family Re- view, 33, 3–10. http://dx.doi.org/10.1300/J002v33n01_02

Phares, V., Fields, S., & Kamboukos, D. (2009). Fathers’ and mothers’ involvement with their adolescents. Journal of Child and Family Stud- ies, 18, 1–9. http://dx.doi.org/10.1007/s10826-008-9200-7

Poirier, C. S., Brendgen, M., Girard, A., Vitaro, F., Dionne, G., & Boivin, M. (2016). Friendship experiences and anxiety among children: A ge- netically informed study. Journal of Clinical Child & Adolescent Psy- chology, 45, 655– 667. http://dx.doi.org/10.1080/15374416.2014 .987382

Rose, A. J., & Rudolph, K. D. (2006). A review of sex differences in peer relationship processes: Potential trade-offs for the emotional and behav- ioral development of girls and boys. Psychological Bulletin, 132, 98 – 131. http://dx.doi.org/10.1037/0033-2909.132.1.98

Saint-Laurent, L. (1990). Étude psychométrique de l’Inventaire de dépres- sion pour enfants de Kovacs auprès d’un échantillon francophone [Psy- chometric study of Kovacs’s Children’s Depression Inventory with a French-speaking sample]. Canadian Journal of Behavioural Science/ Revue Canadienne des Sciences du Comportement, 22, 377. http://dx .doi.org/10.1037/h0078990

Saluja, G., Iachan, R., Scheidt, P. C., Overpeck, M. D., Sun, W., & Giedd, J. N. (2004). Prevalence of and risk factors for depressive symptoms

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

14 BROUILLARD, BRENDGEN, VITARO, DIONNE, AND BOIVIN

among young adolescents. Archives of Pediatrics & Adolescent Med- icine, 158, 760 –765. http://dx.doi.org/10.1001/archpedi.158.8.760

Santé Québec., Jetté, M., Desrosiers, H., & et Tremblay, R. E. (1998). “In 2001 . . . I’ll be 5 years old!” Survey of 5-month old infants. Preliminary report of the Longitudinal Study of Child Development in Québec. Québec, Canada: Bibliothèque Nationale du Québec.

Scarr, S., & McCartney, K. (1983). How people make their own environ- ments: A theory of genotype ¡ environment effects. Child Develop- ment, 54, 424 – 435.

Segrin, C., & Dillard, J. P. (1992). The interactional theory of depression: A meta-analysis of the research literature. Journal of Social and Clinical Psychology, 11, 43–70. http://dx.doi.org/10.1521/jscp.1992.11.1.43

Seiffge-Krenke, I., Overbeek, G., & Vermulst, A. (2010). Parent-child relationship trajectories during adolescence: Longitudinal associations with romantic outcomes in emerging adulthood. Journal of Adolescence, 33, 159 –171. http://dx.doi.org/10.1016/j.adolescence.2009.04.001

Shanahan, L., McHale, S. M., Crouter, A. C., & Osgood, D. W. (2007). Warmth with mothers and fathers from middle childhood to late ado- lescence: Within- and between-families comparisons. Developmental Psychology, 43, 551–563. http://dx.doi.org/10.1037/0012-1649.43.3.551

Sheeber, L. B., Davis, B., Leve, C., Hops, H., & Tildesley, E. (2007). Adolescents’ relationships with their mothers and fathers: Associations with depressive disorder and subdiagnostic symptomatology. Journal of Abnormal Psychology, 116, 144 –154. http://dx.doi.org/10.1037/0021- 843X.116.1.144

Sheeber, L., & Sorensen, E. (1998). Family relationships of depressed adolescents: A multimethod assessment. Journal of Clinical Child Psy- chology, 27, 268 –277. http://dx.doi.org/10.1207/s15374424jccp2703_4

Shoval, G., Mansbach-Kleinfeld, I., Farbstein, I., Kanaaneh, R., Lubin, G., Apter, A., . . . Zalsman, G. (2013). Self versus maternal reports of emotional and behavioral difficulties in suicidal and non-suicidal ado- lescents: An Israeli nationwide survey. European Psychiatry, 28, 235– 239. http://dx.doi.org/10.1016/j.eurpsy.2012.02.009

Smetana, J. G., Campione-Barr, N., & Metzger, A. (2006). Adolescent development in interpersonal and societal contexts. Annual Review of Psychology, 57, 255–284. http://dx.doi.org/10.1146/annurev.psych.57 .102904.190124

Smollar, J., & Youniss, J. (1989). Transformations in adolescents’ percep- tions of parents. International Journal of Behavioral Development, 12, 71– 84.

Smucker, M. R., Craighead, W. E., Craighead, L. W., & Green, B. J. (1986). Normative and reliability data for the Children’s Depression Inventory. Journal of Abnormal Child Psychology, 14, 239. http://dx.doi .org/10.1007/BF00917219

Spanier, G. B. (1976). Measuring dyadic adjustment: New scales for assessing the quality of marriage and similar dyads. Journal of Marriage and the Family, 38, 15–28. http://dx.doi.org/10.2307/350547

Spitz, E., Carlier, M., Vacher-Lavenu, M. C., & Reed, T. (1996). Long- term effect of prenatal heterogeneity among monozygotes. Cahiers de Psychologie Cognitive/Current Psychology of Cognition, 15, 283–308.

Starrels, M. E. (1994). Gender differences in parent-child relations. Jour- nal of Family Issues, 15, 148 –165. http://dx.doi.org/10.1177/ 019251394015001007

Vallerand, R. J. (1989). Vers une methodologie de validation transcul- turelle de questionnaires psychologiques: Implications pour la recherche en langue francaise [Toward a methodology for the transcultural vali- dation of psychological questionnaires: Implications for research in the French language]. Canadian Psychology/Psychologie canadienne, 30, 662– 680. http://dx.doi.org/10.1037/h0079856

Vermunt, J. K. (2010). Latent class modeling with covariates: Two im- proved three-step approaches. Political Analysis, 18, 450 – 469. http:// dx.doi.org/10.1093/pan/mpq025

Wichers, M., Schrijvers, D., Geschwind, N., Jacob, N., Myin-Germeys, I., Thiery, E., . . . van Os, J. (2009). Mechanisms of gene- environment interactions in depression: Evidence that genes poten- tiate multiple sources of adversity. Psychological Medicine, 39, 1077– 1086.

Wilkinson, P. O., Trzaskowski, M., Haworth, C. M., & Eley, T. C. (2013). The role of gene-environment correlations and interactions in middle childhood depressive symptoms. Development and Psycho- pathology, 25, 93–104. http://dx.doi.org/10.1017/S09545794 12000922

Received June 30, 2018 Revision received April 2, 2019

Accepted April 3, 2019 !

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

15TRAJECTORIES OF PARENT-ADOLESCENT RELATIONSHIP QUALITY