Journal Article Summary

The assignment in the folder below.

write 2 to 3 pages

I uploaded chapter 1, 2, 3, 5, 6, 7, and 8 (only the ones we took so far)

and I uploaded 5 articles and you just have to choose one for this assignment

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: Brendgen.Mara@uqam.ca

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

y.

Developmental Psychology © 2019 American Psychological Association 2019, Vol. 1, No. 999, 000 0012-1649/19/$12.00 http://dx.doi.org/10.1037/dev0000751

1

 

 

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

y.

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

y.

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

y.

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

y.

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.