Deviant Peer Affiliation and Problem Behavior: A Test of Genetic and Environmental Influences

  • Bernadette Marie Bullock
  • Kirby Deater-Deckard
  • Leslie D. Leve
Article

This study uses a multitrait, multimethod (MTMM) approach to investigate the genetic and environmental etiologies of childhood deviant peer affiliation (DPA) and problem behavior (PROB). The variability of genetic and environmental estimates by agent and method is also examined. A total of 77 monozygotic and 72 dizygotic twin pairs and each twin's close friend were assessed. The informants included parents, teachers, and twins, and the methods involved questionnaire reports and coder ratings of videotaped dyadic interactions between each twin and their close friend. Twin intraclass correlations and univariate models attributed DPA and PROB to genetic, and shared and nonshared environmental effects. Parameter estimates differed by rater and method, however. Results accentuate the imperative to attend to method effects inherent in MTMM behavioral geneticresearch.

KEY WORDS:

behavior genetic deviant peer affiliation problem behavior children methodology 

 

Peers are important agents of socialization, influencing many domains of psychosocial and cognitive development (Hartup, 1996; Hartup & Stevens, 1997; Parker & Asher, 1997). In adolescence, affiliation with antisocial friends is one of the strongest correlates of delinquency and substance use (Elliott, Huizinga, & Ageton, 1985; Loeber, Stouthamer-Loeber, Van Kammen, & Farrington, 1991; Patterson, Reid, & Dishion, 1992). Furthermore, interpersonal dynamics between antisocial friends are linked to an escalation in deviant behaviors (Dishion, Andrews, Kavanagh, & Soberman, 1996). In particular, observations of positive reinforcement of deviant talk within peer dyads (e.g., discussing aggressive or illegal activities) are linked to escalations in substance use, delinquency, violence, and risky sexual behavior in adolescence and early adulthood (Capaldi, Dishion, Stoolmiller, & Yoerger, 2001; Dishion et al., 1996; Dishion, Capaldi, Spracklen, & Li, 1995;Dishion, Eddy, Haas, Li, & Spracklen, 1997). Although environmental mechanisms are indeed influential, they co-occur within the context of genetic influences (Plomin, 1994). The first aim of the current study was to test whether and how genetic and environmental factors operate in accounting for child deviant peer affiliation and problem behavior.

Similar to psychosocial research, behavioral genetic studies are subject to methodological problems including rater bias and measurement error (O'Connor & Plomin, 2000). For behavioral genetic research, estimates of genetic, and shared and nonshared environmental effects can vary significantly by rater and method (Eaves et al., 1997; Leve, Winebarger, Fagot, Reid, & Goldsmith, 1998; Miles & Carey, 1997), leading some to rely on multimethod latent indicators in their models. Unfortunately, combining data across informants and methods may diffuse distinct patterns of behavior that emerge by context (Bank, Dishion, Skinner, & Patterson, 1990). The second aim of this research was to examine estimates of genetic, and shared and nonshared environmental variance using a multitrait, multimethod (MTMM) assessment strategy to further the understanding of how rater and method effects might vary when considering child deviant peer affiliation and problem behavior.

Family Influences on Deviant Peer Affiliation and Problem Behavior

Deviant peer affiliation and problem behaviors are linked to parenting practices, dynamics embedded within friendship interactions, constitutionally based individual differences including genetic influences, and other factors. The “Nurturance Hypothesis” (Dishion & Bullock, 2002) proposes that adult caregivers influence child problem behavior directly via parenting strategies and indirectly through parents' influence on children's exposure to individuals outside the home, including peers. Migration into a deviant peer group and antisocial child and adolescent behavior can be fostered by premature adolescent autonomy and fueled by parental disengagement and abdication of monitoring and limit setting (Dishion, Nelson, & Bullock, 2004). Although deviant peer affiliation is often conceptualized as a group activity in which youth share similar environments and behaviors, it is possible that parents may also transmit genes that predispose their children to interact with deviant peers. This proposition has yet to be investigated; however, several studies suggest that aggressive and delinquent acts that often serve as organizing behaviors in deviant peer groups are influenced by genetic as well as environmental factors (see recent meta-analyses by Miles & Carey, 1997; Rhee & Waldman, 2002).

Problem behaviors can also be shaped by relational dynamics and contingencies within families, including escape conditioning and positive reinforcement of child negative behavior, which are often referred to as coercive family processes (Patterson, 1982). Once these behavioral dynamics are instantiated, children and adolescents may seek the companionship of, and have greater access to, similar others who foster deviance (Dishion et al., 2004; Loeber & Dishion, 1983; Quay & Peterson, 1987).

In addition to factors within the family environment, constitutionally based individual differences, including genetic influences on aggression, delinquency, and substance use (Miles & Carey, 1997; Rhee & Waldman, 2002) among other factors, can provide an impetus toward antisocial behavior. Recent meta-analyses provide evidence that heritability accounts for roughly 50% of the variance in aggression and multiple forms of antisocial behavior (Miles & Carey, 1997; Rhee & Waldman, 2002). Genetic studies also indicate that shared (i.e., leading to sibling similarity) and nonshared (i.e., leading to sibling differentiation) environmental influences contribute significantly to the etiology of antisocial behaviors (Deater-Deckard & Plomin, 1999; Lyons et al., 1995).

There are few genetic studies of peer affiliation and child and early adolescent socialization, and even fewer that use an MTMM approach. Extant research has emphasized child and adolescent perceptions of peer behavior, finding small-to-moderate genetic effects (Baker & Daniels, 1990; Daniels & Plomin, 1985; Iervolino et al., 2002; Manke, McGuire, Reiss, Hetherington, & Plomin, 1995; Pike, Manke, Reiss, & Plomin, 2000), suggesting that some of the variance in the selection of friends may be influenced by heritable child characteristics in addition to environmental factors. Nevertheless, there is still much to understand regarding the extent to which constitutionally based genetic factors affect friendship selection and promote antisocial behavior, or whether these influences operate via shared and/or nonshared environmental pathways.

MTMM Assessment Strategy and Related Methodological Issues

A unique feature of this study is the use of an MTMM design in combination with a genetically informative sample. MTMM assessments include parent, teacher, and child questionnaire data and coder ratings of videotaped dyadic interactions for each twin and a friend during a semistructured observation task. Most studies of adolescents' relationship processes incorporate an observational component (Dishion et al., 1995, 1996, 1997). However, with the exception of the Nonshared Environment and Adolescent Development study (NEAD; Reiss, Neiderhiser, Hetherington, & Plomin, 2000), we are not aware of other genetically informed studies of adolescents that incorporate observational data.

When considering the full spectrum of development from childhood through adolescence, there are few published genetic observational studies of problem behavior, and most of these provide estimates of heritability that are typically lower for observations, compared with parent or self-reports (Borkenau, Riemann, Angleitner, & Spinath, 2000; Deater-Deckard, 2000; Deater-Deckard et al., 2001; Ge et al., 1996; Hoffman, 1991; Leve et al., 1998; Reiss et al., 2000). These findings, however, are not always consistent (Miles & Carey, 1997). Indeed, there is a need to assess the utility of MTMM approaches for behavior genetic research. There is still much to be learned regarding the prevalence of rater effects for estimates of genetic and environmental influences (Dishion & Bullock, 2002; Miles & Carey, 1997; O’Connor & Plomin, 2000; Patterson et al., 1992).

Overview of Study Questions

Two questions are emphasized in this study: (1) Is there evidence of genetic and/or environmental variance in child deviant peer affiliation and problem behavior? and (2) To what extent does the method of assessment affect the genetic and environmental parameter estimates for measures of problem behavior and deviant peer affiliation when an MTMM strategy is employed?

METHOD

Participants

Data from twin pairs and a close friend of each from the second wave of the Oregon Twin Project (OTP) were used. During the first wave, families with twins were identified via birth announcements, twin organizations, and the public school system between 1993 and 1994. Families were contacted (50% by telephone, 50% by mail), provided with a description of the study, and invited to participate in the project. All twin families with children in the targeted age range (6–14 years) who responded to the request were included, and participants were paid for their participation. The second wave occurred approximately 18 months later and included 136 of the original 158 families (86%). Of those, 119 agreed to participate (88%). In addition, 30 new twin families were added during wave two. There were no significant demographic differences between families who participated in only one or in both waves.

The final sample for analysis included 149 families; 77 had monozygotic or MZ twins and 72 had either same or opposite sex dizygotic or DZ twins (see Table I). The children were 10 years old on average and were predominantly European American (91%). Zygosity was determined by parent report on the Zygosity Questionnaire (Goldsmith, 1991), which draws on developmental and medical history data. This instrument is widely employed and has 94% accuracy compared with blood typing (Goldsmith, 1991). Three independent raters reviewed the questionnaire and photographs of the twins to determine zygosity. In one case, rater agreement was not established, so data for that pair were not included in the study.
Table I.

Demographic Information by Twin Zygosity

Characteristics

Monozygotic (n = 154 twins)

Dizygotic (n = 144 twins)

Twin mean age, years (SD)

10.18 (1.98)

9.93 (1.85)

Twin age range, years

7–14

6–14

Twin mean grade (SD)

4.68 (1.98)

4.40 (1.79)

Twin pair sex (n = twin pairs)

 Male

33

20

 Female

44

21

 Opposite sex

 

31

Twin ethnicity (n)

 Caucasian

140

138

 African American

0

2

 Asian

6

0

 Native American

4

0

 Other

4

4

Peer mean age (SD)

10.31 (2.03)

10.01 (1.83)

Peer mean grade (SD)

4.75 (2.01)

4.36 (1.73)

Peer sex (n)

 Male

66

71

 Female

88

73

Parent mean age (SD)

 Mother

39.80 (4.96)

38.94 (4.95)

 Father

41.26 (5.68)

41.25 (5.80)

Parent education (n)

 Mother

  High school graduate

2

0

  High school diploma

17

18

  Some college

33

23

  College degree

17

18

  Graduate degree/professional

5

11

 Father

  High school graduate

1

0

  High school diploma

7

11

  Some college

20

19

  College degree

23

18

  Graduate degree/professional

11

13

Socioeconomic status

 Median Hollingshead Index (SD)

5.01 (1.68)

5.44 (1.67)

Family structure (n)

 Married

60

58

 Live-in partner

1

1

 Separated

3

1

 Divorced

9

10

 Single

4

2

Note. SD: standard deviation.

Parents were 39 years old on average (range 25–54 years), with an average of 15 years of education. Parent occupational status ranged from 1 to 9 on the Hollingshead occupation code, with a median of 6 for fathers (i.e., semiprofessionals and small business owners) and a median of 5 for mothers (i.e., clerical and sales workers). Twenty percent of the families were single-parent households. Neither ethnicity nor age of child and parent, parent education, parent occupation, or family structure differed by twin zygosity. The demographic background of the families was comparable to median levels in the Pacific Northwest community from which the sample was obtained.

Procedures

Each child selected a close friend of the same age and sex to participate in the study. The friend was required to be a nonrelative who lived in a different household. Within a family, each twin was instructed to select a different friend than that of their co-twin. In three families, twins selected the same friend or their co-twin, and were asked to select a different individual.

The twin--friend dyads engaged in two 5-min conversations, which took place in the laboratory and were videotaped. The conversations involved (a) planning a fun activity that they would enjoy doing together during the next several weeks and (b) the upcoming school year, including what their school experience would be like, what they might like or dislike about it, and information they had learned about it from other children. The order of participation was counterbalanced for the birth order of the twins. In addition, the twins and their friends completed questionnaires, as did twins' and friends' parents and teachers. In many cases, twins and their friends attended the same school. In this sample, 27% of the twin pairs (24% for MZ and 31% for DZ) and 32% of the twin–peer dyads (32% for MZ and 31% for DZ) were rated by the same teacher. There were no systematic differences in the clustering of teacher ratings by zygosity; this happened equally often for MZ as for DZ twins.

Measures

Deviant Peer Affiliation (DPA)

DPA and child problem behaviors were each assessed using three measures. Twins' friends' deviant behaviors were assessed using two teacher-report measures and observers' ratings of videotaped interactions between each twin and their selected friend. Specifically, teachers completed the teacher-report form of the Child Behavior Checklist (CBCL; Achenbach, 1991b) for the friends who participated in the study. The externalizing score (α = .85) provided a measure of the friends' deviant behavior (T-DP1). Using the Teacher Peers/Social Skills questionnaire (Walker & McConnell, 1988), teachers also responded to questions about the type of children with whom the twin associated. These questions pertained to the peer group in to general, rather than to the specific friend who participated in the study with the twin. Three items were used for this analysis: (1) How often is the peer with children who misbehave? (2) How often is the peer with children who get into fights? and (3) How often is the twin with a peer who exerts a negative influence on friends? The responses were rated from 1 (never) to 5 (always), and were averaged and standardized to yield a second teacher-rated deviant peer affiliation scale (T-DP2; α = .80).

Finally, research assistants who were blind to information regarding the twins or their friends, coded the behavior of each participant (either twin or friend) for both the “fun activity” and “school experience” videotaped discussions. Three items from the Coder Impressions global rating instrument (adapted from Forgatch, Fetrow, & Lathrop, 1984) were used: “How at risk for antisocial behavior is the child?’’ (1: no chance, 3: moderate chance, 5: definite chance); and “How does the child rate on the following bipolar adjectives?: rude vs. polite; uncooperative vs. cooperative,” using the following scale (1: very, 2: somewhat, 3: neutral, 4: somewhat, and 5: very). Research assistants were trained to rate each item using their own perceptions of the children's behavior and were provided examples of antisocial behaviors (e.g., discussion of deviant activities, rude or inappropriate behavior, aggression) from which to base their ratings.

Twin and peer bivariate correlations for each of these three items (antisocial, rude–polite, and uncooperative–cooperative) within and across tasks were all statistically significant (from r = .57 to r = .78 p < .05). To create a deviant friendship process score that reflected both twin and peer influences during the interaction, twin and peer scores for each task were summed. These items were then averaged across tasks and standardized to yield an observer-rated deviant friendship process score (O-DFP; α = .83). Interrater reliability was acceptable (α = .78, percent agreement = 90%). The antisocial, rude–polite, and uncooperative–cooperative items were recoded to ensure that the higher the item score, the more negative the outcome. This deviant friendship process score (O-DFP) was significantly correlated with teacher ratings of the externalizing behavior of each twin's friend (DP-1: r = .17, p < .01) and the extent to which each twin affiliated with antisocial peers (DP-2: r = .14, p < .05).

Problem Behavior

Children's antisocial behaviors were assessed using parents', teachers', and twins' reports. Parents completed the parent form of the CBCL (Achenbach, 1991a), a standardized and validated measure that includes 113 items. The externalizing score (delinquent and aggressive behavior; α = .84) provided the parent ratings of problem behavior (P-PB) for each twin independently. The parent identified as the primary caregiver (the mother in all but three cases) provided these reports. Teachers completed the teacher report form of the CBCL (Achenbach, 1991b), and the externalizing score was used as an indicator of child problem behavior (T-PB; α = .85).

Each twin independently completed Harter's (1993) “What I Am Like” Questionnaire, a 36-item measure that assesses the child's self-perceptions. Each question described two orthogonal behavioral dimensions and asked the respondent to identify “Which kids are most like you?” Twins were required to indicate which of the two contradictory descriptions of child behavior is “sort of true for me” or “really true for me.” For this study, the mean of the following four items was used to create a child self-report problem behavior scale (C-PB): (a) “Some kids usually act the way that they know they are supposed to, but other kids often don't act the way that they are supposed to”; (b) “Some kids usually get in trouble because of things they do, but other kids don't do things that get them into trouble”; (c) “Some kids do things they know they shouldn't do, but other kids hardly ever do things they know they shouldn't do”; and (d) “Some kids behave themselves well, but other kids often find it hard to behave themselves.” These items were scored on a 4-point scale with a higher score indicating a higher level of problem behavior. The items were averaged, and the resulting score was standardized to represent child-rated problem behavior (C-PB; α = .78).

RESULTS

Results are provided in two sections. Descriptive statistics and preliminary analyses are presented first, followed by univariate genetic analyses pertaining to the two study questions.

Descriptive Statistics and Preliminary Analyses

Multivariate analyses of variance (MANOVA) were used to test for mean level differences in DPA and PROB by twin zygosity and in sex composition (MZ and DZ same sex and DZ opposite sex pairs). Analyses revealed significant differences by sex for two of the three DPA scales, with boys being significantly more likely to affiliate with peers who get into trouble (T-DP1) and engage in deviant friendship process (O-DFP) than were girls (see Table II). A significant difference by sex was also detected for twin self-reported problem behavior, with boys endorsing significantly more problem behaviors than did girls. No other statistically significant group differences by sex or zygosity, or sex by zygosity interactions for any of the DPA or PROB scales, were found. These data suggest that, although boys tend to affiliate with peers who are rated as more antisocial, the mean levels of deviant peer affiliation and problem behavior do not vary as a function of zygosity of the twin pairs.
Table II.

Multivariate Analysis of Variance (MANOVA) for Standardized Deviant Peer Association and Problem Behavior Scales by Zygosity and Sex of Twin Pair

 

Mean (SD)

             
 

MZ

DZ

 

Significance

        

Scales

Male

Female

Male

Female

Opposite sex

Sex

Zygosity

Sex × Zygosity

        

Deviant peer association (DPA)

        

  T-DP1a (teacher-reported peer externalizing behavior)

−0.05 (0.99)

−0.29 (0.53)

0.23 (1.26)

−0.08 (0.87)

0.92 (0.79)

*

ns

ns

        

  T-DP2b (teacher-reported peer deviance)

0.26 (1.03)

−0.21 (0.83)

0.32 (1.35)

−0.22 (1.03)

0.20 (0.97)

ns

ns

ns

        

  O-DFPc (coder impressions—deviant friendship process)

0.18 (1.24)

−0.16 (0.68)

0.54 (1.50)

−0.34 (0.62)

0.07 (1.02)

*

ns

ns

        

Problem behavior (PROB)

        

  P-PBa (parent-reported twin externalizing behavior)

−0.16 (0.99)

−0.10 (0.90)

0.04 (1.07)

0.04 (1.00)

0.15 (1.15)

ns

ns

ns

        

  T-PBa (teacher-reported twin externalizing behavior)

0.08 (1.05)

−0.29 (0.75)

0.13 (1.13)

0.14 (0.98)

0.38 (1.24)

ns

ns

ns

        

  C-PBd (child self-reported problem behavior)

0.22 (0.83)

−0.30 (0.55)

0.26 (0.80)

0.00 (0.86)

0.09 (0.88)

*

ns

ns

        

Note. ns: not significant; MZ: monozygotic; DZ: dizygotic; SD: standard deviation.aAchenbach (1991a, 1991b).bWalker and McConnell (1988).cForgatch et al. (1984).dHarter (1993).*p < .05.

Preliminary analyses were then conducted to assess whether intraclass correlations for DPA and PROB by zygosity might be confounded with the sex composition of the twin pairs. In light of the fact that only DZ twin pairs can be of opposite sex, it was important to test whether twin intraclass correlations might differ. In large samples with significant power to detect between group differences, this is often accomplished with group invariance modeling. Using this strategy, group equivalence is ascertained by comparing models in which parameter estimates between MZ same and DZ same and opposite sex groups are freely estimated and then systematically constrained to be equal (Neale & Cardon, 1992). Because of this study's small sample size, there was not sufficient statistical power to use this strategy.

As an alternate approach, the current analyses compared two models: one using a full sample model (all MZ and DZ twin pairs) in which sex was statistically controlled, and the second using a same sex only model in which only same sex MZ and DZ twins were included. These models examined whether the residual intraclass correlations for the full sample (after controlling for sex) would differ from those of a same sex only sample. Intraclass correlations indicated that estimates of genetic, and shared and nonshared environmental influences were similar across the two sets of models (see Table III). Nested chi-square model comparisons indicated that there were no statistically significant differences between the same sex only model and the full model controlling for sex. Given the apparent lack of systematic differences based on sex, we conducted the remaining analyses on the full sample and statistically controlled for sex.
Table III.

Twin Intraclass Correlations for MZ/DZ Same and Opposite Sex (All) and MZ/DZ Same Sex Only Twin Pairs

 

MZ/DZ All

MZ/DZ Same sex

Composites

MZ

DZ

MZ

DZ

Deviant peer affiliation (DPA)

 T-DP1a (teacher-reported peer externalizing behavior)

.03

.04

.03

.04

 T-DP2b (teacher-reported peer deviance)

.23

.11

.23

.28

 O-DFPc (coder impressions—deviant friendship process)

.26

.56

.26

.56

Problem behavior (PROB)

 P-PBa (parent-reported twin externalizing behavior)

.87

.63

.84

.63

 T-PBa (teacher-reported twin externalizing behavior)

.42

.30

.52

.11

 C-PBd (child self-reported problem behavior)

.40

.34

.49

.32

Note. MZ: monozygotic; DZ: dizygoticaAchenbach (1991a).bWalker and McConnell (1988).cForgatch et al. (1984).dHarter (1993).

In sum, the decision to control for sex for the remaining analyses was made for two reasons. First, the patterns of twin intraclass correlations did not vary greatly between the MZ–DZ same sex and opposite sex DZ twin pairs (after controlling for sex). Second, elimination of DZ opposite sex twin pairs reduced the total sample by 21% and would have significantly diminished statistical power. Therefore, sex effects were statistically removed by using residuals from the regression of deviant peer affiliation and problem behavior on sex. For the remaining analyses, all scales represent residual scores controlling for sex. The alternative would be to analyze these sex effects within the genetic models; however, the small sample size limited the statistical power and precluded this analysis.

Study variables were then examined by zygosity to test for systematic differences by twin age. Correlations for the MZ and DZ twin pairs did not differ systematically by age, indicating that there was no need to control for child age in the genetic analyses. Because children in twin dyads are the same age, scores do not generally require adjustment for age differences as would be necessary for non-twin sibling dyads (Rowe & Plomin, 1981).

Univariate Genetic Analyses

Intraclass Correlations

Univariate genetic analyses were conducted to examine the genetic and/or environmental variance in deviant peer affiliation and in child problem behavior and to examine variability in parameter estimates by informant and method. Twin intraclass correlations for teacher and observer ratings of deviant peer affiliation (T-DP1, T-DP2, O-DFP) and parent, teacher, and child ratings of problem behavior (P-PB, T-PB, and C-PB) were computed separately for MZ and DZ twins (see Table IV), and yielded estimates of heritability, and shared and nonshared environmental influences.
Table IV.

Summary of Twin Intraclass Correlations and Parameter Estimates for Deviant Peer Association and Problem Behavior by Rater

Scales

MZ

DZ

h2

c2

e2

Deviant peer affiliation (DPA)

 T-DP1a (teacher-reported peer externalizing behavior)

.03

.04

.00

.03

.97

 T-DP2b (teacher-reported peer deviance)

.23

.11

.24

.00

.76

 O-DFPc (coder impressions—deviant friendship process)

.26

.56

.00

.26

.74

Problem behavior (PROB)

 P-PBa (Parent-reported twin externalizing behavior)

.87

.63

.48

.39

.13

 T-PBa (teacher-reported twin externalizing behavior)

.42

.30

.24

.18

.58

 C-PBd (child self-reported problem behavior)

.40

.34

.12

.28

.60

Note. MZ: monozygotic; DZ: dizygotic.*Falconer's formulas for computation of univariate parameter estimates: heritability, h2 = (MZ−DZ)×2; shared environment, c2 = MZ−h2; nonshared environment: e2 = 1−(h2c2); Falconer (1989).aAchenbach (1991a,)bWalker and McConnell (1988).cForgatch et al. (1984).dHarter (1993).

Examination of intraclass correlations by informant suggested variability in parameter estimates. Computed heritability estimates for deviant peer affiliation using teacher reports (T-DP2) revealed moderate genetic and robust nonshared environmental influences but virtually no shared environmental effects. Teacher ratings of peer problem behavior (T-DP1) and coder impressions of twin peer process with best friends (O-DFP) suggested significant nonshared environmental influence, moderate-to-little shared environmental variance, and no genetic effects.

Heritability estimates for problem behavior also varied as a function of informant. Parent reports (P-PB) suggested that twin problem behavior was most strongly influenced by genetic and shared environmental factors. Conversely, teacher ratings (T-PB) and twin self-reported problem behavior (C-PB) indicated that problem behavior was strongly influenced by genetic and nonshared environmental factors. Collectively, these findings suggest the importance of attending to rater and method in the decomposition of peer deviance and problem behavior scores into genetic, and shared and nonshared environmental components.

Univariate Genetic Modeling

Univariate parameter estimates (i.e., estimates of heritability, and shared and nonshared environment influences for DPA and PROB) were then computed using the standard univariate structural equation model (Neale & Cardon, 1992). Three independent latent variables were estimated in this model for each twin, using variance/covariance matrices: additive genetic influences (A), shared environment influences (C), and nonshared environment influences plus error (E). The variance estimates (e.g., genetic variance or heritability, shared environmental variance, and nonshared environmental variance) are derived from path estimates, and confidence intervals are used to interpret their statistical significance (for more information, see Neale & Cardon, 1992).

We used a model-fitting strategy in which several alternative models were compared. These included the full model (ACE), and three submodels—one in which the genetic variance was set to zero (CE), another in which the shared environmental variance was set to zero (AE), and a third in which the genetic and shared environmental variances were set to zero (E). The overall fit of the complement of genetic analyses was determined by examining the χ2 indexes, where the degrees of freedom equaled the total number of estimated parameters less the number of free parameters. Considering that χ2 values are sensitive to sample size, are subject to improvement as more parameters are included in the model, and are likely to reject models that may fit the data well (Neale & Cardon, 1992), we used two additional fit indexes: root mean square error of approximation(RMSEA) and Akaike's Information Criterion(AIC) (Browne & Cudeck, 1989) is a widely accepted alternative for model fit testing. In general, RMSEA values of less than .05 suggest a good model fit, whereas values between .05 and .08 suggest an adequate model fit. The relative goodness of fit was also examined using, AIC for which a higher negative number suggests a better fit (Medsker, Williams, & Holahan, 1994). Differences in model fit were assessed by a direct comparison of the AIC and RMSEA models, with the better fit indicated by the lower values.

For most informants and methods, the univariate models fit the data well or adequately. The best fitting models are highlighted in Table V. For deviant peer affiliation, all three measures indicated substantial nonshared environmental effects ranging from 63 to 100%. In addition, teacher report of deviant peer group affiliation (T-DP2) indicated modest heritability (18%), and coder impressions of peer process (O-DP) revealed moderate shared environmental influences (37%).
Table V.

Univariate ACE Models for Deviant Peer Association and Problem Behavior by Rater

Scales

Model

h2

CI

c2

CI

e2

CI

χ2

df

p value

AIC

RMSEA

Deviant peer association (DPA)

 T–DP1a (teacher-reported peer externalizing behavior)

ACE

.06

0.00–0.32

.00

0.00–0.00

.94

0.68–1.00

7.58

3

.06

0.15

.15

 

CE

.00

0.00–0.00

.04

0.00–0.23

.96

0.77–1.00

7.57

4

.11

−0.43

.09

 

AE

.06

0.00–0.32

.00

0.00–0.00

.94

0.68–1.00

7.58

4

.11

−0.42

.09

 

E

.00

0.00–0.00

.00

0.00–0.00

1.00

1.00–1.00

7.76

5

.17

−2.24

.08

 T–DP2b (teacher-reported deviant peer group)

ACE

.14

0.00–0.39

.03

0.00–0.30

.83

0.61–1.00

3.38

3

.34

−2.62

.04

 

CE

.00

0.00–0.00

.13

0.00–0.31

.87

0.69–.1.00

1.49

4

.82

−6.55

.00

 

AE

.18

0.00–0.40

.00

0.00–0.00

.82

0.61–1.00

1.33

4

.86

−6.67

.00

 

E

.00

0.00–0.00

.00

0.00–0.00

1.00

1.00–1.00

3.38

5

.64

−6.62

.00

 O–DFPc (coder impressions—deviant friendship process)

ACE

.00

0.00–0.25

.37

0.12–0.51

.63

0.49–0.80

6.77

3

.08

0.08

.13

 

CE

.00

0.00–0.00

.37

0.20–0.51

.63

0.49–0.80

6.77

4

.15

−1.23

.09

 

AE

.35

0.15–0.51

.00

0.00–0.00

.65

0.49–0.85

13.24

4

.01

5.24

.19

 

E

.00

0.00–0.00

.00

0.00–0.00

1.00

1.00–1.00

24.42

5

.00

14.42

.24

Problem behavior (PROB)

 P–PBa (parent-reported externalizing behavior)

ACE

.39

0.17–0.70

.46

0.16–0.67

.15

0.10–0.22

5.93

3

.12

−0.07

.11

 

CE

.00

0.00–0.00

.76

0.68–0.82

.24

0.18–0.32

18.70

4

.00

10.70

.22

 

AE

.85

0.79–0.89

.00

0.00–0.00

.15

0.11–0.21

13.45

4

.01

5.48

.18

 

E

.00

0.00–0.00

.00

0.00–0.00

1.00

1.00–1.00

143.43

5

.00

133.43

.60

 T–PBa (teacher-reported externalizing behavior)

ACE

.46

0.00–0.67

.05

0.00–0.45

.49

0.34–0.73

17.10

3

.00

11.10

.28

 

CE

.00

0.00–0.00

.36

0.20–0.52

.64

0.50–0.80

19.61

4

.00

11.61

.25

 

AE

.51

0.31–0.66

.00

0.00–0.00

.49

0.34–0.69

17.14

4

.00

9.14

.23

 

E

.00

0.00–0.00

.00

0.00–0.00

1.00

1.00–1.00

36.67

5

.00

26.67

.32

 C-PBd (child-reported problem behavior)

ACE

.26

0.00–0.61

.19

0.00–0.49

.55

0.39–0.75

5.63

3

.13

−0.37

.09

 

CE

.00

0.00–0.00

.37

0.23–0.50

.63

0.50–0.77

6.61

4

.16

−1.39

.08

 

AE

.48

0.30–0.62

.00

0.00–0.00

.52

0.38–0.70

6.41

4

.17

−1.59

.07

 

E

.00

0.00–0.00

.00

0.00–0.00

1.00

1.00–1.00

27.72

5

.00

18.72

.11

Note. CI: confidence interval; AIC: Akaike's Information Criterion fit index; RMSEA: root mean square error of approximation. ACE model includes h2, c2, and e2; CE model includes c2 and e2 with h2 parameter set to 0; AE model includes a2 and e2 with c2 parameter set to 0. Underlined models represent best-fitting models.aAchenbach (1991a, 1991b).bWalker and McConnell (1988).cForgatch et al. (1984).dHarter (1993).

Genetic and environmental parameter estimates for child problem behavior also varied by rater. All three measures indicated moderate to substantial genetic effects, with heritability estimates ranging from 39 to 51%. For child self-reported and teacher-reported problem behavior (C-PB and T-PB, respectively), the AE model yielded the most robust fit, with 48–51% of the variance in child behavior problems accounted for by genetic variance. In contrast, parental reports of twin problem behavior demonstrated significant levels of genetic (39%) and shared environmental (46%) influence, as well as a modest degree of nonshared environmental (15%) effects. Collectively, these univariate genetic models illustrated noteworthy effects by rater and method, with parent, teacher, and child reports generally yielding higher estimates of genetic variance than coder ratings of observed twin–friend interactions.

DISCUSSION

This study is one of the first to assess multiple domains of child functioning (deviant peer affiliation and problem behavior) using an MTMM behavioral genetic design. Analyses focused on the identification of genetically influenced child behaviors, as well as shared and nonshared environmental mechanisms for deviant peer affiliation and problem behavior. To advance current theories of peer influences and problem behavior, two questions were examined: (1) Is there evidence of genetic and/or environmental variance in deviant peer affiliation, as well as in child problem behavior? and (2) To what extent do estimates of genetic and environmental influence variance differ by rater? Quantitative genetic modeling of peer socialization moves the field one step closer to inferring whether a specific child effect (genetic constitution), as well as shared and nonshared environmental influences, contribute meaningfully to the etiology of deviant peer affiliation and child and early adolescent behavior problems.

Deviant Peers and Problem Behavior: Genetic or Environmental Influence?

Results from this study (as well as existing literature) point to moderate genetic and shared environmental influences and moderate-to-substantial nonshared environmental influences for youth problem behavior. Average (across method and rater) parameter estimates in this study were h2 = .37, c2 = .23, and e2 = .40; values that correspond very closely to those derived in Rhee and Waldman's (2002) meta-analysis: h2 = .32, c2 = .16, and e2 = .43.

A different picture emerges in the extant literature on gene–environment processes in relation to deviant peer affiliation. In this study, much of the variance in peer deviance appears to be attributable to nonshared environmental influences, with little evidence of genetic or shared environmental effects. Average (across method and rater) parameter estimates in the current study were h2 = .07, c2 = .13, and e2 = .80. These findings are consistent with Manke et al. (1995), who reported that two-thirds to nearly all of the variance in positive and negative interactions with best friends was attributable to nonshared environment. However, they also found evidence of some heritable variance in positive friendship interactions. In a subsequent analysis in which that same sample was combined with that of an adoptive sibling study, a similar result emerged with the majority of the variance in adolescent delinquency being attributed to nonshared environmental sources (Iervolino et al., 2002). It is interesting to note that these studies provided a different pattern of results when parents' reports were used. For example, Pike et al. (2000) found that parents' ratings of the delinquency orientation of their teenagers' peer groups showed substantial heritability (.49 and .71) and shared environmental variance (.21 and .35 for fathers and mothers, respectively) and little nonshared environmental variance. Consistent with parent reports, youth self-reports also provided evidence of genetic variance in peer delinquency orientation (Pike et al., 2000).

In the current study, parameter estimates for teacher reports of twin deviant peer affiliation were slightly different than those from teacher reports of the externalizing behavior of each twin's selected friend. Whereas the latter attributed all of the influence in friends' externalizing behavior to nonshared environmental influences, the former suggests that a small amount of the influence in deviant peer affiliation may be genetic. Although these univariate models are not designed to test for gene–environment correlation, these findings raise the possibility of active or evocative child effects in their peer group selection. These effects can be tested using bivariate genetic modeling. Exploratory bivariate ACE models were computed for teacher ratings of deviant peer group affiliation and teacher–reported twin externalizing behavior in the current sample. The fit indices were inadequate, however, a finding that is not surprising given the sample size and the fact that moderate genetic effects were not detected for deviant peer affiliation at the univariate level.

Parameter Estimate Variability by Informant and Method

One of these data's most notable features was the pattern of estimates of heritability and shared and nonshared environment, by informant and method. All three deviant peer affiliation measures were dominated by nonshared environmental variance (63–100%). There were some differences, however, between teacher-reported twin deviant peer group affiliation (T-DP2), in which modest heritability was detected, and coder impressions of deviant friendship process (O-DFP), in which approximately one-third of the variance was attributed to shared environment. These findings are consistent with previous investigations in which observational measures tended to yield higher estimates of shared environmental variance (Borkenau et al., 2000; Deater-Deckard et al., 2001; Deater-Deckard & O’Conner, 2000; Ge et al., 1996; Leve et al., 1998). These within-rater effects suggest that estimates of heritability may vary depending upon the behavior that is being measured and the nature of the assessment.

Parameter estimates of genetic, and shared and nonshared environmental influences for twin behavior problems also differed by rater. Whereas teacher and child reports suggested moderate to substantial genetic influences for twin problem behavior (ranging from 39 to 51%), substantial nonshared environmental influences were also found. Indeed, the total variance in the prediction of twin problem behavior was split fairly equally between genetic and nonshared environmental effects.

Parent-reported twin problem behavior (P-PB) provided a different picture than did child and teacher ratings in that parent reports yielded evidence of significant shared environmental influence. The modest genetic effects found for parent reports of twin externalizing (P-PB) were somewhat inconsistent with those of previous genetically informative studies. Contrary to those investigations, the parent-reported genetic effects in this study were smaller than those reported for child and teacher-rated estimates of genetic effects (Deater-Deckard, 2000; Miles & Carey, 1997; Rhee & Waldman, 2002). The detection of rater differences for genetic and environmental parameter estimates is consistent with a growing number of quantitative genetic studies in which estimates varied by rater and method (Plomin, DeFries, McClearn, & McGuffin, 2000; Rhee & Waldman, 2002).

When considering potential rater effects in this sample, it is important to note that there was not adequate statistical power to test whether ACE model differences among raters were significantly different from one another. Although these findings generally reflect those of previous studies, they should be interpreted with caution. Nevertheless, we encourage researchers to collect MTMM data with larger samples of twins.

There are several possible explanations for the rater differences detected in this investigation. The discrepant estimates of heritable and environmental influences may be a result of behavioral inconsistency across contexts (home and school), characteristics of the sample, method effects, or perceptual biases of the informants. A diverse literature supports the proposition that parent–child and parent–teacher agreement about child behavior is low (Achenbach, McConaughy, & Howell, 1987; Andrews, Garrison, Jackson, Addy, & McKeown, 1993; Bartels et al., 2003; Hudziak et al., 2003; Offord et al., 1996), suggesting that observable behaviors differ by setting, method, or the perception of the informant (Dishion & Patterson, 1999).

From an ecological perspective, the development of deviant peer networks and psychopathology is conceptualized as a function of synchronous levels of influence across settings (Bronfenbrenner, 1989). Therefore, some behavioral consistency is assumed to exist across contexts. Research with clinical populations points to reliable rates of cross-setting consistency for deviant peer process across home and school environments (Dishion, 2000). Thus, the general consistency across raters for nonshared environmental influences on deviant peer affiliation and the genetic influences on problem behavior found in this study may reflect the theorized cross-setting consistency. The discrepancies in the presence or magnitude of genetic and shared environmental influences on deviant peer affiliation and of shared and nonshared environmental effects on problem behavior may represent state-specific child behaviors in which certain patterns of behavior emerge only in contexts in which they are relevant or reinforced. In this nonreferred community twin sample, children who did not exhibit adjustment difficulties may have been more responsive to the environmental contingencies inherent in particular contexts.

It is also important to consider that parents' impressions of their biological offspring are genetically confounded with their children's behaviors, in that parents and their biological children share, on average, 50% of their genes (Turkheimer & Waldron, 2000). It is plausible that this passive gene–environment correlation may create systematic differences between parent, teacher, and observer perceptions of child behavior. This has yet to be empirically tested.

Methodological Issues for Behavioral Genetic Research

The MTMM literature provides some insight into these differences. Although an MTMM approach is encouraged as a means to reduce the possibility of monomethod bias (Cook & Campbell, 1979; Dwyer, 1983), indicators derived from the same measurement method (Bank et al., 1990) and/or method variance may complicate the interpretation of measurement models (Dishion, Burraston, & Li, 2002; Dishion & Patterson, 1999). For example, in an investigation of the relation of parenting practices to adolescent problem behavior using an MTMM strategy, Dishion et al. (2002) found that method variance accounted for as much variability in adolescent problem behavior as parenting characteristics.

Indeed, twin intraclass correlations across DPA and PROB scales in this study (from r = −.03 to r = .87) suggest notable differences by reporting agent (child, parent, and teacher) as well as method (questionnaire vs. coder impressions of observed behavior). When considered in the context of estimates of genetic and environmental influences, the low-to-moderate correlations between scales designed to measure a similar phenomenon (DPA or PROB) raise the possibility that, indeed, rater perceptions, method biases, or an inconsistency of child behaviors across settings may be, in part, responsible for variability in genetic and environmental estimates. Composite indicators are likely to obscure these effects. At the same time, composites work very well when there is good convergent validity across the multiple indicators. When this is the case, composite scores likely yield the most reliable estimates of genetic and environmental sources of variance.

A recent meta-analysis of behavioral genetic studies provides a compelling illustration of potential method effects. The meta-analysis revealed that brief semistructured behavioral observations may elicit a more restrictive sample of behaviors than do naturalistic observations (Rhee & Waldman, 2002). A narrower sampling of child behavior, in turn, may yield different parameter estimates than would questionnaire data. Studies that combine data across informants may diffuse the distinct patterns of behavior that emerge by context, leading to what some researchers have referred to as a “glop” problem (Bank et al., 1990). As such, the variability of quantitative genetic parameter estimates by informant and method in this study draws attention to the viability of the MTMM approach for behavioral genetic research.

In summary, this study suggests that genetic, and shared and nonshared environmental influences play a role in deviant peer affiliation and child problem behavior. Although child and adolescent problem behavior is likely to be embedded within a group dynamic in which children and adolescents promote antisocial norms and behavior, individual differences among youth may also contribute to this process. Parameter estimates of these influences varied by rater and method, suggesting that considerable attention should be paid to the specific measure administered when interpreting results of behavioral genetic, as well as environmental, studies.

Notes

ACKNOWLEDGMENTS

This project is based on the doctoral dissertation of the first author and was supported by Grant P50 MH 46690, National Institutes on Mental Health and Office of Research on Mental Health at the National Institutes of Health, to John B. Reid, and by Grant R03 MH 57053, National Institutes on Mental Health at the National Institutes of Health, to Leslie D. Leve. The authors extend their appreciation to Margaret Grace Cooper and Beverly Fagot, to whom this work is dedicated.

REFERENCES

  1. Achenbach, T. M. (1991a). Manual for the Child Behavior Checklist/4–18 and 1991 Profile. Burlington, VT: University of Vermont.Google Scholar
  2. Achenbach, T. M. (1991b). Manual for the Teacher's Report Form and 1991 Profile. Burlington, VT: University of Vermont.Google Scholar
  3. Achenbach, T. M., McConaughy, S. H., & Howell, C. T. (1987). Child/adolescent behavioral and emotional problems: Implications of cross-informant correlations for situational specificity. Psychological Bulletin, 101, 213–232.CrossRefPubMedGoogle Scholar
  4. Andrews, V. C., Garrison, C. Z., Jackson, K. L., Addy, C. L., & McKeown, R. E. (1993). Mother–adolescent agreement on symptoms and diagnoses of adolescent depression and conduct disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 32, 731–738.PubMedCrossRefGoogle Scholar
  5. Baker, L. A., & Daniels, D. (1990). Nonshared environmental influences and personality differences in adult twins. Journal of Personality and Social Psychology, 58, 103–110.CrossRefPubMedGoogle Scholar
  6. Bank, L., Dishion, T. J., Skinner, M., & Patterson, G. R. (1990). Method variance in structural equation modeling: Living with “glop.” In G. R. Patterson (Ed.), Depression and aggression in family interaction (pp. 247–279). Mahwah, NJ: Erlbaum.Google Scholar
  7. Bartels, M., Hudziak, J. J., van den Oord, E. J. C. G., van Beijsterveldt, C. E. M., Rietveld, M. J. H., & Boomsma, D. I. (2003). Co-occurrence of aggressive behavior and rule-breaking behavior at age 12: Multirater analysis. Behavior Genetics, 33, 607–621.CrossRefPubMedGoogle Scholar
  8. Borkenau, P., Riemann, R., Angleitner, A., & Spinath, F. M. (2000). Genetic and environmental influences on observed personality: Evidence from the German Observational Study of Adult Twins. Journal of Personality and Social Psychology, 80, 655–668.CrossRefGoogle Scholar
  9. Bronfenbrenner, U. (1989). Ecological systems theory. In P. Vasta (Ed.), Annals of child development: Vol. 6. Six theories of child development: Revised formulations and current issues (pp. 187–249). London: Jai Press.Google Scholar
  10. Browne, M. W., & Cudeck, R. (1989). Single sample cross-validation indices for covariance structures. Multivariate Behavioral Research, 24, 445–455.CrossRefGoogle Scholar
  11. Capaldi, D. M., Dishion, T. J., Stoolmiller, M., & Yoerger, K. (2001). Aggression toward female partners by at-risk young men: The contribution of male adolescent friendships. Developmental Psychology, 37, 61–73.CrossRefPubMedGoogle Scholar
  12. Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation design and analysis for field settings. Boston: Houghton Mifflin.Google Scholar
  13. Daniels, D., & Plomin, R. (1985). Differential experience of siblings in the same family. Developmental Psychology, 21, 747–760.CrossRefGoogle Scholar
  14. Deater-Deckard, K. (2000). Parenting and child behavioral adjustment in early childhood: A quantitative genetic approach to studying family process. Child Development, 72, 468–484.CrossRefGoogle Scholar
  15. Deater-Deckard, K., & O’Connor, T. G. (2000). Parent–child mutuality in early childhood: Two behavioral genetic studies. Developmental Psychology, 36, 561–570.CrossRefPubMedGoogle Scholar
  16. Deater-Deckard, K., Pike, A., Petrill, S. A., Cutting, A. L., Hughes, C., & O’Connor, T. G. (2001). Nonshared environmental processes in socio-emotional development: An observational study of identical twin differences in the preschool period. Developmental Science, 4, F1–F6.CrossRefGoogle Scholar
  17. Deater-Deckard, K., & Plomin, R. (1999). An adoption study of the etiology of teacher and parent reports of externalizing problems in middle childhood. Child Development, 70, 144–154.CrossRefPubMedGoogle Scholar
  18. Dishion, T. J. (2000). Cross-setting consistency in early adolescent psychopathology: Deviant friendships and problem behavior sequelae. Journal of Personality, 68, 1109–1126.CrossRefPubMedGoogle Scholar
  19. Dishion, T. J., Andrews, D. W., Kavanagh, K., & Soberman, L. H. (1996). Preventive interventions for high-risk youth: The Adolescent Transitions Program. In R. D. Peters & R. J. McMahon (Eds.), Preventing childhood disorders, substance abuse, and delinquency (pp. 184–214). Thousand Oaks, CA: Sage.Google Scholar
  20. Dishion, T. J., & Bullock, B. M. (2002). Parenting and adolescent problem behavior: An ecological analysis of the nurturance hypothesis. In J. G. Borkowski, S. Ramey, & M. Bristol-Power (Eds.), Parenting and the child's world: Influences on intellectual, academic, and social–emotional development (pp. 231–249). Mahwah, NJ: Erlbaum.Google Scholar
  21. Dishion, T. J., Burraston, B., & Li, F. (2002). A multimethod and multitrait analysis of family management practices: Convergent and predictive validity. In W. Bukoski & Z. Amsel (Eds.), Handbook for drug abuse prevention theory, science, and practice (pp. 587–607). New York: Plenum.Google Scholar
  22. Dishion, T. J., Capaldi, D., Spracklen, K. M., & Li, F. (1995). Peer ecology of male adolescent drug use. Development and Psychopathology, 7, 803–824.CrossRefGoogle Scholar
  23. Dishion, T. J., Eddy, M., Haas, E., Li, F., & Spracklen, K. M. (1997). Friendships and violent behavior during adolescence. Social Development, 6, 207–223.CrossRefGoogle Scholar
  24. Dishion, T. J., Nelson, S. E., & Bullock, B. M. (2004). Premature adolescent autonomy: Parent disengagement and deviant peer process in the amplification of problem behavior. Journal of Adolescence, 27, 515–530.PubMedCrossRefGoogle Scholar
  25. Dishion, T. J., & Patterson, G. R. (1999). Model-building in developmental psychopathology: A pragmatic approach to understanding and intervention. Journal of Clinical Child Psychology, 28, 502–512.CrossRefPubMedGoogle Scholar
  26. Dwyer, J. H. (1983). Statistical models for the social and behavioral sciences. New York: Oxford University Press.Google Scholar
  27. Eaves, L., Silberg, J. L., Maes, H. H., Simonoff, E., Pickles, A., Rutter, M., et al. (1997). Genetics and developmental psychopathology: 2. The main effects of genes and environment on behavioral problems in the Virginia Twin Study of Adolescent Behavioral Development. Journal of Child Psychology and Psychiatry, 38, 965–980.PubMedCrossRefGoogle Scholar
  28. Elliott, D. S., Huizinga, D., & Ageton, S. S. (1985). Explaining delinquency and drug use. Thousand Oaks, CA: Sage.Google Scholar
  29. Falconer, D.S. (1988). Introduction to quantitative genetics (3rd ed.). New York: Longman.Google Scholar
  30. Forgatch, M., Fetrow, B., & Lathrop, M. (1984). Coder impressions. Unpublished training manual, Oregon Social Learning Center, Eugene.Google Scholar
  31. Ge, X., Conger, R. D., Cadoret, R. J., Neiderhiser, J. M., Yates, W., & Troughton, E. (1996). The developmental interface between nature and nurture: A mutual influence model of child antisocial behavior and parent behaviors. Developmental Psychology, 32, 574–589.CrossRefGoogle Scholar
  32. Goldsmith, H. H. (1991). A zygosity questionnaire for young twins: A research note. Behavior Genetics, 21, 257–269.PubMedCrossRefGoogle Scholar
  33. Harter, S. (1993). What I am like. Unpublished coding instrument, University of Denver.Google Scholar
  34. Hartup, W. W. (1996). The company they keep: Friendships and their developmental significance. Child Development, 67, 1–13.PubMedCrossRefGoogle Scholar
  35. Hartup, W. W., & Stevens, N. (1997). Friendships and adaptation in the life course. Psychological Bulletin, 121, 355–370.CrossRefGoogle Scholar
  36. Hoffman, L. W. (1991). The influence of the family environment on personality: Accounting for sibling differences. Psychological Bulletin, 110, 187–203.CrossRefGoogle Scholar
  37. Hudziak, J. J., van Beijsterveldt, C. E. M., Bartels, M., Rietveld, M. J., Rettew, D. C., Derks, E. M., et al. (2003). Behavior Genetics, 33, 575–589.CrossRefPubMedGoogle Scholar
  38. Iervolino, A. C., Pike, A., Manke, B., Reiss, D., Hetherington, E. M., & Plomin, R. (2002). Genetic and environmental influences in adolescent peer socialization: Evidence from two genetically sensitive designs. Child Development, 73, 162–174.CrossRefPubMedGoogle Scholar
  39. Leve, L. D., Winebarger, A. A., Fagot, B. I., Reid, J. B., & Goldsmith, H. H. (1998). Environmental and genetic variance in children's observed and reported maladaptive behavior. Child Development, 69, 1286–1298.PubMedCrossRefGoogle Scholar
  40. Loeber, R., & Dishion, T. (1983). Early predictors of male delinquency: A review. Psychological Bulletin, 94, 68–99.CrossRefPubMedGoogle Scholar
  41. Loeber, R., Stouthamer-Loeber, M., Van Kammen, W., & Farrington, D. P. (1991). Initiation, escalation, and desistance in juvenile offending and their correlates. Journal of Criminal Law and Criminology, 82, 36–82.CrossRefGoogle Scholar
  42. Lyons, M. J., True, W. R., Eisen, S. A., Goldberg, J., Meyer, J. M., Faraone, S. V., et al. (1995). Differential heritability of adult and juvenile antisocial traits. Archives of General Psychiatry, 52, 906–915.PubMedGoogle Scholar
  43. Manke, B., McGuire, S., Reiss, D., Hetherington, E. M., & Plomin, R. (1995). Genetic contributions to adolescents' extrafamilial social interactions: Teachers, best friends, and peers. Social Development, 4, 238–256.CrossRefGoogle Scholar
  44. Medsker, G. J., Williams, L. J., & Holahan, P. J. (1994). A review of current practices for evaluating causal models in organizational behavior and human resources management research. Journal of Management, 20, 439–464.CrossRefGoogle Scholar
  45. Miles, D. R., & Carey, G. (1997). Genetic and environmental architecture of human aggression. Journal of Personality and Social Psychology, 72, 207–217.CrossRefPubMedGoogle Scholar
  46. Neale, M. C., & Cardon, L. R. (1992). Methodology of genetic studies of twins and families. Dordrecht, The Netherlands: Kluwer.Google Scholar
  47. O’Connor, T. G., & Plomin, R. (2000). Developmental behavioral genetics. In A. J. Sameroff & M. Lewis (Eds.), Handbook of developmental psychopathology (2nd ed., pp. 217–235). New York: Kluwer/Plenum.Google Scholar
  48. Offord, D. R., Boyle, M. H., Racine, Y., Szatamari, P., Fleming, J. E., Sanford, M., et al. (1996). Integrating assessment data from multiple informants. Journal of the American Academy of Child and Adolescent Psychiatry, 35, 1078–1085.CrossRefPubMedGoogle Scholar
  49. Parker, J. G., & Asher, S. R. (1997). Peer relations and later personal adjustment: Are low-accepted children at risk? Psychological Bulletin, 102, 357–389.CrossRefGoogle Scholar
  50. Patterson, G. R. (1982). Coercive family process. Eugene, OR: Castalia.Google Scholar
  51. Patterson, G. R., Reid, J., & Dishion, T. J. (1992). A social interactional approach: IV. Antisocial boys. Eugene, OR: Castalia.Google Scholar
  52. Pike, A., Manke, B., Reiss, D., & Plomin, R. (2000). A genetic analysis of differential experiences of adolescent siblings across three years. Social Development, 9, 96–114.CrossRefGoogle Scholar
  53. Plomin, R. (1994). Genetics and experience: The interplay between nature and nurture. Thousand Oaks, CA: Sage.Google Scholar
  54. Plomin, R., DeFries, J., McClearn, G., & McGuffin, P. (2000). Behavioral genetics (4th ed.). New York: Worth Publishers.Google Scholar
  55. Quay, H. C., & Peterson, D. R. (1987). Manual for the Behavior Problem Checklist. Miami, FL: Authors.Google Scholar
  56. Reiss, D., Neiderhiser, J., Hetherington, E. M., & Plomin, R. (2000). The relationship code: Deciphering genetic and social influences on adolescent development. Cambridge, MA: Harvard University Press.Google Scholar
  57. Rhee, S. H., & Waldman, I. D. (2002). Genetic and environmental influences on antisocial behavior: A meta-analysis of twin and adoption studies. Psychological Bulletin, 128, 490–529.CrossRefPubMedGoogle Scholar
  58. Rowe, D. C., & Plomin, R. (1981). The importance of nonshared (E-sub-1) environmental influences in behavioral development. Developmental Psychology, 17, 517–530.CrossRefGoogle Scholar
  59. Turkheimer, E., & Waldron, M. (2000). Nonshared environment: A theoretical, methodological, and quantitative review. Psychological Bulletin, 126, 78–108.CrossRefPubMedGoogle Scholar
  60. Walker, H. M., & McConnell, S. R. (1988). Walker–McConnell scale of social competence and school adjustment. Austin, TX: Pro-Ed, Inc.Google Scholar

Copyright information

© Springer Science+Business Media, Inc. 2006

Authors and Affiliations

  • Bernadette Marie Bullock
    • 1
    • 4
  • Kirby Deater-Deckard
    • 2
  • Leslie D. Leve
    • 3
  1. 1.Child and Family CenterUniversity of OregonEugeneUSA
  2. 2.Department of PsychologyVirginia Polytechnic Institute and State UniversityBlacksburgUSA
  3. 3.Oregon Social Learning CenterEugeneUSA
  4. 4.Child and Family CenterUniversity of OregonEugeneU.S.A

Personalised recommendations