PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Dev Psychol. Author manuscript; available in PMC 2017 May 1.
Published in final edited form as:
Published online 2016 February 25. doi:  10.1037/dev0000109
PMCID: PMC4844769
NIHMSID: NIHMS756824

Antisocial Peer Affiliation and Externalizing Disorders in the Transition from Adolescence to Young Adulthood: Selection versus Socialization Effects

Abstract

Prior research has demonstrated both socialization and selection effects for the relationship between antisocial peer affiliation and externalizing problems in adolescence. Less research has evaluated such effects post-adolescence. In this study, a cross-lagged panel analysis was used to evaluate the extent of socialization (i.e., the effect of antisocial peer affiliation on subsequent externalizing disorders) and selection (i.e., the effect of externalizing disorders on subsequent antisocial peer affiliation) in the prospective relationships between antisocial peer affiliation and externalizing disorders from adolescence through young adulthood. Data from a community sample of 2,769 individuals (52% female) with assessments at ages 17, 20, 24, and 29 were used. Analyses with a latent externalizing measure (estimated using clinical symptom counts of nicotine dependence, alcohol use disorder, illicit drug use disorder, and adult antisocial behavior) and self-reported antisocial peer affiliation revealed significantly stronger socialization effects from age 17 to 20, followed by significantly stronger selection effects from age 20 to 24 and 24 to 29. To better understand the impact of college experience, moderation by college status was evaluated at each developmental transition. Results were generally consistent for those who were in or were not in college. Results suggest selection effects are more important in later developmental periods than earlier periods, particularly in relation to an overall liability towards externalizing disorders, likely due to more freedom in peer selection post-adolescence.

Keywords: Adolescence, Antisocial Peers, Externalizing Disorders, Peer Relationships, Young Adulthood

Externalizing behavior refers to undersocialized behavior broadly and is often characterized as acting out or acting aggressively. Particularly problematic externalizing behavior generally refers to engaging in delinquent or illegal activity, such as adolescent substance use, as well as more disordered psychopathology, such as meeting a clinical threshold for childhood disruptive disorders or substance use disorders (Iacono, Malone, & McGue, 2008). Research has long shown a relationship between affiliation to deviant peers and adolescent externalizing problems (Brendgen, Vitaro, & Bukowski, 2000; Dishion, Véronneau, & Myers, 2010; Gardner & Steinberg, 2005), including substance use and substance use disorders (Curran, Stice, & Chassin, 1997; Dishion & Owen, 2002). Recent research has shown peer affiliation and externalizing problems are related through both processes of selection as well as socialization (Monahan, Steinberg, & Cauffman, 2009; Walden, McGue, Iacono, Burt, & Elkins, 2004; Samek, Keyes, Iacono, & McGue, 2013). Selection refers to how an adolescent’s current externalizing behavior predicts subsequent antisocial peer affiliation. Socialization refers to how an adolescent’s current affiliation to antisocial peers predicts subsequent externalizing behavior (Andrews, Tildesley, Hops, & Li, 2002; Monahan et al., 2009; Curran et al., 1997).

Whereas cross-sectional research is unable to tease apart these effects, a growing number of longitudinal studies have demonstrated reciprocity (i.e., both socialization and selection effects) in the relationship between deviant peer affiliation and adolescent externalizing behaviors (Curran et al., 1997; Kendler et al., 2007; van Ryzin, Fosco, & Dishion, 2012; van Ryzin & Dishion, 2014). For example, using latent growth curve modeling, Curran et al. (1997) found that both the initial status of peer alcohol use predicted subsequent increase of adolescent alcohol use and the initial status of adolescent alcohol use predicted subsequent increase of peer alcohol use, evidencing reciprocal selection and socialization influences. Other research has shown that after accounting for earlier deviant peer affiliation, deviant peer affiliation at ages 16-17 mediated the association between substance use at age 14 and subsequent substance dependence at age 19 (van Ryzin & Dishion, 2014), evidencing a socialization influence above and beyond the effect of a selection influence in mid-to-late adolescence.

On the other hand, limited research has systematically evaluated the relationship between antisocial peer affiliation and externalizing problems post-adolescence. One exception is a study conducted by Monahan et al. (2009), in which socialization and selection were evaluated in the transition from adolescence into young adulthood (age 14 to 22). Using a cross-lagged, structural equation analysis, these authors showed that while selection and socialization effects were both evident from ages 14 to 15 (such that both peer affiliation predicted subsequent antisocial behavior and antisocial behavior predicted subsequent peer affiliation), selection effects were not evident past age 16; only socialization effects were evident from ages 16 through 20. From ages 20 to 22, there was no evidence to support either socialization or selection. This overall pattern of results suggests selection may be less relevant past middle adolescence and that the influence of antisocial peer affiliation may be less relevant to the development of young adult antisocial behavior compared to adolescent antisocial behavior.

These results are somewhat at odds with developmental theory suggesting selection effects should become stronger in the transition of adolescence into young adulthood, as individuals have more freedom in selecting environments of their own choosing the older they become (e.g., see Scarr & McCartney, 1983). Monahan and colleagues (2009) argued that their findings demonstrate that peer influences may operate more in line with a socialization hypothesis in late adolescence because individuals have generally already sorted and established themselves into peer groups by this time period (thus there may be limited change in antisocial peer affiliation after middle adolescence). Additionally, Monahan et al. noted that peer influences (both selection and socialization) are likely less salient in early adulthood because individuals are more mature and thus more resistant to peer pressure as a result of this maturity.

On the other hand, the transition into young adulthood is often accompanied by a marked increase in binge-drinking and other risky behavior (Esser, Hedden, Kanney, Brewer, Gfrorer, & Naimi, 2014; Schulenberg & Maggs, 2002) particularly among college students (Substance Abuse and Mental Health Administration, 2014; Slutske, 2005; Schulenberg & Patrick, 2012; White, Labouvie, & Papadaratsakis, 2005). Thus, it seems somewhat odd that peer influences would not be salient in the development of antisocial behavior at this time. It is important to note that the Monahan et al. sample included individuals who were adjudicated for serious crime and the sample was predominately male. Thus, it is unclear if results will hold up in population-based studies, when increases in antisocial behavior are normative in the transition to young adulthood, and whether results are consistent for both males and females. Also, Monahan et al. evaluated this developmental process through age 22; it is unclear if this pattern continues into later young adulthood (i.e., through age 30 when antisocial behavior and problematic substance use decreases rapidly (Loeber, Menting, Lynam, Moffitt, Stouthamer-Loeber, & Stallings et al., 2012; Moffitt, 1993; SAMSHA, 2014)) or whether socialization or selection may again become relevant to explaining the link between deviant peer affiliation and externalizing disorder development after age 22.

Following prior developmental theory predicting a shift from more passive to active environment selection based on unique genetically-influenced traits and interests (Scarr & McCartney, 1983), we expect to see more influence of socialization than selection in adolescence and more influence of selection than socialization in early and later young adulthood. This is consistent with studies demonstrating that individuals often become more resistant to antisocial peers over time (Albert, Chein, & Steinberg, 2013; Gardner & Steinberg, 2005; Sumter, Bokhorst, Steinberg, & Westenberg, 2009) and fits with developmental theory suggesting that people are more likely to actively select environments as they reach adulthood, in part because of increasing freedom to do so (Scarr & McCartney, 1983).

On the other hand, peers may have unique effects on substance use and related externalizing behavior for college students in particular. It has long been demonstrated that college students use more substances (particularly alcohol) relative to their non-college-attending peers (SAMSHA, 2014; Schulenberg & Patrick, 2012; Schulenburg & Maggs, 2002; Slutske. 2005; White, Labouvie, & Papadaratsakis, 2005). College students are also often exposed to new peers in classrooms, on campus, and/or in conjunction with new living arrangements (e.g., dorm or off-campus apartment). Combined with the notion that college students in particular are more embedded in a culture of excessive drinking or “partying” (e.g., proximity to bars, marketing of drink specials towards college students, frequency of Greek and/or other parties, etc.), and that they are often exposed to new peers (on campus, in the classroom, and possibly in new living arrangements), we hypothesized that antisocial peer socialization processes may be more relevant to the development of externalizing problems and substance use for those in college compared to those not in college.

Taken together, we aimed to extend prior research on the topic of peer influences on substance use and externalizing problems by utilizing a cross-lagged panel design to test the prospective relationships between antisocial peer affiliation and externalizing disorders (nicotine dependence, alcohol use disorder, illicit drug use disorder, and adult antisocial behavior clinical symptom counts) at ages 17, 20, 24, and 29. We also evaluated differences in how college enrollment impacts the relationship between deviant peer affiliation and the development of externalizing disorders in adolescence through young adulthood. Guided by prior developmental theory (Scarr & McCartney, 1983), we expected to see stronger evidence for socialization effects earlier in time (i.e., age 17 to 20) followed by stronger effects for selection later in time (i.e., age 20 to 24 and age 24 to 29). As college experiences include a transition into new environments and new peer affiliations, we also hypothesized that socialization effects may be more apparent for those enrolled in college versus not enrolled in college.

Method

Participants

Data from the Minnesota Twin and Family Study (MTFS) were used. MTFS is a cohort-sequential, longitudinal study designed to evaluate the etiology of substance use and related externalizing behaviors (see Iacono et al., 1999 for a complete overview). Two cohorts of twins and their parents were included: the younger cohort, first assessed at age 11, and the older cohort, first assessed at age 17. Twins and parents were recruited from publically available birth certificates (years born 1972-1984); 90% of twin families were located and ~80% of those located and meeting study eligibility criteria participated. To be eligible, families had to live within a day’s drive to the university laboratory, and children could not have a mental or physical handicap that impaired participation at assessments. Follow-up assessments were completed through age 29 for both cohorts, resulting in overlapping assessments of the younger and older cohort at ages 17, 20, 24, and 29. Participation rates across follow-up assessments ranged from 88%-93% (more details provided in the Missing Data section, below).

In total, this sample includes 2,769 individuals (52% female) from 1,382 twin pairs (65% monozygotic, 35% dizygotic pairs, including 5 triplets), using data from the age 17 (M age = 17.8, SD = .69), age 20 (M age = 21.1, SD = .82), age 24 (M age = 25.0, SD = .90), and age 29 (M age = 29.4, SD = .67) assessments. Consistent with state demographics in the relevant birth years, nearly all participants were of European ancestry (96%), however, there was considerable diversity in other aspects, including socioeconomic status. The highest education completed for the majority of parents was a high school diploma or equivalent (63.5% for fathers, and 62.6% for mothers); 28.5% of fathers and 25.1% of mothers earned at least a BA/BS degree. At the intake assessment (assessment years ranged from 1990-1996), the median household income was $45,001 to $50,000. Twenty-five percent of families earned ≤ $40,000 and 25% of families earned ≥ $60,001; 7% earned ≤ $20,000, meeting federal poverty guidelines for a family of four in the relevant data collection years (US Census Bureau, 1996). Although the majority of rearing parents (79.3%) were married at the original intake assessment, 16.7% were divorced, 2.7% were never married and 1.3% were separated. The MTFS also included families from both rural (40%) and urban (60%) communities (Legrand et al., 2008), as well as approximately equivalent numbers of males (48%) and females (52%). As outlined below, externalizing disorders and antisocial peer affiliation were obtained in a consistent manner within each age-assessment period.

Measures

Externalizing problems

Clinical symptom counts of nicotine dependence, alcohol use disorder, illicit drug use disorder, and adult antisocial behavior were assessed at ages 17, 20, 24, and 29. SUDs are broadly characterized by social impairment and distress (e.g., desire/unable to cut down, continued use despite problems, marked tolerance). DSM III-R nicotine dependence symptom counts range from 0 to 7. Alcohol and illicit drug use disorder included a best estimate of both abuse and dependence symptom counts for a total range of 0 to 10 DSM III-R symptoms. Adult antisocial behavior is characterized by reckless disregard for others; meeting a threshold of both adult antisocial behavior in adulthood and conduct disorder prior to age 15 is required to meet the diagnosis for Antisocial Personality Disorder (American Psychological Association, 1987, 2013). Here, we are using the adult antisocial behavior symptom counts alone (e.g., irritable and aggressive), with a total range of 0 to 10 DSM III-R symptoms at each assessment.

All substance use disorder symptoms were assessed using the Substance Abuse Module of the Composite International Diagnostic Interview (Robins, Wing, Wittchen, Hu, Helzer, Babor, et al., 1988). Adult antisocial behavior symptoms were assessed using an interview comparable to the Structured Clinical Interview for DSM-III-R Axis II (Spitzer, Williams, & Gibbon, 1987). Clinical symptoms were reviewed by at least two individuals with advanced training and consensus by both individuals was necessary to assign symptoms. Kappa coefficients indexing diagnostic reliability were > .90 for all substance use disorders and .79 for adult antisocial behavior.

Antisocial peer affiliation

The Friends self-report survey (Burt, McGue, & Iacono, 2009; Walden et al., 2004) was used to assess antisocial peer affiliation at each time point. Participants rated their friendships on a scale of 1 (All of my friends are like that) to 4 (None of my friends are like that). A 7- to 9-item scale was used to measure affiliation with antisocial peers at each assessment (e.g., “My friends use drugs,” “My friends get into trouble with the police”); items were adapted, dropped, or added based on developmental relevance. At age 29, two items were dropped (“My friends drink alcohol or beer,” My friends smoke cigarettes or chew tobacco”) as having peers that occasionally drink or smoke in later adulthood may not be indicative of antisocial behavior (e.g., getting into trouble with police). A sensitivity analysis showed that scales that excluded vs. included these two items at the earlier assessments were almost perfectly correlated (r’s ranged from .96 to .97, all ps < .001). Follow-up analyses also revealed essentially identical results using scales from earlier assessments that excluded vs. included these two items.

Median scale internal consistency reliability (α) for antisocial peer affiliation was .88 for 17 year-olds, .83 for 20 year-olds, .83 for 24 year-olds, and .78 for 29 year-olds. Scales were highly correlated across time (r’s ranged from .48 to .67, also see Table 2), demonstrating moderate stability. To put antisocial peer measures on roughly the same metric as the estimated externalizing disorder latent factors (M = 0, SD = 1), the antisocial peer scales were standardized (within-assessment) prior to analysis. This is especially useful for purposes of comparison of unstandardized coefficients in cross-lagged panel analyses, described further below.

Table 2
Correlations among Raw Study Variables for Males (n = 1,333) and Females (n = 1,436)

College status

College status was measured using items from the Social Adjustment Interview at ages 20, 24, and 29. Participants were asked to report on their educational experiences. The highest level of college educational attainment was computed at each assessment in adulthood (1 = less than high school/GED, 2 = high school diploma or GED, 3 = some college/business certificate/technical degree, 4 = college degree, 5 = at least some professional or graduate school). For example, at age 29, 1.5% of the sample reported earning less than a high school diploma or GED as the highest level of education completed, 10.3% reported a high school diploma/GED, 37.0% reported completing at least some college, 26.6% had completed a college degree and 14.6% reported some professional/graduate school (10% were missing this information due to non-participation or failure to complete the associated interview or questions at the age 29 assessment).

Because we were interested in whether socialization vs. selection paths were moderated by college status, due to the new peers and friends typically encountered upon college entrance, we examined the effect of having initiated college during the time that preceded each assessment. Thus, at age 20, just over half the sample reported at least some college experience (n = 1,614, 58% of sample). At age 24, only 280 reported having new college experience since age 20 (i.e., had participated in college for the first time after the age 20 assessment, 10% of sample). Following this same line of logic, at age 29, only those who reported being in college for the first time at age 29 were considered to have a college status; n = 56 met this criteria at age 29 (2% of sample).

Missing Data

Little data were missing for externalizing disorders or reports of antisocial peer affiliation across time (missing cases for externalizing disorders at age 17: 5.5%, at age 20: 11.5%, at age 24: 10%, at age 29: 9.8%; missing cases for antisocial peer affiliation at age 17: 15.9%, at age 20: 15.6%, at age 24: 17.4%, at age 29: 12.7%). This was due to non-participation at each assessment (6.9%-11.5% across assessments) or failure to complete self-report surveys (2.9%-9.0% across assessments). To evaluate attrition effects, we compared mean differences in symptoms of externalizing disorders at age 17 for those who did or did not complete adult assessments. Those who participated in adult assessments had slightly fewer externalizing symptoms at age 17 than those who did not participate, but the effect sizes were small with average Cohen’s ds = −0.21, −0.16, and −0.03 at ages 20, 24, and 29, respectively, indicating little evidence of attrition effects.

Analysis Plan

Our primary goal was to examine selection and socialization effects through the use of prospective data analytic methods, specifically, a cross-lagged panel design (for an overview, see Berrington, Smith, & Sturgis, 2006; Kenny & Harackiewicz, 1979). Preliminary analyses were conducted to compare cross-lagged results across clinical symptom counts of nicotine dependence, alcohol use disorder, illicit drug use disorder, and adult antisocial behavior. Based on the similarity of results and high correlations between symptom count variables (see Table 2), we then estimated latent factors of externalizing disorders at ages 17, 20, 24, and 29 using clinical symptom counts of nicotine dependence, alcohol and illicit drug use disorder, and antisocial behavior and evaluated a cross-lagged structural equation model (see Figure 1). In all cross-lagged analyses, the prospective relationship between antisocial peer affiliation and externalizing disorders at ages 17, 20, 24, and 29 was tested, after accounting for autoregressive effects and residual correlations at each time point. All analyses were tested using Mplus, 7.2 (Muthén & Muthén, 1998-2014), using full information maximum likelihood and the maximum likelihood with robust standard errors (MLR) estimator to deal with non-normal distribution of externalizing disorder variables. The CLUSTER specification was used to account for non-independence of (twin) cases.

Figure 1
Study Model Depicting Hypotheses

Selection effects, referring to the effect of externalizing disorders on subsequent antisocial peer affiliation, are shown in bolded gray (see cross-path 1a, 2a, and 3a in Figure 1). Socialization effects, referring to the effect of antisocial peer affiliation on subsequent externalizing disorders, are shown in bolded black (see cross-path 1b, 2b, and 3b). To determine whether socialization effects were significantly stronger in magnitude in comparison to selection effects or vice versa, socialization and selection paths were constrained to be equivalent in three analyses (for each developmental transition; i.e., age 17 → 20, age 20 → 24, age 24 → 29) and the Satorra-Bentler Chi-Square difference test (used in conjunction with the MLR estimator) was used to measure a significant decrement in model fit between the free and constrained models.

A second aim was to test whether socialization and selection effects varied as a function of college enrollment status. To evaluate this, we tested the cross-lagged paths from the cross-lagged model (Figure 1) separately for those who had attended any college versus those who had not attended any college. Separate groupings were used for each developmental transition (college status at age 20 was used as a grouping variable for the analyses evaluating socialization vs. selection from age 17 to age 20, new college status at age 24 was used as a grouping variable for the analyses evaluating socialization vs. selection from age 20 to 24, and new college status at age 29 was used as a grouping variable for the analyses evaluating socialization vs. selection from for age 24 to 29). For example, if someone went to college for the first time at age 26, they would be grouped in the “no college experience” groups for the age 17 to age 20 analysis and the age 20 to age 24 analysis, but in the “new college experience” group for the age 24 to age 29 analysis. Significant differences in paths for those with any college versus no college were tested by constraining paths to be equivalent and using the Satorra-Bentler Chi-Square difference test to test for a significant decrement in model fit from the free and constrained models.

Results

Preliminary Analyses

Table 1 shows descriptive statistics on externalizing disorder symptom counts by sex. On average, males had higher mean symptom counts than females, across all externalizing disorders and across time (Cohen’s d ranged from .13 to .67, M = .39). Table 2 shows correlations among all study variables for both males and females. In general, antisocial peer affiliation and externalizing disorders were significantly and moderately to substantially correlated both within and across time, for both males and females.

Table 1
Descriptive Statistics for DSM-IIR Clinical Symptoms for Males (n = 1,333) and Females (n = 1,436)

Cross-Lagged Panel Analyses

Figure 2 shows results of each individual externalizing disorder (results for nicotine dependence are shown in Panel A, results for alcohol use disorder are shown in Panel B, results for illicit drug use disorder in Panel C, and results for adult antisocial behavior are shown in Panel D). The stability of antisocial peer affiliation was moderate and stable across time (β’s ranged from .59 to .65). The stability of externalizing disorders was smaller in magnitude from ages 17 to 20 (β’s ranged from .15 to .46) but stronger and more stable from ages 20 to 24 (β’s ranged from .43 to .69) and ages 24 to 29 (β’s ranged from .41 to .70).

Figure 2Figure 2
The Prospective Relationship between Antisocial Peer Affiliation and Each Indicator of Externalizing Disorders from Adolescence through Young Adulthood

Across each individual externalizing disorder, results showed that after accounting for the stability of traits and residual associations at each time point, there were stronger socialization than selection effects from ages 17 to 20. For example, the β corresponding to socialization path for nicotine dependence was small-to-moderate in magnitude (β = .24) and significant at p < .001, while the β corresponding to the selection path was estimated to be .00 and was not significantly different than zero at p < .05 (see Panel A of Figure 2). A similar pattern was found for alcohol use disorder (Panel B), illicit drug use disorder (Panel C) and adult antisocial behavior (Panel D). Following this, constraining the socialization and selection paths to be equivalent from age 17 to 20 for the separate nicotine, alcohol, drugs, and antisocial behavior models resulted in a significant decrement in model fit in all cases (nicotine: Δχ2 (1) = 37.80, p < .001; alcohol: Δχ2 (1) = 30.65, p < .001; drugs: Δχ2 (1) = 50.59, p < .001; adult antisocial behavior: Δχ2 (1) = 68.47, p < .001), suggesting socialization effects were significantly stronger than selection effects across the board.

On the other hand, and across each individual externalizing disorder, comparing socialization vs. selection paths from ages 20 to 24 and age 24 to 29 showed more similarity and reciprocity in effects. For example, for age 24 to age 29 the β corresponding to socialization path for nicotine dependence was small in magnitude (β = .05) but significant at p < .05, as was the β corresponding to the selection path (β = .12, p < .001). Similar effects were demonstrated for age 24 to 29 (see Panel A of Figure 2). Nonetheless, for nicotine dependence, constraining the socialization and selection pathways to be equivalent from 20 to 24 resulted in a significant decrement of fit (Δχ2 (1) = 7.15, p = .007), suggesting selection effects were significantly stronger than socialization effects from 20 to 24, as well (see Panel A of Figure 2 for details). However, constraining the socialization and selection pathways to be equivalent 24 to 29 did not result in a significant decrement of fit (Δχ2 (1) = .67, p = .41, suggesting selection and socialization effects were equivalent in magnitude from 24 to 29 for nicotine dependence.

There was more evidence for reciprocity in socialization vs. selection effects for ages 20 to 24 and 24 to 29 for the other externalizing disorders. Across alcohol use disorder, illicit drug use disorder, and adult antisocial behavior, constraining socialization and selection pathways to be equivalent from age 20 to 24 did not result in a significant decrement to model fit (alcohol: Δχ2 (1) = 2.62, p = .11; illicit drug use disorder: Δχ2 (1) = .39, p = .53; adult antisocial behavior: Δχ2 (1) = 1.16, p = .28). The same was true for constraining socialization and selection paths to be equivalent from age 24 to 29 for each of these externalizing disorder symptom counts (alcohol: Δχ2 (1) = 1.51, p = .22; illicit drug use disorder: Δχ2 (1) = .78, p = .38; adult antisocial behavior: Δχ2 (1) = .54, p =.46). This suggests equivalently strong selection and socialization effects involving antisocial peers and these alcohol use disorder, illicit drug use disorder, and adult antisocial behavior beyond age 20.

It is important to note that both selection and socialization effects were varied in effect size across developmental transition and by disorder. Squaring the standardized path coefficient (β) and dividing by the total predicted variance accounted for (R2) illustrates the percentage of explained variance accounted for by that pathway. Although large socialization effects from age 17 to 20 were found across the board (nicotine: antisocial peers at age 17 explained 15.5% of the predicted variance of nicotine dependence, 28% of the predicted variance in alcohol use disorder at age 20, 27% of the predicted variance in illicit drug use disorder at age 20, and 62.8% of the predicted variance in antisocial behavior at age 20), generally smaller socialization effects were found beyond age 20. For example, antisocial peers at age 17 explained <1% of the predicted variance in nicotine dependence at age 20, 11.6% of the predicted variance in alcohol use disorder at age 20, 3.3% of the predicted variance in illicit drug use disorder at age 20, and 12.1% of the predicted variance in adult antisocial behavior at age 20. This suggests that in line with expectations, socialization effects may be more relevant from age 17 to 20 compared to age 20 to 24 or 24 to 29 for each of the externalizing disorders.

Cross-Lagged Panel Results for the Latent Externalizing Disorder Factors

As results were similar across each facet of externalizing disorders, and externalizing disorders were highly correlated (as shown in Table 2), we estimated latent factors of externalizing disorders at ages 17, 20, 24, and 29. Figure 3 shows the cross-lagged panel results. Following the results for each individual externalizing disorder facet, the stability of externalizing disorders from ages 17 to 20 was smaller in effect size (β = .44, p < .001) than the stability of externalizing disorders from ages 20 to 24 (β = .83, p < .001) and 20 to 29 (β = .82, p < .001). Comparatively, the stability of antisocial peer affiliation from ages 17 to 20 was somewhat larger in effect size (β = .58, p < .001) than the stability of externalizing disorders from 17 to 20 (reported above) and relatively stable from ages 20 to 24 (β = .52 p < .001) and ages 24 to 29 (β = .48, p < .001).

Figure 3
The Prospective Relationship between Antisocial Peer Affiliation and Externalizing Disorders from Adolescence through Young Adulthood

After accounting for the stability of externalizing disorders and antisocial peer affiliation over time and the residual correlations between externalizing disorders and antisocial peer at each time point, results showed there were stronger socialization than selection effects from age 17 to 20 (as the path from antisocial peer affiliation at age 17 to externalizing disorders at age 20 was significant and moderate in magnitude and the path from externalizing disorders at age 17 to antisocial peer affiliation at age 20 was not significantly different than zero). Conversely, there were generally stronger selection than socialization effects from age 20 to 24, and from ages 24 to 29 (as the paths from externalizing disorders to subsequent antisocial peer affiliation were significant and moderate in magnitude in both cases, but the paths from antisocial peer affiliation to subsequent externalizing disorders was not significant from zero in either case; see Figure 2 for details).

Constraining these paths to be equivalent resulted in a significant decrement of fit for each developmental transition, suggesting socialization effects were significantly stronger than selection effects from 17 to 20 (Δχ2 (1) = 41.07, p < .001) and that selection effects were significantly stronger than socialization effects from 20 to 24 (Δχ2 (1) = 22.38, p < .001) and 24 to 29 (Δχ2 (1) = 28.07, p < .001). Follow-up tests showed results were consistent across males and females (see Supplementary Materials, eFigures 1-2) and across younger and older MTFS cohorts (see Supplementary Materials, eFigures 3-4). We also confirmed results were consistent when including family-of-origin socioeconomic status as a covariate (see Supplementary Materials, eFigures 5) or a moderator of these prospective relationships (see Supplementary Materials, eFigures 6).

Squaring the standardized path coefficient and dividing by the total predicted variance accounted for (obtained from Figure 1) illustrates the percentage of explained variance in each dependent variable account for by that pathway. Thus, antisocial peers at age 17 explained 25.8% of the predicted variance in externalizing disorders at age 20, a relatively large effect. Comparatively, externalizing disorders at age 17 explained less than 1% of the predicted variance in antisocial peers at age 20. Alternatively, antisocial peers at age 20 explained less than 1% of the predicted variance of externalizing disorders at age 24, whereas externalizing disorders at age 20 explained 13.5% of the predicted variance in antisocial peers at age 24. Similarly, antisocial peers at age 24 explained less than 1% of the predicted variance of the predicted variance of externalizing disorders at age 29, whereas externalizing disorders at age 24 predicted 18.4% of the total predicted variance of antisocial peers at age 29. Thus, there were relatively large socialization effects from age 17 to 20, followed by relatively moderate selection effects from ages 20 to 24 and 24 to 29.

Moderation by College Status

Table 3 shows results for the cross-lagged model by college status. At a descriptive level, socialization paths tended to be stronger in effect size for those with new college experience compared to those without new college experience for each developmental transition, including from ages 17 to 20 (bcollege = .39 vs. bno college = .24), ages 20 to 24 (bcollege = .12 vs. bno college = .05), and ages 24 to 29 (bcollege = .55 vs bno college = .01). However, constraining these paths to be equivalent across college vs. no college status did not result in a significant decrement of model fit for the age 17 to 20 transition (Δχ2 (1) = 1.02, p = .31) or the age 20 to 24 transition (Δχ2 (1) = .92, p = .34), suggesting socialization effects were similar in magnitude across college status during those developmental transitions. Only path coefficients from the age 24 to 29 transition were significantly different (Δχ2 (1) = 4.61, p = .03). Thus, hypotheses were not supported by except for the age 24 to 29 assessment, and even then caution should be warranted as only 56 people (2% of the sample) met criteria for the new college experience criteria post age 24 (suggesting it was a relatively rare occurrence).

Table 3
Prospective Associations between Externalizing Disorders and Antisocial Peer Affiliation by College Status

Moreover, across developmental transitions and for both those with any college compared to those with no college, results showed that socialization effects tended to be stronger from age 17 to 20 (bsocialization’s ranged from .24 to .29 vs. bselection‘s ranged from .04 to .10), and that selection effects tended to be stronger from age 20 to 24 (bselection‘s = .27 vs. socialization’s ranged from .05 to .10), and from 24 to 29 (bselection‘s ranged from .36 to .61 vs. bsocialization’s ranged from .01 to .55). Thus, regardless of college experience, results generally support stronger socialization than selection effects from ages 17 to 20 and stronger selection than socialization effects from ages 20 to 29. Follow-up analyses showed a similar pattern of results across each externalizing disorder facet, as well (see eTable 1 in the Supplementary Materials for details).

Discussion

The purpose of this study was to evaluate the processes of socialization and selection as they apply to the associations between antisocial peer affiliation and externalizing problems in the transition from adolescence through young adulthood. We also evaluated whether these socialization and selection processes varied as a result of college experiences. As a reminder, socialization refers to how an adolescent’s affiliation to antisocial peers predicts subsequent externalizing behavior. Selection refers to how an adolescent’s externalizing behavior predicts subsequent antisocial peer affiliation (Andrews et al., 2002; Monahan et al., 2009; Curran et al., 1997).

Following previous research (Curran et al., 1997; Kendler et al., 2007; van Ryzin, Fosco, & Dhion, 2012; van Ryzin & Dishion, 2014), after accounting for the stability of traits and correlated residual variance at each time point, we found evidence for socialization processes linking antisocial peer affiliation with externalizing disorders from age 17 to 20 and for selection processes linking externalizing disorders with antisocial peer affiliation from age 20 to 24 and 24 to 29. This was true for each of the externalizing disorder facets examined (nicotine dependence, alcohol use disorder, illicit drug use disorder, and adult antisocial behavior) and our estimated latent factor of externalizing disorders.

Contrary to our expectations that we would find significantly stronger selection than socialization effects post-adolescence, we found evidence for essentially equivalent socialization and selection effects from ages 20 to 24 and 24 to 29 for alcohol use disorder, illicit drug use disorder, and adult antisocial behavior, specifically. However, and in line with our expectations, we showed evidence of significantly stronger selection than socialization effects from age 20 to 24 and 24 to 29 using a latent measurement of externalizing disorders. These results were consistent across males and females and across different levels of socioeconomic status. Moreover, this general pattern of results was consistent for those who were in or were not in college. In total, results support developmental theory, suggesting selection effects become more important in later than in earlier developmental periods due to more freedom in environmental/social-relationship selection post-adolescence (Scarr & McCartney, 1983). However, this may be particular to a liability towards externalizing generally, rather than specific drug use and clinical antisocial behavior (see Iacono et al., 2008).

This follows other research from the Minnesota Twin and Family Study. Samek, Hicks, Keyes, McGue, and Iacono (2015) showed evidence of gene x environment interaction involving antisocial peer affiliation and a broad measure of externalizing disorders at age 17, but no such evidence at ages 20, 24, or 29. The broad externalizing measure included several substance use disorders as well as the adult antisocial behavior symptom count, similar to this study. Samek et al. showed the genetic influence on adolescent externalizing disorders was greater under the context of a greater degree of antisocial peer affiliation and smaller under the context of a lesser degree of antisocial peer affiliation in adolescence.

This generally supports a socialization hypothesis in that antisocial peers appear to activate genetic risk for externalizing problems – but only in adolescence. That is because no such genetic moderation was found for the cross-sectional associations at ages 20, 24, or 29, or the longitudinal associations between adolescent antisocial peer affiliation and adult externalizing disorders (Samek et al., 2015). However, subsequent analyses by Samek and colleagues confirmed that there was unique evidence for genetic moderation for several substance use disorders rather than the broader measurement of externalizing problems. In combination, these studies suggest important socializing influence of adolescent antisocial peers that seem to dissipate over time, especially in regards to broad externalizing problems. On the other hand, these studies suggest some continued socializing influence of antisocial peers on individual (substance and behavior specific) externalizing problems.

The present study’s findings are somewhat inconsistent with results reported by Monahan et al. (2009). Monahan and colleagues found evidence of both selection and socialization from 14 to 15, did not find evidence for selection past age 16, and did not find evidence for either selection or socialization from age 20 to 22. Potential reasons for the discrepancy in the Monahan et al. study and this one was that the Monahan et al. study included individuals who were adjudicated of serious crime and the sample was made up of predominately males. Using a large community sample of males and females, we have shown that selection appears to still be operating in a meaningful and developmental pattern through age 29.

It is important to note that in this study, the stability of externalizing disorders was smaller in effect size from ages 17 to 20 than from ages 20 to 24 and 20 to 29. This follows earlier research demonstrating marked instability from adolescence to early adulthood in antisocial behavior and substance use disorders (Loeber et al., 2012; Moffitt, 1993; Moffitt & Caspi, 2001), including marked increases substance use disorders in early adulthood relative to adolescence (SAMSHA, 2014). Comparatively, the stability of antisocial peer affiliation from ages 17 to 20 was generally similar in effect size to the stability of antisocial peer affiliation from ages 20 to 24 and ages 24 to 29. This suggests that adolescents who affiliate with antisocial peers in adolescence are to some extent more likely to continue to affiliate with antisocial peers as they transition into early adulthood (i.e., age 20) and in later young adulthood (i.e., age 24 and 29). Nonetheless, even with relatively stable antisocial peers, our findings suggest the relative socializing influence of antisocial peers appears to decrease across time. This again follows studies a general marked desistence in antisocial behavior and substance use generally as people become older (Loeber et al., 2012; SAMSHA, 2014).

Future Directions

In addition to replicating findings in other community and “at-risk” samples, additional future directions include an analysis of other aspects of socialization and selection that may be relevant to externalizing disorders in young adulthood, such as romantic relationships experiences and romantic relationship affiliations. There is some evidence of this as time spent with friends and peer groups decreases and time spent with romantic partners increases in early adulthood (Reis et al., 1993). Following this, van Dulmen, Goncy, Haydon, and Collins (2008) indicated that romantic relationship security was more strongly and inversely related to externalizing problems in early adulthood (ages 23-24) relative to adolescence (age 16). This suggests that romantic partner characteristics, rather than peers, may be more relevant to adult externalizing disorders.

This study is not without limitations. As this sample was representative of Minnesota for the birth cohorts sampled, it is predominately white. Nonetheless, there was considerable diversity in family-of-origin socioeconomic status and the percentages of those that went to versus did not go to college. It is useful to reiterate that antisocial peer affiliation and externalizing disorders were evaluated at ages 17 and through 29; likely these constructs were impacted by earlier antisocial peer and externalizing problems (i.e., prior to age 17) not captured in the modeling used here. This is supported by prior research, which has evidenced both socialization and selection in earlier adolescent periods (e.g., Curran et al., 1997; van Ryzin & Dishion, 2014). Moreover, there is research to support the notion that peer influence may be even more important to adolescent development in early rather than middle or late adolescence (Albert, Chein, & Steinberg, 2013; Gardner & Steinberg, 2005; Sumter, Bokhorst, Steinberg, & Westenberg, 2009). It will also be important for future research to take advantage of multiple measurements of antisocial peer affiliation to determine whether that may have impacted results. Finally, we were unable to evaluate for variation by type of college experience (e.g., some college experience vs. finished degree) as the majority of cases who met the new college experience criteria evaluated in this study had completed only “at least some college” (92%-100% at ages 20, 24, and 29). It remains important for future studies to better explore this and determine whether there may be specific reasons to expect one college type to have a stronger socialization effect than another. There are also important strengths of this study, including the large sample size, inclusion of both males and females, and the use of clinical interviews to assess externalizing disorders, as well as the longitudinal design, and incorporation of prospective data analysis.

Conclusion

This study showed support for socialization as the predominant mechanism to explain the link between antisocial peer affiliation and a broad measure of externalizing disorders in the developmental transition from adolescence to early adulthood (i.e., age 17 to 20), but past this developmental transition, selection was more important in explaining the prospective links between antisocial peer affiliation and this broad measure of externalizing disorders (i.e., ages 20 to 24 and 24 to 29). Results were generally consistent across males and females and regardless of college status, suggesting these processes are not context-specific to college experience. On the other hand, there was more support for smaller socialization and selection influences beyond age 20 for each individual externalizing disorder (e.g., substance use disorder and adult antisocial behavior). More research is needed to evaluate the socialization vs. selection impact of other mechanisms linked to externalizing disorders in adolescence and young adulthood, including romantic relationship affiliation and experiences.

Supplementary Material

supp

Acknowledgements

This research was supported by Grant DA05147, DA034606, and DA024417 from the National Institute on Drug Abuse and Grant AA09367 from the National Institute on Alcohol Abuse and Alcoholism. The first author was also supported by the USDA National Institute of Food and Agriculture, Hatch project 1006129. The content is solely the responsibility of the authors and does not necessarily represent the official views of the aforementioned funding agencies.

References

  • Albert D, Chein J, Steinberg L. The teenage brain: Peer influences on adolescent decision making. Current Directions in Psychological Science. 2013;22:114–120. [PMC free article] [PubMed]
  • American Psychological Association . Diagnostic and Statistical Manual of Mental Disorders. 5th Author; Washington, DC: 2013.
  • Andrews JA, Tildesley E, Hops H, Li F. The influence of peers on young adult substance use. Health Psychology. 2002;21:349–357. doi: 10.1037//0278-6133.21.4.349. [PubMed]
  • Berrington A, Smith P, Sturgis P. An overview of methods for the analysis of panel data. ESRC National Centre for Research Methods. 2006 Retrieved from http://eprints.ncrm.ac.uk/415/1/MethodsReviewPaperNCRM-007.pdf.
  • Brendgen M, Vitaro F, M. Bukowski W. Deviant friends and early adolescents' emotional and behavioral adjustment. Journal of Research on Adolescence. 2000;10:173–189.
  • Burt SA, McGue M, Iacono WG. Nonshared environmental mediation of the association between deviant peer affiliation and adolescent externalizing behaviors over time: Results from a cross-lagged monozygotic twin differences design. Developmental Psychology. 2009;45:1752–1760. doi: 10.1037/a0016687. [PMC free article] [PubMed]
  • Curran PJ, Stice E, Chassin L. The relation between adolescent alcohol use and peer alcohol use: A longitudinal random coefficients model. Journal of Consulting and Clinical Psychology. 1997;65:130–140. [PubMed]
  • Dishion TJ, Owen LD. A longitudinal analysis of friendships and substance use: bidirectional influence from adolescence to adulthood. Developmental Psychology. 2002;38:480–491. doi: 10.1037//0012-1649.38.4.480. [PubMed]
  • Dishion TJ, Véronneau MH, Myers MW. Cascading peer dynamics underlying the progression from problem behavior to violence in early to late adolescence. Development and Psychopathology. 2010;22:603–619. doi:10.1017/S0954579410000313. [PubMed]
  • Esser MB, Hedden SL, Kanny D, Brewer RD, Gfroerer JC, Naimi TS. Prevalence of alcohol dependence among US adult drinkers, 2009–2011. Preventing Chronic Disease: Public Health Research, Practice and Policy. 2014;11(E206):1–11. doi: 10.5888/pcd11.140329. [PMC free article] [PubMed]
  • Gardner M, Steinberg L. Peer influence on risk taking, risk preference, and risky decision making in adolescence and adulthood: an experimental study. Developmental Psychology. 2005;41:625–635. doi: 10.1037/0012-1649.41.4.625. [PubMed]
  • Hicks BM, South SC, DiRago AC, Iacono WG, McGue M. Environmental adversity and increasing genetic risk for externalizing disorders. Archives of General Psychiatry. 2009;66:640–648. doi:10.1001/archgenpsychiatry.2008.55. [PMC free article] [PubMed]
  • Iacono WG, Carlson SR, Taylor J, Elkins IJ, McGue M. Behavioral disinhibition and the development of substance-use disorders: Findings from the Minnesota Twin Family Study. Development and Psychopathology. 1999;11:869–900. doi: 10.1017/S0954579499002369. [PubMed]
  • Iacono WG, Malone SM, McGue M. Behavioral disinhibition and the development of early-onset addiction: Common and specific influences. Annual Review of Clinical Psychology. 2008;4:325–348. doi: 10.1146/annurev.clinpsy.4.022007.141157. [PubMed]
  • Kendler KS, Jacobson KC, Gardner CO, Gillespie N, Aggen SA, Prescott CA. Creating a social world: A developmental twin study of peer-group deviance. Archives of General Psychiatry. 2007;64:958–965. [PMC free article] [PubMed]
  • Kenny DA, Harackiewicz JM. Cross-lagged panel correlation: Practice and promise. Journal of Applied Psychology. 1979;64:372–379. doi: 10.1037/0021-9010.64.4.372.
  • Legrand LN, Keyes M, McGue M, Iacono WG, Krueger RF. Rural environments reduce the genetic influence on adolescent substance use and rule-breaking behavior. Psychological Medicine. 2008;38:1341–1350. doi: 10.1017/S0033291707001596, [PMC free article] [PubMed]
  • Loeber R, Menting B, Lynam DR, Moffitt TE, Stouthamer-Loeber M, Stallings R, et al. Findings from the Pittsburgh Youth Study: Cognitive impulsivity and intelligence as predictors of the age-crime curve. Journal of the American Academy of Child and Adolescent Psychiatry. 2012;51:1136–1149. [PubMed]
  • Moffitt TE. Adolescence-limited and life-course-persitent antisocial behavior: A developmental taxonomy. Psychological Review. 1993;100:674–701. [PubMed]
  • Moffitt TE, Caspi A. Childhood predictors differentiate life-course persistent and adolescence-limited antisocial pathways among males and females. Development and Psychopathology. 2001;13:355–375. [PubMed]
  • Monahan KC, Steinberg L, Cauffman E. Affiliation with antisocial peers, susceptibility to peer influence, and antisocial behavior during the transition to adulthood. Developmental Psychology. 2009;45:1520–1530. doi: 10.1037/a0017417. [PMC free article] [PubMed]
  • Muthén LK, Muthén BO. Mplus user’s guide. 7th Muthén & Muthén; Los Angeles, CA: 1998–2014.
  • National Institute on Alcohol Abuse and Alcoholism Alcohol Use Disorder: A Comparison Between DSM–IV and DSM–5. NIH Publication No. 13–7999. 2013 Retrieved June 22, 2015, from http://pubs.niaaa.nih.gov/publications/dsmfactsheet/dsmfact.pdf.
  • Reis HT, Yi-Cheng L, Bennett ME, Nezlek JB. Change and consistency in social participation during early adulthood. Developmental Psychology. 1993;29:633–645.
  • Robins LN, Wing J, Wittchen HU, Helzer JE, Babor TF, Burke J, Towle LH. The Composite International Diagnostic Interview: An epidemiologic instrument suitable for use in conjunction with different diagnostic systems and in different cultures. Archives of General Psychiatry. 1988;45:1069–1077. [PubMed]
  • Samek DR, Hicks B, Keyes MA, McGue M, Iacono WG. Antisocial peer affiliation and externalizing disorders: Evidence for gene × environment × development interaction. Development and Psychopathology. 2015 Manuscript under revision at.
  • Samek DR, Keyes MA, Iacono WG, McGue M. Peer deviance, alcohol expectancies, and adolescent alcohol use: Explaining shared and nonshared environmental effects using an adoptive sibling pair design. Behavior Genetics. 2013;43:286–296. doi: 10.1007/s10519-013-9595-9. [PMC free article] [PubMed]
  • Substance Abuse and Mental Health Association (SAMSHA) NSDUH Series H-48, HHS Publication No. (SMA) 14-4863. Rockville, MD: 2014. Results from the 2013 national survey on drug use and health: Summary of national findings. Retrieved from http://www.samhsa.gov/data/sites/default/files/NSDUHresultsPDFWHTML2013/Web/NSDUHresults2013.pdf.
  • Scarr S, McCartney K. How people make their own environments: A theory of genotype→ environment effects. Child Development. 1983;23:424–435. [PubMed]
  • Schulenberg JE, Maggs JL. A developmental perspective on alcohol use and heavy drinking during adolescence and the transition to young adulthood. Journal of Studies on Alcohol and Drugs. 2002;(14):54–70. [PubMed]
  • Schulenberg JE, Patrick ME. Historical and developmental patterns of alcohol and drug use among college students: Framing the problem. In: White HR, Rabiner D, editors. College Drinking and Drug Use. Guilford; New York: 2012. pp. 13–35.
  • Sher KJ, Gotham HJ. Pathological alcohol involvement: A developmental disorder of young adulthood. Development and Psychopathology. 1999;11:933–956. [PubMed]
  • Slutske WS. Alcohol use disorders among US college students and their non–college-attending peers. Archives of General Psychiatry. 2005;62:321–327. doi: 10.1001/archpsyc.62.3.321. [PubMed]
  • Spitzer RL, Williams JBW, Gibbon M. Structured Clinical Interview for DSM-III-R (SCID) New York State Psychiatric Institute, Biometrics Research; New York: 1987.
  • Stone AL, Becker LG, Huber AM, Catalano RF. Review of risk and protective factors of substance use and problem use in emerging adulthood. Addictive Behaviors. 2012;37:747–775. doi: 10.1016/j.addbeh.2012.02.014. [PubMed]
  • Substance Abuse and Mental Health Administration . NSDUH Series H-48, HHS Publication No. (SMA) 14-4863. Substance Abuse and Mental Health Servicies Administration; Rockville, MD: 2014. Results from the 2013 National survey on drug use and health: Summary of national findings.
  • Sumter SR, Bokhorst CL, Steinberg L, Westenberg PM. The developmental pattern of resistance to peer influence in adolescence: Will the teenager ever be able to resist? Journal of Adolescence. 2009;32:1009–1021. [PubMed]
  • U.S. Census Bureau Poverty thresholds 1996. 1996 Retrieved from https://www.census.gov/hhes/www/poverty/data/threshld/thresh96.html.
  • Van Dulmen MH, Goncy EA, Haydon KC, Collins WA. Distinctiveness of adolescent and emerging adult romantic relationship features in predicting externalizing behavior problems. Journal of Youth and Adolescence. 2008;37:336–345.
  • van Ryzin MJ, Dishion TJ. Adolescent deviant peer clustering as an amplifying mechanism underlying the progression from early substance use to late adolescent dependence. Journal of Child Psychology and Psychiatry. 2014;55:1153–1161. doi:10.1111/jcpp.12211. [PMC free article] [PubMed]
  • Van Ryzin MJ, Fosco GM, Dishion TJ. Family and peer predictors of substance use from early adolescence to early adulthood: An 11-year prospective analysis. Addictive Behaviors. 2012;37:1314–1324. doi:10.1016/j.addbeh.2012.06.020. [PMC free article] [PubMed]
  • Walden B, McGue M, Iacono WG, Burt SA, Elkins I. Identifying shared environmental contributions to early substance use: the respective roles of peers and parents. Journal of Abnormal Psychology. 2004;113:440–450. DOI: 10.1037/0021-843X.113.3.440. [PubMed]
  • White HR, Labouvie EW, Papadaratsakis V. Changes in substance use during the transition to adulthood: A comparison of college students and their noncollege age peers. Journal of Drug Issues. 2005;35:281–306.