Premature autonomy describes a developmental dynamic where parents of high‐risk adolescents reduce their involvement and guidance when confronted with challenges of problem behaviour and the influence of deviant friendships. This dynamic was tested on the sample of Oregon Youth Study boys (N=206), whose family management practices and friendships were observed on videotaped interaction tasks. Latent growth curve models were used to examine longitudinal trends between deviant friendship interactions and family management. Direct observations of deviant friendship process at age 14 were associated with degradation in family management during adolescence. A comparison of antisocial and well‐adjusted boys clarified that parents of antisocial boys (started early and persisted) decreased family management around puberty, in comparison to parents of well‐adjusted boys who maintained high levels of family management through adolescence. In predicting late adolescent problem behaviour, there was a statistically reliable interaction between family management degradation and deviant peer involvement in adolescence in support of the premature autonomy hypothesis. Adolescent males involved in deviant friendships, and whose parents decreased their family management, were most likely to use marijuana and commit antisocial acts at age 18. The implications for interventions that target adolescents are discussed.

A dynamic systems framework was applied to understand the influence of friendship on antisocial behavior from childhood (age 9-10) through adulthood (age 24-25) for Oregon Youth Study males (N = 206). Boys were videotaped interacting with a friend at ages 14, 16, and 18, and deviant content and interpersonal processes were independently coded. Conditional dyadic interpersonal processes were studied as a communication system and summarized by an index of information entropy (F. Attneave, 1959). High entropy scores represent disorganized, unpredictable patterns of interaction, whereas low entropy scores reflect an organized dialogue. Conversations of early-onset antisocial boys and their best friends were less organized and included more deviant content than those of well-adjusted controls. Prediction analyses, however, revealed an interaction between entropy and deviant talk. Consistent with expectation, males with well-organized interactions (i.e., low entropy) but elevated levels of deviant content were most likely to continue antisocial behavior into adulthood. Findings suggest that individual risk for maladaptation may be amplified by early adolescent friendship dynamics organized around deviance.

This paper reviews the model-building approach to developmental psychopathology as it applies to understanding the influence of peers on adolescent problem behavior. Considerable longitudinal evidence suggests that deviant peer involvement is prognostic of escalations in various forms of problem behavior, including substance use, delinquent behavior, and violence. Most convincing, however, are data from randomized intervention trials that show alarming iatrogenic effects for aggregating high-risk peers. This paper shows that intervention research can also clarify the role of parents in reducing deviant peer involvement. Results from a 4-year prevention trial are reported for a community sample of multiethnic middle school youth. Random assignment to family-centered services in a school Family Resource center produced reduced growth in deviant peer involvement. The extensiveness of the parenting intervention was correlated with reduced growth in deviant peer involvement, especially among high-risk youth. Findings are discussed with respect to developmental theory and the design of future intervention trials that clarify the role of adults in structuring the emerging adolescents' world of peers.

This study used latent transition analysis (LTA) to examine changes in early emotional and behavioral problems in children age 2 to 4 years resulting from participation in a family-centered intervention. A sample of 731 economically disadvantaged families was recruited from among participants in a national food supplement and nutrition program. Families with toddlers between age 2 and 3 were randomized either to the Family Check-Up (FCU) or to a nonintervention control group. The FCU's linked interventions were tailored to each family's needs. Assessments occurred at age 2, 3, and 4. The FCU followed age 2 and age 3 assessments. Latent class analyses were conducted on mother reports of behavior and emotional problems from age 2 to 4 to study transitions among the following four groups: (a) externalizing only, (b) internalizing only, (c) comorbid internalizing and externalizing, and (d) normative. LTA results revealed that participation in the FCU increased the likelihood of transitioning from either the comorbid or the internalizing class into the normative class by age 4. These results suggest family interventions in early childhood can potentially disrupt the early emergence of both emotional and behavioral problems.