This study examines how multivariate trajectory patterns of overt and relational peer and dating violence perpetration, alcohol use, and nonviolent deviant behavior during high school predict convictions in adulthood. Adolescent data are from an accelerated cohort design study that spanned four waves in 2003–2005. In 2019, conviction records were obtained for a random subsample of 1,579 individuals from the original study. We identified latent classes that were jointly characterized by distinct behavioral trajectories and adult conviction status, and described the demographic and psychosocial profiles of each class. The best-fitting model comprised four trajectory classes: Low Deviance (44%), Moderate Stable Deviance (40%), Increasing Deviance (8%), and Dating Violence Perpetrators (8%). Adolescents whose deviance increased during adolescence had substantially higher risk of convictions, including violent convictions, than all other groups. Classes were differentiated by gender, household structure, parental education, school bonding, grades, emotional dysregulation, sensation-seeking, family conflict, and prosocial values. The Increasing Deviance class was predominantly male, had an elevated probability of coming from a single-parent household and of having parents with low education, but values on psychosocial indicators were not extreme. Dating Violence Perpetrators were also more likely to come from a single-parent household, but their parents tended to have more education. This group was the most extreme on several psychosocial indicators that indicate low school and family bonds, and poor emotional regulation. The implications of these patterns in relation to interactional and strain theories, theories of cognitive maturation, and theories of social bonds and social control are discussed.
Household structure and urban opportunities: Evaluating differences in the accessibility to jobs, education and leisure in São Paulo
