Deviance as an antecedent and consequence of early transitions to adulthood: mediating effects and moderating conditions

Drawing from concepts in criminological literature and sociological life-course perspective literature, data from adolescent and young adult measurements collected as part of a longitudinal panel study conducted on a cohort from Harris County, Texas, were used to estimate Structural Equation Models, testing the unmediated and mediated relationships between adolescent deviance, early timing of transitions to adult roles, and adult deviance. First, a simplified three latent variable model was estimated using the full sample (N= 3,379) to examine direct associations among adolescent deviance, early transitions to adulthood, and adult deviance while controlling for prior involvement in deviant behavior in adolescence. An expanded seven latent variable model was then estimated adding mechanisms in adolescence through which the relationships previously observed in the simplified model are mediated. Lastly, both the simplified and expanded models were estimated on eight subgroups in the sample to examine whether the relationships observed for the full sample are moderated by gender, race/ethnicity, paternal level of education, and expectations for future failure in conventional adolescent roles. For the full sample, the simplified model produced significant direct relationshipsbetween adolescent deviance and early transitions to adulthood, and between early transitions to adulthood and adult deviance. When this simplified model was estimated on the eight subgroups, the first relationship remained stable for each of the eight moderating subgroups, while the second relationship did not. When several intervening variables were added between adolescent deviance and early transitions to adulthood in the expanded model, the parameters added to the model using the intervening variables formed a chain of significant direct relationships fully mediating the relationship between adolescent deviance and early transitions to adulthood for the full sample. This chain of significant direct relationships remained stable for five of the eight subgroups, and the three subgroups that did not experience full mediation underwent great attenuation of the relationship. These intervening variables offer avenues for altering the trajectory of behavior seen in the simplified model.