IS AGE IMPORTANT? TESTING A GENERAL VERSUS A DEVELOPMENTAL THEORY OF ANTISOCIAL BEHAVIOR*

We tested competing hypotheses derived from Gottfredson and Hirschis (1990) general theory and Moffitt's (1993a) developmental theory of antisocial behavior. The developmental theory argues that different factors give rise to antisocial behavior at different points in the life course. In contrast, the general theory maintains that the factor underlying antisocial behavior (i.e., criminal propensity) is the same at all ages. To test these competing predictions, we used longitudinal data spanning from age 5 to age 18 for the male subjects in the Dunedin Multidisciplinary Health and Development Study. Using reports from three sources (parents, teachers, and the boys themselves), we estimated second-order confirmatory factor models of antisocial behavior. These models provided consistent support for the developmental theory, showing that separate latent factors underlie childhood and adolescent antisocial behavior. Moreover, we found that these childhood and adolescent factors related in ways predicted by Moffitt's developmental theory to four correlates of antisocial behavior: Childhood antisocial behavior was related more strongly than adolescent antisocial behavior to low verbal ability, by per activity, and negative/impulsive personality, whereas adolescent antisocial behavior was related more strongly than childhood antisocial behavior to peer delinquency. The two underlying latent factors also showed the predicted differential relations to later criminal convictions: Childhood antisocial behavior was significantly more strongly associated with convictions for violence, while adolescent antisocial behavior was significantly more strongly associated with convictions for nonviolent offenses.     

Age Adjustment and Recall Bias in the Analysis of Domestic Violence Data: Methodological Improvements Through the Application of Survival Analysis Methods

This methodological paper presents the utility of survival analysis methods to provide age adjustment in the analysis of domestic violence data. These methods improve the estimation of lifetime probability of domestic violence, improve identification of patterns of first victimization over the lifespan, and provide methods of testing risk factors for first victimization while adjusting for the respondents' age. Most importantly, these methods allow a new investigation of recall bias. Results suggest that lifetime probability of abuse may have been substantially underestimated in previous studies because of problems in recall/disclosure encountered by middle-aged women.