| This dissertation concerns criminal careers, i.e., criminal behavior by individuals over the life course, and the question of whether it is fair to assume that behavioral trajectories fall into discrete trajectory classes, as some popular group-based models assume. I review the literature on the age-crime relationship, longitudinal research and developmental criminology, and I make some basic technical points about finite mixture models and growth mixture models. I also review several publications citing methodological and theoretical problems with the assumption of discrete groups in criminal trajectories.;Using the Cambridge Study in Delinquent Development, a cohort of 411 boys followed from age 8-10 to age 50, I conduct an analysis using an alternative method to obtain estimates of trajectories and analyze the whole distribution of these trajectories. This alternative consists of a combination of some conventional methods commonly available to applied researchers and assumes a continuous distribution of estimated trajectory parameters. I find some evidence of groups in the trajectory distributions, but not as many as in prior research on criminal trajectories, even those based on the same data source. These findings are confirmed by findings from a series of finite mixture models estimated with the same data. |
