Modeling Long-term Criminal Careers

Abstract

The challenge for criminologists interested in studying change after the criminal career volume was fairly clear—create models that could capture the obvious heterogeneity that existed across the population, describe changes in offending over age, and capture the apparent intermittency in the data in offending frequency for individuals. In our review, we find that the current most commonly used models have, in fact, done a good job of capturing heterogeneity and describing changes in latent rates with age. However, there has been little apparent attention paid to the issue of intermittency, despite the prominence of the issue in the original criminal careers volume. This had led to an overall vision of change in the literature that has “smoothed out” or flattened the actual paths described in this research. In the example provided in this article, using administrative data from New York, we show that there appear to be uneven periods of time between arrests even for people with fixed number of arrests in a short period of time, an observation that echoes back to the original volume. We conclude with new ideas for more fully studying dynamic change and a warning about models that confuse and confound involvement in the criminal justice system and offending itself.