Analyzing the Presence and Consequences of Unobserved Heterogeneity in Recidivism Research

This study extends research on recidivism by examining the phenomenon via a survival mixture modeling approach, a method that is analogous to mixture modeling approaches used in trajectory analyses. Using this approach, the authors discover that multiple recidivism profiles are identifiable within a random sample of inmates released from Florida prisons between January 1998 and June 2001. Findings revealed that certain covariates predicted class membership and operated differently across the groups in predicting the hazard of recidivism, suggesting that a unilateral approach to reducing recidivism risk is an ineffective strategy. The Discussion section presents the findings in the context of theory, research, and policy.