Recidivism Risk and the Impact of Prison-Based Programs for Women: A Survival Analysis

• In Event: Poster Session 2

Abstract

Recidivism is a significant public concern, and prison-based programs to deter reoffending have been a focus for scholars. However, predictors for women's recidivism may be overlooked. This study examined women's recidivism trends, emphasizing the impact of prison-based programs. Using a sample of 8,393 women with a recidivism rate of 9.22%, this research employs the random forest method for feature selection in R. It identifies ten types of prison-based programs as potentially influential. Discrete-time hazard models are applied to estimate the trend in the recidivism risk. This analysis differentiates between two recidivism types: technical violations and committing new crimes. Results reveal that the hazard probability for technical violation peaks between the sixth and twelfth months post-release before declining. The hazard probability of committing new crimes increases over time, with a sharp rise in the first eighteen months after release. Further analysis incorporating predictors indicates that race, age at offense, duration of imprisonment, and offense count are associated with women’s recidivism risk. Importantly, participating in re-entry plans and job programs has been found to significantly reduce the risk of committing new crimes for women. This research highlighted the need to address unique predictors of recidivism among female offenders.

Author

• Sian Zhu, University of Massachusetts Lowell