The marginal effect of parole supervision on recidivism
Dr Evarn Ooi1, Dr Joanna Wang2
1Bureau Of Crime Statistics And Research
2University of Technology Sydney
AIM: To estimate the causal effect of parole supervision on recidivism among offenders sentenced to short-term prison sentences.
METHOD: Recidivism is compared between parolees and inmates released from prison unconditionally. To measure the causal effect of parole supervision, this study uses the variation in the sentencing leniency of quasi-randomly assigned Local Court magistrates as an instrument of release on parole. This magistrate leniency measure is used as an instrumental variable (IV) in a two-stage least squares (2SLS) model to measure the local average treatment effect (LATE) of parole supervision on recidivism. Using this methodology, the study compares three recidivism outcomes. These are the probability of re-conviction; probability of committing a personal, property, or serious drug offence; and probability of re-imprisonment.
RESULTS: A range of tests suggest that the magistrate leniency measure satisfies the criteria for a valid instrument. The IV estimates reveal that parolees are substantially less likely to re-offend than inmates released unconditionally. The main results show that parolees are 10.0 percentage points less likely to be re-convicted, 10.3 percentage points less likely to commit a personal, property, or serious drug offence and 5 percentage points less likely to be re-imprisoned within 12 months of release compared with other ex-inmates. These reductions in recidivism are statistically significant and persist after 24 months after release from prison. Additional findings reveal statistically significant reductions in recidivism among parolees with a Level of Service Inventory – Revised (LSI-R) score of Medium or above and below Medium, and Aboriginal and non-Aboriginal parolees.
CONCLUSION: Parolees are substantially less likely to re-offend than inmates released unconditionally and the reduction in recidivism persists after 24 months of release from prison.
Dr. Joanna Wang is a Senior Lecturer in School of Mathematical and Physical Sciences at UTS. She was previously a Research Statistician at BOCSAR. Her research interest include using various statistical and econometric models to analyse crime data and to assess the effectiveness of Justice programs and policy.