Author Efty Stavrou, Suzanne Poynton and Don Weatherburn
Published July 2016
Report Type Crime and Justice Bulletin No. 194
Subject Parole; Recidivism / Re-offending; Statistical methods and modelling
Keywords re-offending, parole, time-varying covariate

Download this publication



To determine whether recidivism was associated with parole release authority; and to determine whether re-offending was also related to being under supervision or not.


Time to first proven re-offence was examined for 1,644 matched offenders who served between 18 and 36 months in custody using Cox proportional hazard regression. Being on-parole was included as a time-varying covariate in the overall model. Estimates of those who re-offended overall and on-parole or off-parole were calculated using Kaplan-Meier estimates as separate outcomes.


One in four offenders re-offended whilst on-parole and 20 per cent re-offended after their parole period had expired. Parolees released by the court were significantly more likely to re-offend overall and when offenders were offparole (when separately examined). Although the overall model showed that offenders were less likely to re-offend when on-parole, this was not statistically significant. No difference was seen in the type of re-offences committed between court-released and SPA parolees.


SPA released parolees appear to be less likely to re-offend, particularly once the parole order has expired. This may be due to the selective processes of the SPA in choosing who should be granted parole or because SPA parolees are more motivated to participate in rehabilitation programs whilst in custody.

Download this publication