Comparing risk instruments’ predictive strength

Based on comparative analyses, where available, Statistical Information on Recidivism – Revised 1 (SIR-R1) and reintegration potential ratings should be prioritized above static factor risk when considering an offender’s likelihood of successfully remaining in the community. This information will be useful to Correctional Service of Canada (CSC) in its continued efforts to maximize efficiencies and accuracy with respect to offender risk assessment.

For non-Aboriginal male offenders (the only group for whom the SIR-R1 is completed), both the SIR-R1 and reintegration potential were strongly associated with returns to custody, both with a new offence and overall, after two years of community follow-up. Static factor risk did not attain satisfactory levels of predictive accuracy.

For all offenders, reintegration potential was a better predictor of community success than static factor risk. The predictive strength of all three instruments was stronger for non-Aboriginal male offenders than for other offender groups.

Risk Instruments’ Associations with Returns to Custody

Note.SIR-R1 = Statistical Information on Recidivism – Revised 1. Associations were measured using Areas Under the Curve (AUCs), measures of the ability of an instrument to predict an outcome of interest. Possible results range between .50 and 1.00, with .60 being considered the minimal acceptable value and .70 generally considered satisfactory.

CSC uses a number of risk instruments to predict offenders’ post-release outcome. However, little is known about their relative predictive strength. This comparative examination focused on the association of three common measures with returns to custody both with an offence and overall.

Offenders released between April 2006 and March 2008 were followed for two years. Of interest was the association between various risk instrument ratings and whether offenders returned to custody either due to a new offence or for any reason (including technical violations, outstanding offences, and new offences). A total of 3,248 offenders who were under community supervision for the full two years were included.

The following risk instruments were examined:

1. SIR-R1: estimates the likelihood of committing an indictable offence within three years of release (available for male non-Aboriginal offenders only).
2. Static Factor Risk: estimates risk based on a structured professional assessment of factors associated with recidivism.
3. Reintegration Potential: estimates the likelihood of an offender successfully reintegrating into the community. It is a composite measure calculated based on, for non-Aboriginal male offenders, scores on a security classification measure, the SIR-R1, and static factor risk. For other offenders, scores on the security classification measure, static factor risk, and dynamic factor risk (i.e., need) are used. However, it can be changed by parole officers during the sentence. The assessment closest to release was used in analyses.

Please e-mail the Research Branch or contact us by phone at (613) 995-3975.

You can also visit the website for a full list of research publications.

Prepared by: Renée Gobeil and Trina K. Forrester