ABSTRACT
This paper reports results from a three-year research project that followed 654 Federal male prisoners released during the period 1973-1993 who had been students in the post-secondary prison education program that operated in British Columbia Federal Prisons during those years. Utilizing the Statistical Information on Recidivism (SIR) risk prediction scheme developed for the Correctional Service of Canada by Nuffield (1982), the research seeks to discover which sub-groups of the sample did better than their predicted rate of recidivism. The sub-groups are constructed from over 60 variables compiled during extensive reviews of offender files, university records, and staff interviews. The three-year follow-up after release shows that while the entire group of prisoner-students did significantly better than was predicted by SIR (75% success compared to a SIR-predicted 58%), certain sub-groups “beat SIR” by even greater margins while others merely matched their prediction. Various hypotheses are explored in an attempt to account for the success and the variations in success.