Prediction of Recidivism Among Federally Sentenced Offenders: A Re-Validation of the SIR (Statistical Information on Recidivism) Scale

• NCJ Number: 160106
• Journal: Canadian Journal of Criminology Volume: 38 Issue: 1 Dated: (January 1996) Pages: 61-79
• Author(s): J Bonta; W G Harman; R G Hann; R B Cormier
• Date Published: 1996
• Length: 19 pages
• Publication Format: Article
• Publication Type: Measurement/Evaluation Device
• Language: English
• Country: Canada

Annotation

This study conducted a revalidation of the Statistical Information on Recidivism Scale (SIR) in Canada, along with an examination of the scale's ability to predict violent recidivism.

Abstract

The assessment of offender risk plays an important role in corrections, both in terms of its influence on program allocation and promoting the safe release of offenders into the community. Paroling authorities are particularly sensitive to offender risk, and risk assessment forms a critical part of their decision to release an offender. Objective offender risk-assessment instruments have been developed to assist in the release decisionmaking process and, in Canada, one such instrument is the SIR Scale, which was developed in the early 1970's. The re- validation of the SIR Scale used a sample of 3,267 inmates released from Federal penitentiaries during the fiscal year 1983- 84. Information was collected on a range of general personal and social demographic factors and offense data required to calculate SIR scores. Criminal records were provided by the RCMP Fingerprint Service and the Correctional Service of Canada. This information was used for coding recidivism. General recidivism was defined as a custodial admission within 3 years following release; this included new indictable offenses and revocation of supervision; two measures of violent recidivism were adopted. The narrow definition of violence included homicide, sexual assault, and aggravated assault. Findings show that the validity of the SIR Scale for the prediction of general recidivism was reaffirmed. Receiver Operating Characteristic (ROC) analysis also confirmed the current use of cut-off scores that define the various risk categories. Although SIR scores predicted violent reoffending, the magnitude of the relationship was less than that reported with other violent risk-predicting instruments. Problems and options for improving predictions of violent reoffending are discussed, with an emphasis on practical applications. 4 tables, 1 figure, and 24 references