NCJ Number
195643
Journal
Criminal Justice and Behavior Volume: 29 Issue: 3 Dated: June 2002 Pages: 235-249
Date Published
2002
Length
15 pages
Annotation
This article compares the accuracy of several instruments used to predict the recidivism rate for offenders upon release from prison.
Abstract
The authors compared the predictive value of 10 risk measurement instruments in their assessment of the risk of general and violent recidivism among a sample of Federal offenders. The participants included 106 male offenders who were referred to the study because their parole officers considered them to be at high risk for re-offending. Their crimes varied and included robbery, assault, murder, sex offenses, weapons offenses, and drug-related offenses. All offenders were scored on the 10 risk measurement instruments as to how great of a threat they posed in terms of re-offending. The participants were then followed for an average of 713 days, with the maximum follow-up time being 2,160 days. During the follow-up time, 28 offenders recidivated nonviolently while 34 recidivated violently. Results of statistical tests, including a stepwise multiple regression analysis, showed that two of the instruments, the GSIR-R and the VRAG, were more accurate in predicting violent recidivism than the other eight instruments. The authors note that all of the instruments they examined consisted entirely of static-risk items; therefore, future research should focus on the predictive value of dynamic factors. Tables, notes, references