Do Standard Risk Prediction Instruments Apply to Female Parolees?

• NCJ Number: 195838
• Journal: Women & Criminal Justice Volume: 13 Issue: 2/3 Dated: 2002 Pages: 163-182
• Author(s): Marilyn D. McShane; Frank P. Williams III; H. Michael Dolny
• Date Published: 2002
• Length: 20 pages
• Publication Format: Article
• Publication Type: Test/Measurement
• Language: English
• Country: United States of America
• Note: An earlier version of this paper was presented at the annual meeting of the Academy of Criminal Justice Sciences, New Orleans, LA, March 2000.

Annotation

This paper reports on an analysis of 546 females in a randomly selected sample of 4,047 parolees, with attention to whether a classification system created for the dominantly male population will work equally well for females.

Abstract

The data collection instrument was constructed from multiple sources. First, variables were incorporated from a preliminary departmental study that correlated inmate information with return-to-prison. Second, items were used from McShane and Williams' (1997) review of research on classification variables, and instruments throughout the United States and Canada over the past 20 years were used. Third, items suggested through interviews with parole agents and parole administrators were included. Lastly, the project staff at parole headquarters suggested possible items. All items were originally incorporated into a data collection instrument, and approaches to their measurement were refined over a period of 6 months. A final reduction of variables from the preliminary instrument occurred in a training session for coders. The criterion variable, success on parole, was measured as either "fail" (reincarcerated) or "succeed" (remained on parole) for a period of 12 months following release. Independent variables -- developed for the classification instrument and derived primarily from existing popular classification models -- were age at release on parole, age at first arrest, number of prior arrests, inmate classification level at release, whether the parolee was a new release or a parole violator being re-released, and street/prison gang affiliation. The analysis used multivariate logistic regression. The analysis found that the classification variables and the classification system were suitable for use with females. Alternative variables likely to affect female parolees were tested with the classification model and failed to offer better predictive results for females. 3 tables, 16 notes, and 23 references