Differences in the Predictive Validity of Actuarial Risk Assessments in Relation to Sex Offender Type

• NCJ Number: 201923
• Journal: International Journal of Offender Therapy and Comparative Criminology Volume: 47 Issue: 4 Dated: August 2003 Pages: 422-438
• Author(s): Darci L. Bartosh; Tina Garby; Deborah Lewis; Steve Gray
• Date Published: August 2003
• Length: 17 pages
• Publication Format: Article
• Publication Type: Report (Study/Research)
• Language: English
• Country: United States of America

Annotation

This article discusses risk assessment for sex offenders.

Abstract

Sex offender risk assessment is an evolving process. A clinician’s ability to predict which offenders will re-offend has limitations. This study sought to increase accuracy of risk assessment by considering four commonly used actuarials with regard to sex-offender type. These actuarials are RRASOR, Static-99, MnSOST-R, and SORAG. The study focused on sexual offenders released from the Arizona Department of Corrections (ADC) that were subject to registration and notification via the Arizona Sex Offender Notification Program. All offenders released between June 1996 and December 31, 1996, were included in the study. Subjects were scored on all four assessment instruments as well as the Child and Adolescent Taxon Scale (CATS). Data collection was conducted through ADC file review exclusively. The findings suggest that taking into account offender type may potentially increase predictive accuracy. When considering the entire sample population, the Static-99 was significantly predictive of sexual recidivism, violent recidivism, and any recidivism. The RRASOR was significantly predictive of sexual re-offenders as well as offenders in the sexual and/or violent category but not of offenders in the violent-only category. The SORAG was moderately predictive of all types of recidivism except for sexual. The MnSOST-R failed to establish predictive validity above change when considering the entire sample population except in terms of any recidivism. The results support the use of RRASOR, Static-99, and SORAG with child molesters for predicting sexual, violent, and any recidivism. The findings of the Static-99 and the SORAG neared significance for rapists in regard to sexual offending. None of the tests were significantly valid for hands-off offenders. Differences between ROC values between types of offenders, and most specifically between child molesters (incest and extra-familial) and rapist and hands-off offenders, were considerable and may warrant further examination. 8 tables, 32 references