Evaluating and Improving Risk Assessment Schemes for Sexual Recidivism: A Long-Term Follow-Up of Convicted Sexual Offenders

• NCJ Number: 217618
• Author(s): Raymond A. Knight Ph.D.; David Thornton Ph.D.
• Date Published: March 2007
• Length: 155 pages
• Grant Number(s): 2003-WG-BX-1002
• Agencies: NIJ-Sponsored
• Publication Type: Report (Study/Research)

Annotation

This federally supported study sought to evaluate and to improve the decisionmaking algorithms that have been generated to assess risk in sexual offenders by evaluating the existing risk assessment measures in a sample of sexual offenders on whom long-term follow up were available.

Abstract

Highlights of the results were: (1) for the average predictability over all measures, the Bridgewater Observations sample (BOs) were better predicted than Bridgewater Treatments sample (BTs), despite a significantly lower recidivism rate; (2) the cross-temporal pattern of prediction differed between rapists and child molesters, with rapists being predicted better at shorter follow-up periods and child molesters better at longer intervals; (3) all actuarials showed moderate reliability and predictive accuracy with few significant differences; (4) five factors accounted for all of the predictive variance in the existing actuarials; (5) the Structured Risk Assessment (SRA) Needs Assessment consistently had the highest area under the Receiver Operating Characteristic (ROC) curve (AUCs) for the entire sample and for rapists and child molesters separately; (6) age was not found to constitute an important moderator for predicting outcome, and a complex relation among age at index offense, age at discharge, and outcome status emerged; and (7) promising additional subgroup specific predictors for child molesters and rapists were identified. Risk assessment plays a central role in the management of sexual offenders. The study assessed the comparative accuracy of the major risk instruments over time and over sub samples, explored their underlying factor structure, examined the accuracy of a new assessment protocol, the SRA Needs Assessment, and explored the potential for generating improved predictive instruments. To accomplish this, archival files from a prior study which followed 599 offenders referred to the Massachusetts Treatment Center (MTC) from 1959 and 1984 were used. The offenders were referred from both Bridgewater Treatment (BT) and Bridgewater Observation (BO). They were coded on the modern actuarials that have been developed since 1998 and on a new experimental measure, the SRA Needs Assessment. References and tables