Actuarial Risk Assessment Models: A Review of Critical Issues Related to Violence and Sex-Offender Recidivism Assessments

• NCJ Number: 187050
• Journal: Journal of the American Academy of Psychiatry and the Law Volume: 28 Issue: 4 Dated: 2000 Pages: 438-448
• Author(s): Shoba Sreenivasan Ph.D.; Patricia Kirkish Ph.D.; Thomas Garrick M.D.; Linda E. Weinberger Ph.D.; Amy Phenix Ph.D.
• Date Published: 2000
• Length: 11 pages
• Publication Format: Article
• Publication Type: Issue Overview
• Language: English
• Country: United States of America

Annotation

Neither clinical judgment unguided by the research literature nor the use of a sole actuarial model would meet judicially determined or professional standards of practice in predicting violent behavior or sex offender recidivism.

Abstract

Risk assessment in the area of identification of violence has been dichotomized by several prominent researchers as the "clinical approach" versus the "actuarial method." The proponents of the actuarial approach argue for actuarially derived decisions to replace existing clinical practice. The actuarial method requires no clinical input, just a translation of the relevant material from the records to calculate the risk score. A risk appraisal approach based upon a sole actuarial method raises several question: those of public safety, peer-accepted standards of practice, liability issues, and concordance with evidence-based medicine practice. Although researchers have contributed significantly to the identification of some general factors highly associated with violent recidivism, the responsibility of the clinician is to understand how these factors are represented in the specific patient. The best strategies are based upon a solid theoretical framework, clinical acumen, and actuarial data. Risk appraisal should be based upon guided clinical judgment by forensically trained clinicians. Use of a single model, based exclusively on group norms, is tantamount to predicting that every individual is best described by a mean. Clearly this is both clinically and statistically incorrect. Integrating clinical and actuarial data is the acceptable standard of practice and underlies every cautionary caveat inscribed in computer-generated assessment tools. The clinician is responsible for using the individual's unique factors to interpret the results of a given test. This same caution must be applied to actuarial tools. 50 references