Degree Type

Dissertation

Date of Award

2007

Degree Name

Doctor of Philosophy

Department

Psychology

First Advisor

Douglas L. Epperson

Abstract

As evidenced by the passage of the Adam Walsh Act and the comprehensive media coverage of several high-profile cases, "The prediction of criminal behavior is perhaps one of the most central issues in the criminal justice system" (Andrews & Bonta, 1998, p. 211). Recent research has generally documented the superiority of actuarial prediction over clinical judgment in decision-making across various domains including risk assessment (Harris, et al., 1993; Grove & Meehl, 1996; Grove, et al., 2000). Several actuarial sexual recidivism risk assessment tools are now well-established (e.g. Static-99; RRASOR; SORAG; and MnSOST-R).;Despite this, Hanson (1998) noted that clinicians may still question the comprehensiveness of actuarial methods and "Those skeptical of actuarial predictions will always find reasons to adjust actuarial estimates" (p. 65). In response to this debate, many clinicians have argued for the use of adjusted actuarial assessment. However, there is currently not a single study documenting the accuracy of adjusted actuarial risk assessment.;The current examines the impact of these overrides on predictive accuracy. Specifically, the study examines the overall accuracy of the purely actuarial assessments versus the adjusted actuarial assessments. Because adjusted actuarial assessment is more costly in terms of personnel and time and because it opens the door for subjective bias, justification for this method should require it to result in significantly greater accuracy that purely actuarial adjustment.;The full sample for this study was comprised of 383 cases in which there had been a clinical override of the Minnesota Sex Offender Screening Tool-Revised (MnSOST-R) in the assigned risk level. ROC Analyses indicated that adjusting MnSOST-R actuarial risk levels based on clinical factors did not result in increased accuracy. Although MnSOST-R levels were nominally more accurate than adjusted risk levels, the difference was not significant. There was no evidence that the direction of the adjustment significantly affected accuracy. These results must be considered to be exploratory because of the small number of sexual recidivists (19) due to the short period of follow-up, which was as short as 2 to 3 years for many in the sample.

DOI

https://doi.org/10.31274/rtd-180813-16754

Publisher

Digital Repository @ Iowa State University, http://lib.dr.iastate.edu/

Copyright Owner

Kathleen Spencer Gore

Language

en

Proquest ID

AAI3274898

OCLC Number

191802337

ISBN

9780549154952

File Format

application/pdf

File Size

85 pages

Share

COinS