Amy Veith

Date of Award

January 2018

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Psychology

First Advisor

Alan R. King

Abstract

The current study explored the accuracy of Static-99R and STABLE-2007 risk evaluations in predicting sex offender recidivism (n = 136) over an average 7.8 year follow-up period. These protocols are differentiated on the basis of their reliance on actuarial versus

dynamic risk factors respectively. The present analysis tested their predictive validity both independently and in combination with one another. It was hypothesized that the Static-99R and the STABLE-2007 would provide moderate predictive accuracy that would be enhanced by their combination. Support was not found for either hypothesis in regard to sexual recidivism, but the Static-99-R did show fair accuracy (AUC = .76, p = .008) in predicting violent re-offenses. Raw scores from both protocols served as weak (AUCs ~ .61) but statistically significant predictors of total recidivism. The occurrence of total recidivism was elevated substantially for offenders

classified in either the Static-99R Moderate to High (RR = 3.4, p = .02) or High (RR = 3.2, p = .04) risk categories. The combination of Static-99R and STABLE-2007 scores generated a metric that was comparable in predictive accuracy to either used in isolation. The STABLE-2007 did not evidence any incremental validity not provided by the Static-99R in this sample. These results contribute to a growing literature regarding both the value and limitations of structured actuarial and dynamic risk assessments in the prediction of criminal recidivism. Both protocols were found to underachieve in their predictive validity in this small community-based ten-year

outcome assessment. These results were discussed within the context of a number of research design limitations evident in this community sample analysis.

Download

Share

COinS