I Think, Therefore I Am (a Criminal?): The Predictive Validity of the LSI-R's Attitudes and Orientation Domain

• In Event: Technology and Data in Community Corrections

Abstract

The Level of Service Inventory-Revised (LSI-R) is a frequently used assessment tool that identifies offenders’ risks and criminogenic needs across 10 domains. The predictive validity of the LSI-R has been established through multiple studies. Many of these studies, however, focus on the overall LSI-R score; fewer examine the predictive validity of individual domain scores. Multiple criminological theories posit that those who have antisocial attitudes and beliefs are more likely to engage in criminal behavior. As such, in this paper we use a sample of 760 offenders released from state prisons in 2013 to test whether the “Attitudes and Orientation” domain score is predictive of recidivism using a three year follow-up. We measure recidivism and test predictive validity in three ways. First, we use logistic regression to examine the relationship with a dichotomous (yes/no) measure of any new arrest during the follow-up. Second, we use ordinary least squares (OLS) regression with a continuous measure of the total number of arrests during the follow-up. Finally, to address possible right censoring, we examine the time to first arrest using survival analysis. Controls for demographic, criminal history, and offense information will be included in each analysis.

Authors

• Shaun M. Gann, Boise State University

• Lisa Growette Bostaph, Boise State University