Date of Award

2015

Document Type

Thesis - ECU Access Only

Publisher

Edith Cowan University

Degree Name

Doctor of Philosophy

School

School of Psychology and Social Science

Faculty

Faculty of Health, Engineering and Science

First Supervisor

Professor Alfred Allan

Second Supervisor

Dr Ricks Allan

Abstract

The prediction of risk of violent and sexual reoffending is very important for the mental health and correctional practitioners making treatment decisions and providing opinions to the courts. Currently, Australian practitioners use risk assessments that were developed in other countries despite little evidence that they can be validly used locally, especially with Aboriginal and Torres Strait Islander people. Both scholars and the courts have expressed concern about the use of data generated with these assessments to make important decisions about risk and risk management. The purpose of the current study was to develop a risk assessment for West Australian male Aboriginal violent and sexual offenders. During the first stage of the study, focus groups were used to identify risk factors that Aboriginal correctional workers believed could be unique to Aboriginal men in differing Western Australian geographical regions. The researcher then compiled a list of these predictors; those routinely used in existing risk measures; and others identified in the research literature to compile a list of possible predictors of reoffending. During three subsequent quantitative studies the researcher used retrospective file data of 1838 male Aboriginal participants obtained from the Western Australian Department of Corrective Services to examine the predictive accuracy of the identified predictors. The predictors that most accurately distinguished reoffenders from non-reoffenders were then used to develop a risk assessment. It was not possible to develop an assessment for violent offenders, but one was developed specifically for sexual offenders. This assessment was comprised of three risk items. They were unrealistic long-term goals, unfeasible release plans and poor coping skills (the 3-Predictor Model). The predictive accuracy of recidivism (sensitivity) of themodel was 92.3%, while the predictive accuracy of desisting (specificity) was 94.3%. The model also outperformed the Western Australian Community Corrections case needs model and adult actuarial risk instrument, the level of service need inventory, and the rapid risk assessment for sexual offence recidivism. Given the relative accuracy and the ability of the 3-Predictor model to outperform other risk assessments, further validation appears warranted.

Share

Paper Location

 
COinS