Date of Award
Bachelor of Arts Honours
Faculty of Health and Human Sciences
A retrospective study of 243 male probationers who had been on community based orders in Western Australia for a mean time of 15 months, was undertaken to explore differences between re-offenders and non re-offenders. Discriminant function analyses were employed in a series of designs where the mediating effects of geographic location and Aboriginality and non Aboriginality were investigated. The analyses revealed that the best static predictor item for distinguishing between non re-offenders and re-offenders in the entire sample was offence type (Wilks Lambda, .88, chi-square 25.589, df = 6, p < .0005) and the best criminogenic need item was employment (Wilks Lambda, .96, chi-square 7.566, df = 2, p < .05). In regional areas, drug use was !he primary predictor contributing to a function which significantly discriminated between and re-offenders and non re-offenders (Wilks Lambda, .78, chi-square 12.557, df = 4, p < .05). The classification accuracy was 68% for grouped cases. This result was unexpected, as previous studies have consistently found static predictors to be primary predictors of risk. Analysis of the metropolitan area sub-sample produced results more consistent with previous findings. Offence type and number of breached orders loaded highly on a statistically significant function which satisfactorily discriminated between outcomes (Wilks Lambda, .81, chi-square 31.226, df= 6, p < .0005). The analysis of race produced similar results. The variables which had tile highest loadings on the derived functions for both sub-samples were al11 static predictors of risk. Based on meta-analytic research outcomes of Andrews et al. (1990), it was also hypothesised that a chi-square analysis of court sanctioned probation conditions would reveal differences across re-offending outcome and the nature of the probation conditions. The results were consistent with the finding that general correctional service combined with a judicial alternative produced greater reductions in recidivism than a judicial alternative alone. The outcomes related to geographic location and race reinforced the importance of assessing risk of recidivism on the basis of population-specific attributes. Despite several limitations associated with the research design, the exploration provided future directions for the development of risk models and the use of judicial alternatives to reduce recidivism.
Dawson, D. (1997). Preliminary Exploration of Risk Assessment Predictors : Can Probationers who Re-Offend be Identified?. Retrieved from http://ro.ecu.edu.au/theses_hons/769