Violent offending is of concern to both the public in general and the justice system in particular. Increasing trends in the numbers of offenders convicted of violent offences have placed increasing pressures on the Justice Department to address the problem and make decisions concerning possible solutions. The main aims of the present study were to firstly, assess the extent to which violent offenders present a problem to the Justice system using base rates, reconviction probabilities and seriousness data. Secondly, to outline how this information can be used to develop priority guidelines for rehabilitative efforts with violent offenders.

Utilizing information from the Wanganui computer, criminal histories of a randomly selected sample of 950 violent offenders were obtained. All offenders selected had a conviction for a violent offence between the dates, 1st January 1979 and 31st December 1981. These data were used to compile person-based offence records which were manually coded onto a database on a desktop computer. Data analysis was designed so as to address base rates, reconviction probabilities and seriousness of subsequent offending for each category of violent offender. Predictor variables related to reoffending were also examined and evaluated using discriminant function analysis.

The results obtained indicated the following: Base rate information identified three categories characterized by significantly higher base rates than the other five categories. These categories were Minor Assault, Serious Assault and Intimidation offenders. Reconviction probabilities were consistently high for all eight categories of violent offenders (i.e. around 0.65). Further, seriousness of subsequent offending was similar for all categories of violent offender. By combining these three information sources it was decided that Minor Assault, Serious Assault and Intimidation categories of violent offender represent a significant group of violent offenders for whom prioritization guidelines for rehabilitation should be established.

In order to increase the likelihood of correctly targeting reoffenders for treatment within these three categories a prediction tool was developed using a discriminant function analysis. This analysis found the following variables to be significant predictors of reconviction for all three categories at the 0.01 level of significance: age at first conviction, age at first violent conviction, age at criterion offence, prior rate of offending (both general and violent) and race. The power of the discriminant functions were evaluated using ROC curves which found the Minor Assault discriminant function to be the most effective in predicting reoffending.The predictive tool was demonstrated to be valuable in improving prioritization procedures for rehabilitation and increasing the effectiveness of limited rehabilitative resources. Implications of employing the tool are discussed.