Discussion

Collectively the results of this analysis suggest the following conclusions:

As a group Maori offenders appeared to be at a disadvantage compared to European offenders in terms of rates of access to HD at both the sentencing and parole hearing stages.

Much of this apparent disadvantage could be explained by ethnic differences in the personal characteristics, offending history and related factors that were also predictive of access to HD.

Nevertheless, even when all of the available covariate factors were taken into account statistically, there remained evidence of a modest bias against Maori in rates of access to HD. The adjusted estimates for leave to apply for HD at sentencing suggest that rates of access to leave for Maori may be in the region of 3.6% lower than for European offenders, and rates of attainment of leave at the Parole Board hearing may be in the region of 2.4% lower than for European offenders. One implication of these findings is that for Maori in this sample to have had the same rates of access to HD as European offenders there would need to have been somewhere in the region of 145 additional Maori granted leave to apply for HD at sentencing and approximately 96 additional Maori granted HD at a parole hearing.

The above findings illustrate the apparent paradox by which what are relatively small differences in rates of access to HD can affect a relatively large number of individuals. This arises because, for this sample of prisoners, the bias in access to HD is against Maori, and Maori constitute the majority (59%) of the sample.

At the same time, the analysis is not without its limitations. In particular, there may be residual sources of confounding that were not well represented in the model and which could explain any remaining ethnic differences in access. For example, the analysis has been unable to account for ethnic differences in such factors such as prisoner demeanour, behaviour or presentation at either the sentencing or parole hearing stages, and these factors may well be related to HD outcomes.

Collectively, these findings would suggest that any ethnic bias in the judicial processes related to access to HD is likely to be small. Nevertheless, to the extent that Maori constitute the majority of the prison population any evidence of bias against Maori in access to HD should be a source of concern. For this reason it would be prudent to continue to monitor the system for evidence of bias.

Reference

Lee J (1981) Covariate adjustment of rates based on the multiple logistic regression model. Journal of Chronic Diseases, 34: 415-426.