One of the main objectives of the New Zealand court-referred restorative justice pilot was to reduce the rate of reoffending by offenders referred to restorative justice conferences, compared to similar offenders dealt with through the conventional court process. In May 2005 the Ministry of Justice published the main evaluation findings of the court-referred restorative justice pilot which included a preliminary reconviction analysis.
The main focus of the current report is reoffending by offenders who attended a conference (the conferenced group). Reoffending, as measured by the reconviction rate, was assessed throughout the two year follow-up period (using survival analysis) and at the end of the one year and two year follow-up periods. The seriousness of reoffending and the subsequent imprisonment rate were also examined. The results for the conferenced group were compared with their predicted reconviction rates, which were derived from a logistic regression model. The results for the conferenced group were also compared with the results for ten matched comparison groups, selected from eligible offenders who were not referred to the pilot.
The main finding was that there appeared to be a small overall decrease in the reconviction rate of the conferenced group compared to matched comparison groups. The decrease was around four percent in absolute terms, for both the one year and two year follow-up periods, compared to both the predicted reconviction rate and the average reconviction rate for the matched comparison groups. While these differences were not statistically significant, the slight decrease in reoffending does appear to be real, based on the finding that the conferenced group had a lower reconviction rate than all ten matched comparison groups throughout the two year period.
Key Results
- The actual two year reconviction rate for the conferenced group was 41%, compared to their predicted rate of 45%. The predicted rate was estimated from a range of variables, such as criminal history, demographics and offence type, using a logistic regression model. The actual two year reconviction rate for the ten matched comparison groups ranged from 42% to 49%, with an average of 45%. The comparison groups were matched as closely as possible to the conferenced group by their predicted reconviction rate, gender, age, offence group, first offender status, and ethnicity.
- The survival curve (the proportion of the group who had not yet reoffended over time) for the conferenced group was above that for all ten comparison groups throughout the two year period, but not significantly so.
- The subsequent imprisonment rate was 10% for the conferenced group, which was not significantly different to the average of 12% for the comparison groups. The imprisonment rate for the ten comparison groups ranged from 8% to 16%.
- The distribution of the seriousness of reoffending did not differ significantly between the conferenced and comparison groups.
The conferenced group was also compared to offenders who were referred to the pilot, but who did not attend a conference (non-conferenced group) and to a more general group of all offenders potentially eligible for referral (other eligible group). The main findings were:
- Certain types of offenders were more likely than others to be referred to the restorative justice pilot and to participate in a conference. In particular, conferenced offenders were more likely to have fewer and less serious previous convictions and were more likely to be traffic offenders than both non-conferenced offenders and other offenders potentially eligible for referral.
- These characteristics meant that the conferenced group had a significantly lower actual and predicted two year reconviction rate compared to both the non-conferenced group (60%) and the group of other eligible offenders (62%). The survival curves of both groups were also significantly different from that of the conferenced group.
- The seriousness of subsequent offending and the subsequent imprisonment rate were also significantly lower for the conferenced group than both the non-conferenced group and the other eligible group