“Restorative justice” is a concept denoting a wide range of justice practices with common values, but widely varying procedures (Braithwaite, 2002). These values encourage offenders to take responsibility for their actions and to repair the harms they have caused, usually (although not always) in communication with their personal victims. This review focuses on the subset of restorative justice procedures that has been tested most carefully and extensively: face-to-face restorative justice conferencing (RJC). In these conferences, victims and offenders involved in a crime meet in the presence of a trained facilitator with their families and friends or others affected by the crime, to discuss and resolve the offense and its consequences.
The reviewers sought to assess the effect of face-to-face restorative justice conferencing on repeat offending and on available measures of victim impact.
The review includes only studies that employed a randomized design to test the effects of conferencing between at least one personal victim and one or more of their offenders on repeat offending or on victim impact, with the random assignment following both offenders’ and victims’ consent to participate in an RJC if selected to do so. Ten eligible studies on three continents were identified, with a total of 1,879 offenders and 734 interviewed victims. The training for the RJC facilitators was provided by the same trainer in all ten trials, but that was not a criterion for selection. Cases were referred to the eligible experiments at various stages of the criminal justice process, including diversion from prosecution, post-conviction RCJs prior to sentencing, and post-sentencing RJCs in prison and probation.
The eligible tests included both violent and property crime, as well as youth and adult crime, with RJCs offered as an alternative or as a supplement to prosecution in court. These variations provide a basis for moderator analyses as well as main effects on subsequent convictions (or in one case, arrests).
The reviewers report the results of the ten eligible experiments identified. These experiments all reported post treatment data only of repeat crime measures at two years after random assignment (the only measurement period of offending common to the ten eligible trials). Measures for victim impact were also post-treatment, as measured by personal interviews with subsets of all victims who consented to random assignment.
All data analyses included in this review examined the effects of Intention-To-Treat (ITT), with wide variations in the percentage of both RJC and control cases receiving treatment as assigned. Many offenders assigned to prosecution, for example, failed to appear in court, just as many offenders assigned to an RJC failed to complete one. The analysis employs the ITT method to provide estimates of effectiveness under real-world conditions, at the expense of likely under-estimates of the efficacy of RJCs when actually delivered. All studies reported effects on individual offenders and victims, while in all cases random assignment was done at the case level. In most trials the ratio of cases to offenders or victims was 1:1, while in others (the two Canberra experiments) that ratio ranged up to 1:1.25.
The evidence of a relationship between conferencing and subsequent convictions or arrests over two years post-random assignment is clear and compelling, with nine out of 10 results in the predicted direction and a standardized mean difference for the ten experiments combined (Cohen’s d = -.155; p = .001). The impact of RJCs on 2-year convictions was reported to be cost-effective in the 7 UK experiments, with up to 14 times as much benefit in costs of the crimes prevented (in London), and 8 times overall, as the cost of delivering RJCs. The effect of conferencing on victims’ satisfaction with the handling of their cases is uniformly positive (d = .327; p<.05), as are several other measures of victim impact.
RJCs delivered in the manner tested by the ten eligible tests in this review appear likely to reduce future detected crimes among the kinds of offenders who are willing to consent to RJCs, and whose victims are also willing to consent. The condition of consent is crucial not just to the research, but also to the aim of its generalizability. The operational basis of holding such conferences at all depends upon consent, since RJCs without consent are arguably unethical and breach accepted principles of restorative justice. The conclusions are appropriately limited to the kinds of cases in which RJCs would be ethical and appropriate. Among the kinds of cases in which both offenders and victims are willing to meet, RJCs seem likely to reduce future crime. Victims’ satisfaction with the handling of their cases is consistently higher for victims assigned to RJCs than for victims whose cases were assigned to normal criminal justice processing.