Intern Andrea Pownall’s first summary investigates the key findings from the paper by Jonas-van Dijk et al., Victim–Offender Mediation and Reduced Reoffending: Gauging the Self-Selection Bias (2020, Crime & Delinquency, 66(6–7), 949–972). This work examines whether self-selection-bias explains the research findings that mediation can reduce recidivism.
It is unlikely that self-selection bias, emerging from the voluntary nature of restorative justice, fully explains the research findings that suggest victim-offender mediation can reduce rates of reoffending among participating offenders.
While reducing reoffending is not necessarily the primary goal of restorative justice, research suggests that offenders who participate in restorative justice are less likely to reoffend than similar persons who do not. To the authors’ knowledge, no randomised controlled trials have been done to examine the effect of victim-offender mediation (VOM) on recidivism. However, it is important to analyse if reductions in reoffending rates are caused by VOM itself, or by other differences between participating and non-participating offenders. VOM allows a victim and offender to communicate in the presence of a trained mediator. The discussion focuses on the harm inflicted and on the parties’ needs, potentially aiding victim healing and heightening victim empathy among offenders.
This paper aimed to address the self-selection bias that likely affects previous studies because VOM is voluntary. The research on which the paper is based is a reinvestigation of the data collected by Claessen et al. (2015), whose control group did not differentiate between offenders who did not want to participate in VOM, and those who were willing to participate, but were unable to do so because the other party declined. By involving a valid control group and doing in-depth, secondary analyses on this dataset, the research aimed to eliminate this bias.
1275 cases were analysed where mediation was offered as part of the criminal justice system in Southern Netherlands, from 2000-2010. Four different offender groups were examined. The first group (n=633) consisted of offenders who participated in direct or indirect forms of VOM (mediation group). Direct mediation involves a face-to-face meeting between the victim and offender; indirect mediation encompasses an exchange of letters or shuttle mediation between them. The second offender group (n=291) concerned those who participated in semi-mediation, where the offender does not meet the victim but comes to an agreement with the prosecutor in a mediator’s presence (semi-mediation group). The third group (n=206) comprised offenders who were not willing to participate in VOM (court group). The fourth group (n=145) consisted of those who were willing to participate in VOM, but the other party declined (control group).
The cases examined contained demographic and judicial information related to the offender, which was used to compare the four groups. There were no significant differences between the four offender groups in terms of gender, age or country of birth. 81% of cases were male. The mean age of offenders was 36, and most were born in the Netherlands. There were also no significant differences in the number of previous contacts with the criminal justice system between the four groups, although there were significant differences in their offence types: the mediation group included more violent and serious offences than the other groups, while the court group included more minor offences than the other groups.
The Recidivism Monitor in the Netherlands enabled the scholars to analyse the prevalence of reoffending of all offenders in their dataset. The general prevalence of reoffending across the complete time at risk period was 35% (i.e. 441 of 1275 persons reoffended). The time at risk varied from 3.5-13.5 years, which was the time from the case first being judicially registered, until the time that a new criminal case was registered. In cases where no reoffending occurred, the time at risk period was until the end of the observation (July 2014).
A logistic regression analysis was used to investigate differences in reoffending rates between the four groups. This revealed that offenders in the court group were significantly more likely to reoffend, compared to offenders in the mediation and semi-mediation groups. Offenders in the court group were only marginally more likely to reoffend than the control group. As such, those who were not willing to participate in mediation had a much higher risk of reoffending than offenders who participated in mediation, and a slightly higher risk compared to offenders who were willing, but unable, to engage in mediation. The reoffending rate of the control group lay in between the court group, and the semi-mediation and mediation groups.
A Cox survival analysis was used to control for the time at risk period. This found that the court group were still significantly more likely to reoffend compared to the mediation and semi-mediation groups, and marginally more likely to reoffend compared to the control group. There were no significant differences between the semi-mediation and control groups, the mediation and semi-mediation groups, or the mediation and control groups. Offenders in the control group had a reoffending risk between the court group, and the semi-mediation and mediation groups. Controlling for the demographic and offender-related variables (i.e. gender, age, place of birth, previous contact and type of offence committed) showed the same pattern: the court group had the highest predicted risk to reoffend, and the mediation and semi-mediation groups had a significantly lower risk to reoffend. The control group again had a recidivism risk in between the court group, and the mediation and semi-mediation groups, suggesting that even among those who volunteered to mediate, those who ultimately participated had a lower reoffending rate than those who did not.
A limitation of the research is that its design did not allow for the mediation process itself to be examined, so “further systemic research is needed to examine how mediation unfolds and how and when this has an impact on offenders” (p.966). Shapland et al. (2008) conducted post-mediation interviews with offenders and found that, for mediation to be as effective as possible, the offender needs to be actively involved, must want to meet the victim, and must understand the harm inflicted. Another limitation is that the participants’ motivations could not be analysed before VOM. Further empirical research could also examine the effect of VOM on participants’ feelings and the change in offenders’ criminogenic profiles.
Offenders who were not willing to participate in VOM had a much higher risk of reoffending than those who participated in VOM, and a slightly higher risk of reoffending than those who were willing to participate in VOM but were unable to do so. The higher risk of reoffending for those unwilling to participate in VOM could be explained by the impact of mediation and/or by other differences between participating and non-participating offenders. The research aimed to remove the self-selection bias by including a control group of offenders who likely had the same criminogenic profile as the group who did participate in mediation. The intermediate position of the control group suggests that the impact of VOM on reoffending may be explained partly by one’s willingness to participate in VOM, and partly by the mediation process itself. Overall, it is unlikely that self-selection bias fully explains the reduced rates of reoffending connected to victim-offender mediation.