Is Restorative Justice Effective in the U.S.? Evaluating Program Methods and Findings Using Meta-analysis

The presence of restorative justice (RJ) in the United States has grown steadily within the last five decades. The dynamics of RJ programs are meant to more holistically address the harms caused by crime in comparison to the traditional criminal justice system (CJS). Yet, evaluative research has provided inconsistent evidence of their effectiveness and the quality of empirical study has gone untested. The current study sought to fill the gaps within past research by examining how success has been measured, assessing the rigor of study methodology using the Maryland Scientific Methods Scale (SMS), and determining the impact of RJ programs on recidivism, victim satisfaction and restitution compliance using meta-analysis. A systematic search of past literature identified a sample of 121 studies whose dependent measures were coded, and methodological designs were rated using the SMS. Most studies failed to include community-based measures of success or measures which reflect the goals of RJ to undue harms and restore relationships. SMS scores were well distributed within the sample. Despite restricted sample sizes, meta-analyses used extracted data from 35 case-control, quasi-experimental and experimental studies to generate 43 unique treatment contrasts and 3 summary effects. Meta-analytic findings favored RJ treatment over CJS control groups across all dependent measures. Heterogeneity between subsequent arrest studies was scrutinized using subgroup analysis. The fewest subsequent arrests were associated with adult offenders, mandated participation, mediation and hybrid programs, and the most rigorous methodologies. Findings support continued efforts to improve the methodological rigor of evaluations, targeted focus on specific program types and delivery characteristics. Future meta-analyses would benefit from the inclusion of non-American RJ program evaluations to enlarge pooled sample populations and better detect moderating influences. Other suggestions for research design improvements include the use of more holistic and stakeholder-centric measures for success, use of continuous measures, and refined indicator variables for heterogeneity testing (e.g., crime type severity, characteristics of program fidelity). The author recommends continued use of these programs, specifically with adult offenders and incidents of serious crime toward a better understanding of the true impacts of RJ on stakeholders. More detailed results, study limitations and implications are discussed herein.