| ||Literature Coverage Dates||Number of Studies||Number of Study Participants|
|Meta-Analysis 1||1976 - 2016||84||0|
|Meta-Analysis 2||1990 - 2013||21||18258|
Wilson and colleagues (2017) evaluated the effectiveness of restorative justice programs for youth across relevant outcomes. To find identify studies, specific keywords related to restorative justice were used to search bibliographic databases. The search strategy was executed from January 5 through January 22, 2016. A total of 41 electronic databases and 50 state websites were searched. Additionally, numerous organization websites were searched for less formally published works.
To be eligible for inclusion, studies had to have tested the effectiveness of a restorative justice program, or a juvenile justice program that included at least one restorative justice component. Studies were also eligible if the samples consisted of juveniles involved in the juvenile justice system at any level or juveniles engaged in problem behaviors without contact with the justice system. Juveniles were defined as aged 18 or younger. Eligible studies were limited to experimental or quasi-experimental designs that compared youths who received restorative justice services with either a comparison group that received traditional formal processing or a comparison group that received an alternative program. Studies had to have reported data related to at least one of the following outcomes: criminal behavior, participant satisfaction, perception of fairness, restitution compliance, reparation of harm, and/or juvenile justice system costs. Both published and unpublished reports were included in the search, as were studies conducted in other countries. However, only studies that were in English were considered.
A total of 31,019 titles were identified, and 1,312 references were subsequently screened against the eligibility criteria, producing 99 eligible references. Following full-text screening of identified references, 60 unique research studies were considered eligible for review. Most of these studies were conducted in the United States (N=46), followed by Australia (N=8), United Kingdom (N=3), European Union (N=2), and Canada (N=1). Of the 60 studies, 52 reported results from a single evaluation. The other 8 studies included multiple independent evaluations resulting in a total of 84 unique “substudies.” Of these, 67 used quasi-experimental designs, whereas 17 used random assignment. In addition, 4 studies offered multiple treatment conditions, compared with a single comparison condition. To maximize the number of available contrasts within each restorative justice program type, the authors kept these as separate comparison contrasts. The final number of comparisons was 91.
In terms of study characteristics, most studies included treatment and comparison conditions that each had at least 100 participants. Most studies had a mix of males and females, with only three restricted solely to males and one restricted solely to females. Most of the youths in these studies had some form of contact with the juvenile justice system, either as non-adjudicated (e.g., diverted from formal processing) or as adjudicated (e.g., formally processed). Six studies were based on samples obtained from a non-juvenile justice setting (i.e., school-based). Finally, one study (Do 2006) examined a victim-awareness class in an institutional setting. It should be noted that the racial and ethnic makeup of the samples was missing from too many studies to provide a reasonable description of this characteristic across the included studies.
The majority of studies (56 percent) examined a specific restorative justice program, with the victim-offender conferencing being the most commonly used restorative justice program. Other programs not explicitly defined as restorative justice programs, but which included restorative justice elements/principles, were restitution, teen courts, cautioning/diversion programs, and programs that included a combination of restorative justice programs or practices.
A standardized mean difference (Hedges’ g) was calculated for all outcomes in the meta-analysis. Given that most of the outcomes were scaled variables, dichotomous outcomes (e.g., measures of delinquency) were converted to the Hedges’ g using the Cox method. To analyze the effect sizes, inverse variance weights were used, as well as random effects models that assume variance in treatment effects across the included studies.Meta-Analysis 2
Wong and colleagues (2016) evaluated the effectiveness of restorative justice programs for diverted youth on recidivism using meta-analytic techniques. To identify studies, the researchers searched 20 bibliographic databases from January 1990 to April 21, 2015. In addition, they used the Ancestry method by reviewing the bibliographies of narrative literature reviews or existing meta-analyses on diversion strategies or other related studies on diversion programs.
To be eligible for inclusion, evaluations of restorative justice programs had to have reported on at least one individual-level outcome of crime/delinquency for at-risk youth. Furthermore, studies had to have included a treatment sample size of at least 20 at-risk youths aged 12 to 18 years. Regarding research design requirements, eligible studies included those that used random assignment to treatment/comparison conditions or quasi-experimental designs, which used a matched control to ensure the control group was appropriate for comparison purposes. Evaluations were limited to those that took place in Canada, the United States, Australia, New Zealand, or a Western European country, and those that were published in English or French between January 1990 and April 2015. To be eligible for inclusion, the program also had to be delivered in a non-closed community setting. Thus, programs that were delivered in schools during the school day, in custody facilities, or in hospitals were not eligible for inclusion.
After screening identified studies using the exclusion/inclusion criteria, 21 studies contributing 21 comparisons were considered eligible for inclusion. These comprised studies conducted between 1990 and 2013, with a total of 5,209 treatment group participants and 13,049 comparison group participants. Of these studies, 13 were technical reports, 4 were theses or dissertations, 3 were peer-reviewed journals, and 1 study was published in a book chapter. Of the 21 studies, 15 took place in the United States, while the remaining 6 took place in Australia, New Zealand, or Europe. Regarding research design, 3 studies used random assignment, while 18 studies used quasi-experimental designs. Nearly all studies provided official reports of delinquent/criminal behavior (e.g., police/court contact or referral, arrest); however, 1 study used participant self-report. In terms of gender, 15 of the 21 studies used samples composed predominately of males, 2 had an even mix of males and females, and 4 did not include information on the gender breakdown of participants. Treatment group sample sizes varied substantially among studies, with a range from 25 to 917 participants. Sample size was dichotomized for the purposes of analysis, with 8 studies (38%) using a total sample of less than 100 students and 13 (62%) using a sample of 100 participants or more. Finally, more than half of the samples were composed of mostly white youths (56%, n = 9), while 44% (n = 7) of the samples contained primarily minority youth or were fairly evenly distributed with respect to race/ethnicity (five studies did not include data on race/ethnicity).
Given the level of diversity in programs included in the meta-analysis, a random-effects model was used to account for the heterogeneity across studies. Given that the majority of the studies in the current analysis present 2 × 2 outcomes in terms of youths who recidivated or did not recidivate, effect sizes were calculated as log odds ratios.