William Bradshaw, Ph.D.
University of Minnesota
David Roseborough, MSW
University of St. Thomas
THE DEVELOPMENT OF effective programs and interventions to reduce juvenile recidivism is a national priority. Juvenile criminal offenses are a significant societal problem with great financial and social costs. Adolescent boys commit higher rates of criminal acts than any other age group and use much of the resources of youth services systems (National Council of Juvenile and Family Court Judges, 1996). Antisocial behavior has significant negative emotional, physical, and financial effects on victims, their families, and communities. Child mental health, youth services, juvenile justice, and child welfare systems have been involved in providing a range of correctional, rehabilitative, and psychological approaches to reduce juvenile recidivism.
Traditionally, the juvenile justice system in the United States has been dominated by two different approaches in responding to juvenile offenses, the retributive justice model and the rehabilitation or treatment model. The retributive model defines a juvenile offense as a crime against the state and the state provides suitable punishment to the offender. The assumption of the retributive model is that punishment will deter future offenses. However, the retributive model often creates situations that increase the likelihood of further delinquent activity (Crouch, 1993; Link, 1987; May & Pitts, 1999). The juvenile offender is also at high risk of lowered educational and occupational opportunities and delinquent behavior is a strong predictor that the offender himself will be victimized (Lauritsen, Laub, & Sampson, 1991).
The rehabilitative model focuses on the treatment of the offender with the assumption that interventions such as probation supervision, work readiness training, cognitive skills training, and behavior therapy will change behavior and reduce the frequency of juvenile offenses. Historically, however, there has been little evidence for the success of these methods in reducing recidivism. Henggleger (1989), in his review of two decades of juvenile justice system attempts to reduce recidivism, concluded that “nothing works.” Lipsey’s (1995) meta-analysis of 400 outcome studies that involved 40,000 juvenile offenders showed only a small average reduction of 10 percent in recidivism.
Restorative justice is an increasingly important alternative approach to responding to criminal offenses (Bazemore & Umbreit, 1995). While the retributive and rehabilitative models focus on the punishment or rehabilitation of the offender, they neglect the needs of the victims. In contrast, for hundreds of years, indigenous populations in New Zealand, the United States, and Canada used rituals to bring together family and friends of both victims and offender to search for a resolution to the problem that was acceptable to all involved. Initial restorative justice programs focused largely on victim offender mediation and on providing restitution to victims. The conceptualization of restorative justice has expanded both its initial formulation and program services over the last 20 years with a broader range of policies and practices being adapted by an increasing number of jurisdictions (Bazemore & Schiff, 2001; Umbreit, Coates & Voss, 2002). Restorative justice assumes that criminal offenses are first a violation of people and relationships and not just in the domain of the state. The restorative model reconceptualizes the purpose of justice by focusing on the three major stakeholders in the process of restoration and healing: the victim, offender, and community (Zehr, 2002).
The aim of restorative justice is to repair the harm done by the crime by bringing together the people most affected by the offense to determine how to deal with the offence. Dialogue, reparation, and accountability are critical components of all restorative interventions (Bazemore, 1996; Umbreit, 2000; Zehr, 1990). This process aims to benefit victim, offender, and the community. The victims are able to express their feelings, get questions answered regarding the crime, and have input into the reparation plan. The offender is held personally accountable in providing restitution and the restorative process also promotes the support and reintegration of the victim and offender into the community.
There are currently three types of uniquely restorative justice dialogue programs that receive a good deal of attention: victim-offender mediation, family group conferencing, and peacemaking circles. Victim- offender mediation (VOM) is the most established intervention model of the restorative justice movement, with more than 1300 VOM programs in 18 countries (Umbreit & Greenwood, 1999). The practice of VOM is grounded in restorative justice theory that emphasizes that crime first should be perceived as an act against individuals within the context of community. While not denying that the state clearly has an interest in preventing and resolving criminal conflict, restorative justice offers a process by which those most directly affected by crime—the victim, community, and offender—have an opportunity to be involved directly in responding to the offense, holding the offender accountable, offering emotional and material assistance to the victim, and working toward the development of a safe and caring community for victim and offender.
The heart of VOM is a guided face-to-face meeting between a crime victim and the person or persons who victimized him or her, along with parents or other support people, if desired by the victim or offender. The goal of these meetings is to provide a safe place for genuine dialogue between the involved parties that can address emotional and informational needs and develop a restitution plan. VOM programs typically involve victims and perpetrators of juvenile property offenses and minor assaults, and their parents. Some programs have expanded the focus of VOM and provide mediated dialogue for crimes of severe violence (Umbreit, 1994).
Current research on VOM has focused primarily on specific victim and offender outcomes, satisfaction, fairness, and restitution completion. Numerous studies have found uniformly high levels of satisfaction with mediation for both victims and offenders. Umbreit, Coates, and Voss (2002) review two decades of research in VOM and note that typically 80-90 percent of participants report being satisfied with the process and 90 percent of these meetings resulted in restitution agreements. Of these restitution agreements 80 to 90 percent have been reported as completed. These findings are consistent across sites, cultures, and severity of offence. Similarly, in a study that examined the experience of fairness in the justice system, 80 percent of burglary victims who participated in VOM reported they experienced the criminal justice system as fair, compared with only 37 percent of victims who did not participate in VOM (Umbreit, 1989).
Family group conferencing (FGC), also called community conferencing, originated in New Zealand as a means of diverting young offenders from formal adjudication. FGC was based largely on the ancient tradition of indigenous people of New Zealand, the Maori. It later evolved in Australia into police-based conferencing that allowed police to bring together juvenile offenders, their families, and supporters on the one hand, and the crime victim and their family and supporters. The goals of the conference are to help offenders understand the impact of the offense on the victims and take responsibility for their actions. The conference provides victims with the opportunity to move toward forgiveness and empowers the community to resolve the problem in ways fitting the situation and stakeholders (McCold & Wachtel, 1998). FGC has become increasingly popular in the United States, Canada, Europe, and South Africa.
The philosophy of restorative conferencing is based on Braithwaite’s (1989) theory of reintegrative shaming, control, and deterrence (Hirschi, 1969) and problem-oriented policing (Goldstein, 1990). Although there are different models of restorative justice conferencing, there are four fundamental assumptions of conferences. These include 1) family and extended family are respected and the focus must be on strengthening family and social supports; 2) power must be given to all participants; 3) conferences must be culturally sensitive and respectful to families; and 4) victims must be involved in the process and get what is needed to repair the harm done to them (McGarrell, Olivares, Crawford, & Kroovand, 2000). Several studies have reported high levels of victim satisfaction (over 90 percent), offender satisfaction, and victim and offender experience of fairness with the conference process (McCold & Wachtel, 1998).
The primary goal of peacemaking circles is to promote accountability, healing, and compassion through community participation in resolving conflicts. Peacemaking circles are based on the process of dialogue, relationship building, and the communication of moral values in order to accomplish the key outcomes of reparation of harm and improvement in social well being (Presser & Van-Voorhis, 2002). Peacemaking circles, sometimes referred to as sentencing circles, have been used in Canada to empower Native peoples and to transfer some aspects of the judges’ role to Aboriginal communities (Jaccoud & Walgrave, 1999). Jacoud and Walgrave (1999) have also suggested that sentencing circles may provide a restorative justice solution to what some consider two limitations of victim offender mediation: the lack of concern for larger community safety and the limitations of voluntary settlements.
The literature on peacemaking and sentencing circles is primarily descriptive (Morris, 2000; Green, 1998; Umbreit, Coates & Voss, 2002; Stuart, 2001, 1996). There are only two known studies of Circles (Umbreit, Coates & Voss; Stuart, 1996 ). Both studies report many positive impacts of peacemaking circles. Neither study, however, examined the effect of peacemaking circles on recidivism.
Restorative justice principles have broad appeal and advocates of restorative justice practices point to many benefits of restorative interventions. Victim needs are more fully met, offenders are held more directly accountable for their actions and there is the possibility of enhanced support for victims and offenders in the community. To what extent are restorative justice programs effective in achieving their goals? McCold and Wachtel (1999) have recommended that measures of restorative justice include 1) the percent of victims and offenders expressing satisfaction with the way their case was handled, 2) the percent of victims and offenders who rate their experience as fair, and 3) the balance of ratings between victims and offenders. In this regard, restorative justice literature routinely shows high levels of satisfaction and fairness with the process across different restorative approaches, VOM and FGC, as reported by both victims and offenders. Unfortunately the concept of “measures of restorative-ness” is significantly limited by the exclusion of the clearly essential variable of recidivism. The goals of healing and restoration for victims, offender, and community are limited if there is no change in criminal behavior and increased community safety.
Is restorative justice an effective response to juvenile criminal behavior? There have been several comprehensive literature reviews of VOM that report varied effects of VOM on juvenile recidivism (Umbreit, Coates, Voss, 2002; Latimer & Klienknecht, 2001; Braithwaite, 1999; Marshall, 1999). These studies have investigated the impact of VOM on juvenile recidivism, primarily by comparing re-offense rates of VOM participants with non-participants (Umbreit, Coates, & Voss, 2001). Much of this research is limited by the lack of control groups, nonequivalent control groups, and self-selection bias of those who choose to participate in VOM and varied definitions of re-offense. In addition, these narrative reviews are limited due to the lack of clarity and inconsistency of how the literature was selected, and their inability to aggregate the empirical knowledge regarding recidivism and interpret discrepant findings in the literature.
Nugent, Umbreit, Winamaki, and Paddock (2001) conducted a rigorous reanalysis of recidivism data reported in four previous well-designed studies. The sample consisted of 1,298 juvenile offenders (619 participated in VOM and 679 did not). Results of the logistic regression showed that VOM youth recidivated at a statistically significant rate, 32 percent lower that non-VOM youth. In addition, when VOM youth did re-offend they did so for less serious offenses than non-VOM youth. This replication study is an important step in the literature in substantiating the effectiveness of VOM in reducing juvenile recidivism.
There have been four research studies evaluating the effectiveness of FGC on juvenile recidivism. The Wagga Wagga study by Moore and Forsythe (1995) used a single group pre-test/post-test design with 693 subjects to evaluate changes in re-offense in this sample. Results show statistically significant reductions in re-offense at nine months follow- up. However, when controlling for time, there was no treatment effect on re-offense.
McCold and Wachtel (1998) report on the results of a random assignment of 150 juveniles with property offenses to FGC or a control group and 75 juveniles with violent offenses similarly assigned to FGC or control. Although there was random assignment, there was non-equivalence between groups that limit the study results. Results indicate that juveniles with property offenses who participated in FGC actually had greater rates of recidivism than the control group at one year follow-up. Regarding violent offenses, there was a statistically significant reduction in re-offense for FGC participants. Further analysis of this data found a self-selection process that negated the hypothesis of a significant FGC treatment effect.
The McGarrell, Olivares, Crawford, and Kroovand (2000) study used an experimental design with random assignment of juveniles to experimental intervention, FGC (232 subjects), or control group (226), diversion programs. Results at six months found a statistically significant reduction in recidivism in FGC participants: re-offense 20 percent and diversion 34 percent. At twelve months, FGC participants had a 30 percent re-offense rate compared to diversion programs (41 percent), which was also statistically significant.
Sherman, Strang, and Woods (2000) used four experimental studies to evaluate the effectiveness of FGC compared to a control group. These studies examined 1) driving while intoxicated (N= 900), 2) juvenile property offenses (N=249), 3) juvenile shoplifting (N=143), and 4) violent offenses (N=110). Results found a reduction of 38 crimes (in a 100 per year in driving while intoxicated) for FGC. For property crimes, there was a small increase of 6 crimes in 100 per year for FGC participants. There was no significant difference between groups in juvenile shoplifting. Regarding violent offenses, there was a reduction of 38 (in a 100 per year) for FGC.
The overall methodological quality of restorative justice research shows considerable variety in the quality of studies, which makes it difficult to compare results across studies (Bradshaw & Roseborough, 2003; Latimer, Dowden, & Muise, 2001; Nugent et al., 2003). Meta-analysis, however, is a useful method for summarizing research findings across studies and synthesizing these findings in a more objective manner than expert opinion that leaves reviewer bias essentially uncontrolled (Rosenthal, 1999).
In a meta-analysis, the strength of the intervention effect on the outcome is described by the effect size. An effect size is a statistical method that was developed to evaluate in a standardized manner how much, on the average, a given treatment program reduced the severity of the target symptoms. The effect size method enables us to compare the efficacy of different types of treatment across studies. For example, an effect size of +.10 can be interpreted as the intervention accounted for 10 percent of the change in outcome.
Latimer, Dowden, and Muise (2001) conducted a recent meta-analysis on restorative justice interventions. However, adults and juveniles were included together in the analysis and there was no differentiation of the types of restorative justice interventions that were used in the study. Due to the lack of evaluative data regarding peacemaking circles, this study focuses on the effectiveness of victim offender mediation and family group conferencing.
The purpose of this meta-analytic study is 1) to synthesize the results of existing studies of the effectiveness of restorative justice dialogue practices on juvenile recidivism to determine the overall intervention effect of restorative justice, 2) to compare intervention effects between VOM and FGC on recidivism, and 3) to examine moderating variables that might affect rates of re-offense.
Methodology
Sample
A literature search was conducted following the guidelines described by Sowers, Ellis, and Meyer-Adams (2001). Two procedures were used to search for studies. First, computer searches were done of PsychInfo, Social Sciences Abstracts, Dissertation Abstracts, and the National Criminal Justice Reference Service databases. Key words included victim offender mediation, victim offender reconciliation, restorative justice, mediated dialogue, victim-offender mediation programs, peacemaking circles, sentencing circles, family group conferencing, community conferencing, police conferencing, and problem-oriented policing. Second, reference lists from each study were examined and experts in the field were contacted to identify unpublished research in this area.
To be included in the meta-analysis, each study had to 1) focus on juvenile offenders, 2) examine restorative justice intervention outcomes on recidivism, and 3) utilize a restorative justice intervention group and a comparison group. The search identified 33 articles in the area of restorative justice and recidivism. Of these, 19 studies met the selection criteria. Studies were excluded because they did not have a comparison group or included both adults and juveniles in the sample. The sample for this study included 11,950 juveniles from 25 different service sites.
Outcome Measure
The outcome measure in the study was reoffense. Re-offense was defined differently in these studies. Some defined re-offense conservatively as an offense for which the youth was adjudicated guilty during a one-year period after the original offense (Umbreit, 1993, 1994; Nugent & Paddock, 1996; Winamaki, 1997; Sneider, 1990). The other studies used a broader, more liberal definition of re-offense that was defined as any other contact with the criminal justice system.
Moderating Variables
Several variables were identified that might moderate the impact of restorative justice on re-offense. These include 1) quality of Restorative Justice and Juvenile Recidivism 17 research design, 2) type of comparison group, 3) type of offense, 4) definition of re-offense, 5) source of the study, 6) sample, and 7) length of follow-up. Differences in re-offense rates between groups could be caused by the lack of equivalence in the initial formation of the groups. Therefore it is important that methodological procedures such as random assignment, matching, and use of statistical evaluation of equivalence between groups are used in the study. Quality of research design was coded 1 = use of methodological procedures to ensure equivalent groups, and 2 = no methods were used.
Some comparison groups consisted of youth who refused to participate in restorative justice programs, while other youth were assigned to and participated in other traditional justice treatment programs. Differences in re-offense rates could be influenced by the nature of the comparison group. Comparison groups of youth who refused participation were coded 0 and groups that were assigned to other treatments were coded 1. Type of offense, property offense vs. person related offense, could moderate rates of re-offense. Samples that had only property offenses were coded 0 and those with property and person offenses were coded 1. Definition of re-offense was coded 1 if adjudicated guilty during the follow-up period and coded 0 if re-offense was defined as any official contact with law enforcement, court, or arrest. Research that reports negative or non-significant results, program evaluation reports, master’s theses, and dissertations are frequently not published and can bias results due to their exclusion from the literature. Published articles were coded 1 and unpublished research was coded 0. Samples in some studies consisted of only restorative justice program referred clients. This could bias toward re-offense rates if persons with more serious history of offenses were excluded. If a study contained all restorative justice referrals, they were coded 1 and if the sample included not all restorative justice referrals it was coded 0.
Analysis
The primary analyses consisted of computation of effect sizes as outlined by Cohen (1977). Effect sizes were computed as the mean of the criterion group, cognitive behavioral treatment, minus the mean of the contrast group, divided by the pooled standard deviation of the treatment and contrast groups. For studies that reported percentages, the effect sizes were computed using the probit transformation of differences in proportion to effect size calculations (Glass, 1981). For those studies that reported non-significant differences, but did not report means and standard deviations, a zero was entered for the effect size of that outcome measure. Effect sizes were adjusted to correct for bias attributable to studies with small sample sizes and to ensure these studies did not inflate overall effect sizes. Weighting procedures described by Hedges and Olin (1985) were used to combine effect sizes from different studies to give greater weight to studies with larger sample sizes that provide greater reliability.
The effect size reflects the distance the average restorative justice client was from the average contrast client expressed in standard deviation units. An effect size of 1.00 would indicate that the average restorative justice client would have been one standard deviation higher than the contrast group. Effect sizes approximating zero would indicate no differential advantage for either treatment. Negative effect sizes would suggest that restorative justice treatment was less effective than the contrast treatment.
The unit of analysis for this study was conceptualized in two different ways (Durlak, 1995). First, separate effect sizes were calculated for each study. Second, effect sizes were calculated across studies. In addition to the computation of effect sizes, comparisons of group means were done based on the moderating variables: 1) quality of research design, 2) type of comparison group, 3) type offense, 4) definition of re-offense, 5) source of the study, and 6) sample. Length of follow-up was correlated with re-offense effect sizes.
Results
The inter-observer agreement for assessing the inclusion criteria for the studies was good (ICC= .96). The sample consisted of 11,950 juveniles who received service in 25 different sites and four countries. The duration of follow-up ranged from nine months to 48 months with an average of M = 17.08, SD = 9.01.
Effect Size Analysis
The average effect size for all studies was M =.26, SD= .39. Restorative justice dialogue programs, VOM and FGC, contributed to a 26 percent reduction in recidivism. There were 15 studies with positive effect sizes, five studies with no treatment effect, and three studies with negative effect sizes. See Figure 1 for graphic depiction of the effect sizes. The comparison between VOM and FGC effect sizes found higher effect sizes for VOM (M = .34, SD = .46) than for FGC (M = .11, SD = .12). This difference was statistically significant (t (20) = 2.79, p =<.05).
Moderating Variables
Following the primary analysis of effect sizes, the first step was to test the homogeneity of effect sizes combined across all of the studies. As Durlak (1995) has noted, the Q statistic assesses whether the effects in the meta-analysis vary primarily due to sampling error or due to systematic differences among the studies and sampling error. If the effects of the group of studies are homogenous, it suggests that they come from the same population and analysis of group means and correlations is warranted. The Q statistic for all studies was Q=18.45, p > .05 ns, indicating homogeneity of the reviewed studies. The influence of six moderating variables on effect sizes was then examined. These variables were 1) quality of research design, 2) type of comparison group, 3) type of offense, 4) definition of re-offense, 5) source of the study, 6) sample, and 7) length of follow-up. Analysis of group means was done by t-tests. There was a significant difference in effect sizes based on type of control group. Studies that utilized a control group consisting of juveniles referred to a restorative justice intervention but who refused participation had significantly higher effect sizes (M = .46, SD = .41) compared to those control groups that were made up of juveniles participating in an alternative treatment, e.g. diversion programs (M = .11, SD = 31). This difference was statistically significant (t (21) = 2.36, p =<.05) There were no significant differences in overall effect sizes between groups based on comparison group, quality of design, type offense, definition of re-offense, source of the study, or sample. There was also no significant correlation between length of follow- up and overall effect size.
Results from meta-analyses can be positively biased in the estimation of treatment effects because journals rarely publish papers that report on non-significant or negative results. This enhances the possibility of Type I error in finding more positive results than would be the case if all existing studies were included in the review. Fail-Safe Ns for each group of effect sizes were calculated using the formula outlined by Rosenthal (1979). The Fail-Safe N represents the number of additional studies in a meta-analysis that would be necessary to reduce the mean effect size to .20, a small effect size (Wolf, 1986). The Fail-Safe Ns reported in this study indicate that confidence can be placed in the findings of the effect sizes (cf. Table 2).
Discussion
This is the first meta-analysis that examined the effectiveness of the two most prominent restorative justice dialogue programs in reducing juvenile recidivism. The use of meta-analytic methods provides a useful means for summarizing diverse research findings across restorative justice studies and synthesizing these findings in an objective manner. The use of an effect size is an easily interpreted way of assessing the strength of an intervention effect.
The average effect size of .26 found in this study represents an intervention effect that is double that of the previously reported effect sizes of .10 found in traditional justice programs (Lipsey, 1995). These results are particularly meaningful given the typical brevity of restorative justice dialogue interventions. They add to the empirical base of the effectiveness of restorative justice dialogue programs in reducing juvenile recidivism and support the use of restorative justice programs as empirically supported interventions for juvenile offenses.
The empirically supported practice movement is an international attempt to identify the best practices in a field of service that is based on the researched efficacy of an intervention. This movement is also grounded in the right of clients to know about and have access to effective treatments (Thyer & Meyers, 1999).
The significant difference in effect sizes between VOM (.34) and FGC (.11) have important implications for the future development of restorative justice practices. The effect of VOM on recidivism has been researched in 15 studies of which 11 show positive intervention effects on reducing recidivism, two show no treatment effect, and two show negative effect sizes. There is sufficient data to support VOM as a well-established, empirically-supported intervention for reducing juvenile recidivism (Chambless et al., 1998).
FGC research is currently limited to the four studies previously reviewed, of which only two show positive intervention effects. Using similar criteria of efficacy, FGC would be considered as a promising, but experimental, intervention for juvenile re-offense. FGC proponents need to continue research efforts to evaluate its effectiveness in reducing juvenile recidivism in order to make FGC more feasible and broaden the empirically supported options available in the field of restorative justice.
The type of control group has a moderating effect on intervention effect size. Studies that utilized a control group consisting of juveniles referred to a restorative justice intervention but who refused participation had significantly higher effect sizes than those control groups that were made up of juveniles participating in an alternative treatment such as diversion programs. This is in contrast to Latimer, Dowden, and Muises’ (2001) findings that alternative treatment control groups had higher mean effect sizes compared to non-participation control groups. Further research on the potential moderating influences of type of control is needed.
Three issues of the methodological quality of these research studies create difficulties in interpreting the results reported in this set of studies. First, only ten studies used random assignment, matching, or statistical methods to create equivalent groups. Second, half of the studies included in the meta-analysis used the broader definition of re-offense: arrest, contact with police, or violation of probation. This definition may increase the number of false positives regarding re-offense. The other studies used a narrowly defined measure of re-offense, adjudicated guilty. Use of this more conservative measure may decrease the number of false positives in re-offenses. Third, due to the voluntary nature of participation in most restorative justice dialogue programs, there is an inherent self-selection bias that makes interpretation of results difficult. The addition of measures that assess the youth’s motivation for participation may provide a means to control for differences in motivation and openness to mediation.
There are several issues that need to be addressed in future research in juvenile recidivism. First, it is essential to do more evaluation of restorative justice dialogue programs, particularly FGC and peacemaking circles, using random assignment and other methodological procedures to enhance equivalence of initial treatment and comparison groups. Second, the use of quantified measures of the severity and number of previous antisocial behaviors is needed to increase the validity of results and identify more accurately subgroups that may have differential responses to different restorative justice programs. For example, in most studies, the sample is described by property or person/violent offenses. If two juveniles are referred to restorative justice dialogue programs for similar property offenses, but one has a significant history of severe and frequent offenses and the other is a first time offender, they may look equivalent as a property offense, but the likelihood of re-offense is greater for the youth with the prior history. Use of a severity rating system (Nugent & Paddock, 1995) allows for greater accuracy in describing offense histories. Third, the use of other self-report and multi-informant approaches such as the Child Behavior Checklist (Achenbach, 1991) to measure juvenile behavior is important to broaden the scope of measuring delinquent behavior beyond re-offense. Fourth, while there were no significant moderating effects by type of offense, definition of re-offense, source of the study, and sample on effect size, further research using other types of designs are needed to identify potential moderating variables and begin to describe what works for whom in different restorative justice programs. Fifth, criminal justice programs or psychologically based treatment interventions are the common responses to delinquent behavior. It is an unfortunate dichotomy given the fact that many juveniles adjudicated guilty who participate in restorative justice programs also have co-morbid psychiatric conditions that need treatment. For those juveniles who are receiving treatment and participate in restorative justice, there is a potential factor of multiple treatments that confound interpretation of research results. On the other hand, it may be an important area of research to evaluate the effectiveness of a combination of restorative justice approaches and empirically supported psychological treatments.
In traditional retributive and rehabilitative models of justice, reduction of recidivism is the gold standard of outcomes. However, within the field of restorative justice there are concerns regarding which outcomes are truly restorative. Some authors have described a model for the evaluation of restorative justice programs that is more congruent with restorative justice values (Presser & Van-Voorhis, 2002). This model focuses on the process of restorative justice and less on traditional outcomes, as well as utilizing qualitative methods of inquiry. Others have emphasized that in restorative justice programs, recidivism is important, but not central, to the practice of restorative justice (McCold & Wachtel, 1998). They note that even if recidivism is not reduced, restorative justice programs could be justified if they meet other needs of victims, offenders, and the community.
In a time of accountability, tight budgets, and closing of restorative justice programs, restorative justice practitioners and policy makers need to utilize effectiveness data to make decisions regarding program development and funding. As McCold and Wachtel (1998) have concluded, “Restorative justice programs which reduce recidivism are to be preferred over programs which have no measurable effect on recidivism.”