By Laura Mirsky

“A Survey of Assessment Research on Mediation and Restorative Justice,” by International Institute for Restorative Practices director of research Paul McCold, has been published as a chapter in the book Repositioning Restorative Justice: Restorative Justice, Criminal Justice and Social Context. The paper provides an overview of 30 years of evaluation research of restorative justice programs from 1971 to 2001. The survey is limited to program assessments available in English and is representative of mediation and conferencing programs that have conducted and published the results of those assessments.Paul McCold Director of Research, IIRP

Said McCold, “This is one of a series of very informative articles published as a result of the Fifth Conference of the International Network for Research on Restorative Justice [held September 2001] and available from Willan Publishing [in Devon, England].” The book can be ordered through the website for Willan Publishing:

McCold's objective in producing the paper was, he writes, “to see what evidence we can bring to bear on the probable truth or falsity of restorative justice as a credible response to crime and conflicts in society. It is his view that restorative justice must be measured against existing practice. “ There is a tendency to compare new programs to perfection and to criticize them when they fail to reach the ideal. To succeed, restorative justice does not need to be perfect. To be preferred, it need only demonstrate superiority, on average, to traditional adjudicatory approaches.”

“...If a justice program is effective, it should be possible to measure these effects.”

Commented McCold, “I think this paper is especially significant and really deserves to be highlighted. It brings together in one place and in a dramatic graphic form the results of research on restorative justice from all over the world and summarizes what currently can be said that we know scientifically about restorative practices.” Askedwhat the research shows overall, McCold said, “that restorative practices are a very popular and effective alternative form of responding to criminal and civil conflict.” One reason he wrote the paper, he said, was to provide “the range of outcomes achieved so programs that are evaluating themselves will know what constitutes normal results.”Writes McCold: “Primary restorative justice practices bring offenders and those directly affected by their behavior together to agree mutually on a plan to repair the harm done. These practices include at least two distinctive versions of mediation-'community mediation' and 'victim-offender mediation'; two distinctive versions of conferencing- 'family group conferencing' and 'community group conferencing'; and at least three distinct versions of peace-making circles.

Writes McCold: “Primary restorative justice practices bring offenders and those directly affected by their behavior together to agree mutually on a plan to repair the harm done. These practices include at least two distinctive versions of mediation-‘community mediation’ and ‘victim-offender mediation’; two distinctive versions of conferencing- ‘family group conferencing’ and ‘community group conferencing’; and at least three distinct versions of peace-making circles.

“A consistent finding, from the initial to the most recent assessment of restorative justice programs, is that when the parties come together, the success rate is high.”

In the paper’s summary, McCold writes, “Results from the 98 restorative program samples and 21 court samples surveyed provide strong empirical support for a few generalizations. Disputing parties typically hold positive views of restorative justice programs; they feel satisfied with the process and would return if a dispute arose in the future. Studies involving different settings and types of disputes found disputants perceived the outcomes of restorative justice to be significantly fairer than those of court proceedings.”

The paper's conclusions include:

    • There is no significant public opposition to restorative justice.
    • There is a high degree of support among victims of crime and the public for offender reparation and for victims to have an opportunity to meet with their offender.
    • While participation rates vary widely from program to program, most victims and offenders will choose to participate, given the opportunity.
    • Offenders are somewhat less likely to participate than victims.
    • Participation rates differ for type of offenses, age of offender, type of victim and the relationship between victim and offender.
    • Many mediation programs offer the option of indirect mediation (conciliation).
    • Victims rate direct mediation higher than indirect mediation or court, and report being less fearful when they meet their offender face to face than when they are not given the opportunity to do so.
    • At least one conferencing program, the New Zealand government's, requires participation of qualifying juvenile offenders. This program and some other conferencing programs proceed with or without the attendance of the victim. Victim participation and satisfaction rating for New Zealand's youth-justice family group conferencing was lower than for most restorative programs.
    • Where victim and offender participate in restorative programs, the rates of agreement and compliance with that agreement are very high.
    • There is no consistent relationship between a program's participation rate and either the agreement or compliance rates.
    • There is no intrinsic limitation to the type of dispute or disputants for which restorative justice can bring a reparative response and no empirical limitation reported in the evaluation research:
        • Mediation and conferencing have reported successful resolutions in violent and property cases, adult felony and first-time juvenile cases, and between strangers or among family members.
        • For everything from consumer complaints to domestic violence, program evaluations have documented the positive outcomes of restorative justice.
        • Justice does not need to be a trade-off between victims and offenders.
        • Both victims and offenders rate restorative justice as more fair and satisfying than court. This is especially true for victims, and for models that directly involve communities of care.
        • Several recent restorative justice programs report fairness and satisfaction ratings from both offenders and victims above 95 percent.
    • Reoffending rates for offenders is no higher for restorative justice than it is for court adjudication:
        • The effects of the program on reoffending depend upon crime type and are related to participation rates.
        • While there appears to be a strong self-selection effect for the voluntary programs (i.e., those offenders less likely to reoffend are more likely to participate), reoffending following restorative justice processing seems to be reduced more among offenses against persons than among property offenses or victimless offenses.

Difficulties arise when one attempts to compare different types of restorative justice programs to each other, writes McCold. One of the tools used to compare different programs is the service delivery rate formula, the estimated average number of cases processed in one year per 10,000 people in a given geographic area, which is a way to compare programs of varying durations, serving diverse populations. Explaining the logic behind the service-delivery rate calculation, McCold writes, “Assessment research varies from single-year program evaluations to multi-year summaries of programs for a whole country. To standardize comparison across assessment schemes, an annualized number of mediations/conferences was estimated for each program sample.”

The paper includes several graphs depicting service delivery rates for various different sizes of program samples. This is significant, writes McCold, because even though small programs process relatively few cases per year, “the impact of the program may be hidden behind the absolute numbers served. A small program serving a small community or directed towards a small finite population may have a larger aggregate effect than a small program serving a large metropolitan area.”

The paper's Figure 4.3 (not shown here) depicts estimated service delivery rates for small program samples - those processing 100 cases per year or fewer. The two highest service delivery rates were for the community mediation center (CMC) scheme in Rotterdam, Netherlands, which provided an intensive program to the residents of a target housing district “with an astounding service delivery rate of 135,” and the small town of Sparwood, British Columbia, Canada, which processed every juvenile arrest using a community group conference (CGC), and had a service delivery rate of 120.

Estimated service delivery rates for mid-sized program samples, processing between 100 and 500 cases per year, are depicted in the paper's Figure 4.4 (not shown here). The two highest service delivery rates in this category were seen in CMCs in the inner-city neighborhood of Dorchester, in Boston, Massachusetts, with a rate of 80 per year per 10,000 population, and the CGC program in Wagga Wagga, New South Wales, Australia, with a rate of 43.

Figure 4.5 (not shown here) depicts restorative justice program samples with more than 500 mediations or conferences per year, including city-based CMC samples and state and national CMC, victim-offender mediation (VOM) and family group conference (FGC) samples. The graph shows that city-based programs provided more intensive service delivery rates than state or national programs. Although France provided 8,000 mediations per year, the service delivery rate was only 1.5. At the other end of the spectrum, the graph shows that the city of Columbus, Ohio, U.S.A., had a service delivery rate of 68 CMCs per 10,000 population. None of the state or national programs had a rate higher than 20.Figure A. Estimated annual meetings: state and national program samples.

The actual number of conferences held can be directly compared to service delivery rates by examining the paper's Figure 4.6 (shown here as Figure A), and the paper's Figure 4.7, (shown here as Figure B). (All graphics have been reproduced with permission of Willan Publishing.) Figure A shows that New York State's 28 CMC pro-grams held a total of 16,000 mediations in one year, and New Zealand held 5,000 FGCs.Figure B. Estimated service delivery rates per 10,000 population: state and national program samples.






Figure B shows the program service rate for these samples, revealing a different pattern. New York State had a program service delivery rate of 8.7, but “New Zealand, with the highest service delivery rate of 13.7, mandates a family group conference for nearly 100 percent of their serious juvenile offenders, or about 20 percent of all young offenders.”

Figure C. Offender perceptions by victim perceptions by program sample.Figure 4.17, (shown here as Figure C) illustrates a collection of surveys of participants' satisfaction with various justice processes, including different types of restorative justice programs, as well as the conventional criminal justice system. McCold writes that there is no accepted standard for measuring participant satisfaction, but that primary topics and questions covered in the surveys compared in figure C included variations on the following:

    • Satisfaction with the way their case was handled.
    • Satisfaction with the outcome of their case.
    • Satisfaction with the facilitator.
    • Fairness of the process.
    • Fairness of the outcome.
    • Neutrality of the facilitator.
    • Are they glad they participated?
    • Would they recommend this program to others?
    • Would they participate again under similar circumstances?

McCold writes that these questions were asked of victims and offenders in mediation assessments and also of the parents of offenders and other participants in conferencing programs. In addition, the same type of questions were asked of two groups of participants in conventional court programs: (1) those who were referred to a restorative justice program but did not take part, and (2) those who were not referred to a restorative justice program - effect, a matched comparison group. The questions asked and measurements used for these assessments varied widely. To arrive at Figure C, within each program the question that received the maximum rating from victims was compared to the question that received the maximum rating from offenders.

“The U.S.A. has the largest, most expensive criminal justice system in the world. In spite of huge government spending on a variety of programs and approaches, the American public remains an unsatisfied customer.”

About the information illustrated in Figure C (see previous page), McCold writes, “Ideally, restorative justice programs would transform the relationship between parties and produce equally high ratings from both victims and offenders. Programs with much higher ratings from offenders than from victims might be considered offender focused, and vice versa.”

Figure C shows that “all but three of the court samples are offender focused, in that the process is more preferred by offenders than by victims. All but one restorative justice program demonstrate a more balanced approach.” McCold concludes that these findings generally confirm that “disputing parties typically hold positive views of restorative justice programs; they feel satisfied with the process and would return under similar circumstances in the future. Assessments involving different settings and types of disputes found that disputants perceived the outcomes of mediation hearings to be significantly fairer than those of court proceedings.”

Among the paper's highlights, said McCold, are data about public opinion regarding the current justice system's approach to crime and about attitudes toward restorative approaches to crime. “The U.S.A.,” he writes, “has the largest, most expensive criminal justice system in the world. In spite of huge government spending on a variety of programs and approaches, the American public remains an unsatisfied customer.” He presents results of a 1998 study by Schulman, Ronca and Bucuvalas, Inc., which surveyed a randomly selected sample of 4,015 adults in nine northeastern states of the United States, asking, “A number of states are considering significant changes in the way the criminal justice system works. Without knowing any specific details, do you like the idea of totally revamping the way the system works, or do you feel the present system works well enough the way it is?” The survey showed that 75 percent of the public, including victims of crime, favored revamping the system and 16 percent felt the system works well enough as it is.

Respondents were also asked what they thought about the idea of victims and offenders getting together to talk about “why the offender committed the crime and whether the offender accepts the consequences of what he or she did.” A very large percentage of both crime victims and the general public expressed high support for the idea of victim-offender encounters. McCold writes that widespread public support for restorative approaches to crime has been replicated in surveys conducted in New Zealand, the United Kingdom and Germany. These surveys concluded that the public at large displays an intrinsic acceptance of private conflict-solving following an offense, and that it is a myth that the public demands tough punishment.

About these results, McCold commented, “It all has to do with how you ask the public how punitive they are. If you ask them about restorative practices they tend to be very supportive of it. If you also ask them about punishment, they all want more of it. It's part of the strangeness of the public. But it explains why there's no strong opposition to restorative justice anywhere. There's no organized opposition, no people saying that you shouldn't be doing this. Every time we've asked the public about this in a reasonable way, the support numbers are huge.”

Asked what constitutes a “reasonable way” to ask, McCold said, “Here's an unreasonable way: 'Should a serial murderer and child molester be allowed to escape punishment by attending a conference where they are confronted or confront the victims of their crimes?' Then everybody would say, 'No, I'm against it!' But a lot of research on public opinion has been done that way, because it's been done to show how tough the public is. So if we want to measure how punitive the public is, you start with the wrong framework, and you shape your question and then determine your outcome.”

The right approach, said McCold, is to ask these questions in “a neutral way. The way I think they did, for example, in the two questions from the public survey of the northeastern United States [noted above]. I think that's a fair way to ask it. Because then what it is, as I see it, is a service for victims. I think we've just over-focused on the offender and recidivism rates and all the things offenders get out of restorative justice - what it could do to offenders and for them. I think those are all kind of missing the mark.” McCold said that we need to focus instead on “why we would want to do this for victims. It's a very different question: Which victim should be eligible for it? It's a better question.”

Continued McCold, “I'm not surprised to find the high support levels in the public for restorative practices. But it is a surprise to others and probably is an unknown thing to most politicians. How many elected officials from those nine northeastern states know this when they're campaigning and talking about getting tough on crime? I think if the campaign managers knew these results they might fashion a different statement.”

Another of the paper's highlights, said McCold, is the mediation study by Mark Umbreit and Warner Roberts comparing direct and indirect mediation, which sampled 68 victims and 51 offenders among the referrals to VOM programs in Coventry and Leeds (United Kingdom) who participated in direct mediation, indirect mediation and court. He explained the difference between direct and indirect programs: “In the U.K., one of their favorite ways of doing victim-offender mediation is through a shuttle mediator, so that the victim and offender actually never meet or talk directly to each other, but a third party shuttles between them doing negotiations about an agreement. I think of it as 'shuttle negotiation' more than mediation. Only about one in eight cases end up in a direct session.”

He writes, “Offenders rated the satisfaction and fairness of indirect mediation as high or higher than direct mediation, and both higher than court. Victims rated indirect mediation as fair and as satisfying as court, but much less fair and satisfying than direct mediation. Both victims and offenders were more satisfied with the outcomes from direct mediation than indirect mediation, and both were more likely to rate their participation as voluntary for face-to-face mediation than third-party conciliation. Most significantly, victims whose cases were processed by court were more likely to report fearing revictimization (33 percent) than victims whose cases were conciliated (21 percent), and were least likely to fear revictimization after meeting their offender in direct mediation (11 percent).”

In fact, said McCold, one of his objectives for writing the paper was to put the above results into perspective, because they have been used by some to devalue restorative justice. In shuttle mediation, he said, victim participation and satisfaction rates are low because that process doesn't put people together, and thus does little to restore the emotional or relational harm. These results, he said, “really make our point that you don't get very good results [with indirect mediation]. The offender loves the indirect stuff. Victims don't like it. For them, they might as well go to court. It supports the argument about our restorative practices typology.” To read an article about the restorative practices typology, “In Pursuit of Paradigm: A Theory of Restorative Justice,” by McCold and Ted Wachtel, please go to:

“The future development of restorative justice should be accompanied by carefully controlled scientific assessments using standardized measures across a variety of settings and practices.”

In conversation, McCold also pointed out the results of the Reintegrative Shaming Experiment (RISE) in Canberra, Australia. This experiment, he wrote, “was specifically designed to detect the effect of police-facilitated CGC [community group conferencing] on offender recidivism for four offense types (juvenile personal property, juvenile retail theft, young violent offenders and 'drink driving' - adults driving over the legal limit for alcohol). The Australian National Police facilitated nearly 1,300 community group conferences over a five-year period ending July 2000.... Recidivism patterns of both juvenile personal property and shoplifting offenders revealed that the deterrent effect of conferencing and court was equivalent. More dramatically, reoffending rates by violent offenders dropped significantly among the CGC group by 38 percent in the 12 months following the conference (Sherman et al., 2000). This represents the first scientifically defensible evidence that CGCs can reduce recidivism, at least among violent offenders.”

Regarding the future of restorative justice research, McCold commented, “I continue to compile data as it comes out. Each time a new program releases an evaluation I get a copy and add their numbers into my database. So I've got a growing database of empirical results, and I always continue to keep that going.” In his paper, he makes several recommendations for research in this field, writing, “In spite of many program assessments, research on restorative justice practice today is a mile wide but only an inch deep.... The future development of restorative justice should be accompanied by carefully controlled scientific assessments using standardized measures across a variety of settings and practices. If research on restorative justice practice is to evolve, findings need to be comparable across programs. Assessments need to report conditions for program eligibility, the number of cases referred, the number of cases participating, reasons for non-participation, number and nature of agreements reached and the rate of compliance with agreements. Results need to be disaggregated by offense type and disputant relationship. Programs providing indirect conciliation and conferences without victim presence should report these data separately. Referred but nonparticipating cases need to be included in follow-up surveys and incorporated into the analyses of program outcomes. Participant satisfaction/fairness findings reported should allow for computing percentage positive response. Without concerted governmental guidance, program assessments are likely to continue to reconfirm already well established findings while contributing little to the cumulative understanding of the practice of restorative justice.”

McCold recommended two other major surveys of restorative justice research. International Review of Restorative Justice, by David Miers, was commissioned by the Crime and Criminal Justice Unit of the Home Office Research, Development and Statistics Directorate, United Kingdom. This review provides an overview of restorative justice programs in 12 European jurisdictions: Austria, Belgium, the Czech Republic, Denmark, Finland, France, Germany, the Netherlands, Norway, Poland, Slovenia, and Spain, together with summaries and examples of programs in Australia, Canada, New Zealand and the United States. Other European jurisdictions included in the study were Ireland, Italy, Russia and Sweden. The review, reads its introduction, “draws some lessons about good practice in restorative justice provision. [It] places the work currently being undertaken within wider theoretical debates about the nature and scope of restorative justice, and highlights some of the strengths and weaknesses of evaluative research into its impact.” This report may be downloaded at:, or ordered by emailing: This email address is being protected from spambots. You need JavaScript enabled to view it..

Victim-Offender Mediation In Europe: Making Restorative Justice Work, edited by Tony Peters, Leuven University Press, is an initiative of the European Forum for Victim-Offender Mediation and Restorative Justice and draws from presentations at its first conference, in Leuven, Belgium, 1999.

The first part of this collection considers victim-offender mediation and restorative justice from a theoretical point of view, with essays by leading scholars in the field. The second consists of overviews of victim-offender mediation in the eight European countries in which it was currently most developed (Austria, Belgium, Finland, France, Germany, Norway, Poland and the United Kingdom). The book can be ordered at:

Graphics and citations from “A Survey of Assessment Research on Mediation and Restorative Justice,” by Paul McCold, used by permission of Willan Publishing, Culmcott House, Mill Street, Uffculme, Cullompton, Devon EX15 3AT, UK. Tel: 44 (0)1884 840337. Fax: 44 (0)1884 840251. Email: This email address is being protected from spambots. You need JavaScript enabled to view it..


Schulman, Ronca and Bucuvalas, Inc. (1999, August). What Do we Want (and What are we Getting) from the Criminal Justice System? Comparing the General Public's Expectations and Perceptions with Crime Victims' Experiences. Conducted for Council of State Governments Eastern Regional Conference, New York, NY.

Sherman, L., Strang, H. and Woods, D. (2000). Recidivism Patterns in the Canberra Reintegrative Shaming Experiments (RISE). Centre for Restorative Justice, Research School of Social Sciences, Australian National University, (

Umbreit, M. and Warner Roberts, A. (1996). Mediation of Criminal Conflict in England: An Assessment of Services In Coventry and Leeds. Minneapolis, MN: Center for Restorative Justice and Mediation, School of Social Work, University of Minnesota.

Restorative Works Cover
Restorative Works Year in Review 2023 (PDF)

All our donors are acknowledged annually in Restorative Works.