The Effects of Restorative Justice Programming:
A Review of the Empirical



An abundance of non-traditional justice programs have been implemented throughout the United States, Canada, Europe, New Zealand, Africa and Australia.  Many are considered restorative in nature; however, these programs may not fully conform to restorative principles. The scope of this paper will be on those programs that have adopted the aforementioned principles. That is to say, the empirical results that are discussed in this section are from programs that attempt to restore the relationship between the victim, the community and the offender and attempt to repair the harm caused by crime.

3.1 Research Issues

In the present criminal justice environment of rapid change, research is essential to the success of any new movement or ‘wave’ such as restorative justice (Bonta, Wallace-Capretta & Rooney, 1998; La Prairie, 1999).  Unfortunately, empirical scrutiny of the outcomes of such movements are rarely encouraged.  What do we actually know about the effectiveness of restorative programming?  And how do we define success?

There are several obvious definitions of a successful program.  First, since public safety remains the paramount concern of the criminal justice system, programs should attempt to reduce recidivism.  If a program were to actually increase the chances of further criminal behaviour, most would agree that this would not be a success.  Second, the needs of victims should be adequately addressed.  This is easily measured through controlled experiments testing the satisfaction levels of victims in the traditional system compared to a restorative program. Third, the effects of a program on the community should be considered.  For example, does the program reduce fear of crime and increase the perception of safety within a neighbourhood?

The effects of restorative justice are not solely limited to the participants.  The criminal justice system, as a whole, may be significantly affected by the proliferation of restorative practices. First, what are the financial costs associated with restorative justice programming compared to traditional responses to crime?  Second, how are the roles of the actors within the justice system affected?  It is reasonable to assume, for example, that the options now available to a police officer or crown attorney when dealing with an offender are different with the increasing accessibility to restorative justice programming. It is also reasonable to assume that the role of duty counsel will be affected as more cases are diverted away from the courts. Will parole decision makers be more likely to grant parole to an offender who has participated in a pre-release restorative justice program? And should this be a factor? Third, the collection of crime statistics will ultimately be affected by the increasing number of pre-judicial alternatives available to offenders.  Currently, crime statistics are collected primarily through official court surveys and police charge data.  As more and more criminal behaviour is dealt with outside the system, the accuracy of the official data becomes questionable.  It is useful to organise these outcomes using the following overarching framework.

Table 3.1 Research Framework
PARTICIPANT LEVEL Victims Offenders Community members Satisfaction Recidivism, satisfaction Fear of crime, sense of security
SYSTEM LEVEL Cost-benefit analysis Data collection Criminal justice system Financial costs, net-widening Crime statistics Police, attorneys, courts, corrections

There are also a number of moderating variables that may affect the outcomes and processes of restorative justice programs and should therefore be examined.  Ideally, the most comprehensive method of synthesising the literature would be through the use of meta­analytic techniques.  Unfortunately, there is not an adequate number of studies using comparison groups to invest the necessary resources at this time.  Table 3.2 highlights a number of moderating variables that could be examined in the future.  Each one of these variables can ultimately affect the outcomes of a program.  For example, the age of the offender and the seriousness of the offence might be related to recidivism rates or victim satisfaction rates. Or the level of training of the mediator and the involvement of family members might effect a participating community member’s subsequent fear of crime.

Table 3.2 Moderating Variables
ENTRY POINTS Police Crown Courts Corrections
MODELS OF PRACTICE Conferences Mediations Circles
VICTIM Age of victim Relationship to offender Gender of victim
MEDIATOR Level of training Voluntary or paid position Justice versus social service
COMMUNITY INVOLVEMENT Mediator only Family involvement Restricted community involvement Open community involvement
OFFENCE Seriousness of the offence Family violence Sexual offences
OFFENDER Age of offender Criminal history Gender of offender

3.2 Empirical Effects at the Participant Level

3.2.1 Recidivism

To date, the number of evaluations that examined the issue of recidivism is insufficient to form any definitive conclusions.  In fact, the research methods utilised in most studies were not rigorous and lacked randomised control groups.  And where random assignment was used, the voluntary nature of the programs still created a self-selected treatment group as referred subjects could simply refuse to participate.  This bias and subsequent lack of adequate comparative recidivism rates offers limited generalisable results.  The findings that are available, however, tend to indicate slight reductions in the recidivism rates of offenders within restorative justice programs compared to the traditional system.  The following data represent a sample of the empirical results in the literature.

  • The results of a meta-analysis completed by Bonta, Wallace-Capretta & Rooney (1998) indicated slight reductions in the recidivism rates of offenders participating in programs using community service, restitution or mediation compared to programs without these features.  While not directly an evaluation of restorative justice programs, these results do indicate the possibility of decreases in recidivism for programs with restorative features.
  • Morris and Maxwell (1998), in a family-group conferencing study with young offenders in New Zealand, found a reconviction rate of 26%.  While this is encouraging, considering the program dealt with moderate and serious offences, there was no comparison group used in this study to place the results in context.
  • In a study by Nuffield (1997), which looked at adult victim-offender mediations in Saskatoon, the recidivism rates of the mediated group were slightly higher than the comparison group.  Nuffield noted, however, that the mediated group of offenders had a larger proportion of cases with a prior record, which is a strong predictive factor for recidivism.
  • The evaluation of the Restorative Resolutions Project in Manitoba (Bonta, Wallace-Capretta & Rooney, 1998) used matched comparison groups and found statistically significant reductions in the recidivism rates of project participants.
  • Prenzler and Wortley (1998) found a very low rate of recidivism (7%) within the Queensland Community Conferencing project for young offenders.  There was, however, no comparison group and the follow-up time period for measuring recidivism was relatively short.
  • Recidivism was examined in four victim-offender mediation sites in the U.S. using comparison groups of referred but non-mediated offenders and non-referred offenders. Eighteen percent of the mediated group of offenders re-offended within one year compared to 27% of the non-mediated offenders (Umbreit, Coates & Kalanj, 1994).  In addition, 41% of the new offences committed by mediated offenders were classified as ‘less serious’ compared to only 12% of the non-mediated group.
  • McCold and Wachtel (1998) evaluated a police led family group conference project with young offenders using random assignment and found that the recidivism rates of project participants after 12 months was 20% while those who refused to participate demonstrated a 48% recidivism rate.  McCold and Wachtel also found, however, that the randomised control group of non-referred offenders had a 35% recidivism rate suggesting that there was little additional treatment effect beyond a self-selection effect.

There is a clear need to better understand the effects of restorative justice programs on re-offending.  While some data were available on victim-offender mediation programs and conferences, we could not locate data on the recidivism rates of offenders who have participated in sentencing circles.  Conferences quite frequently occur at the pre-conviction entry point whereas circles are often organised at the post-conviction stage.  The experiences of the offenders and their penetration into the traditional system are therefore different. If restorative justice programs are to be considered a reasonable alternative to certain aspects of the formal criminal justice system, more research is required on the long-term effects along all four entry points.

3.2.2 Victim satisfaction and perceptions of fairness

It is clear that victims tend to be satisfied following their involvement in a restorative justice program. This is perhaps the most critical piece of evidence to support the development of restorative approaches.  Programs using restorative principles achieve the central goal of addressing the needs of victims.  It is also likely, although somewhat less so, that victims in the traditional justice system are less satisfied than victims in a restorative program.  Again, however, the issue of self-selection should be considered.  It should be noted as well that there were some cases that did result in victim dissatisfaction.  These were typically a result of the offender’s failure to follow through with the agreed restitution, however, and not a reflection of the utility or success of the restorative process.

  • Umbreit, Coates and Kalanj (1994) found that 79% of mediated victims were satisfied with the processing of their case compared to 57% of the victims within a court sample.  The mediated victims were also more likely to perceive their case to be handled fairly by the justice system (83% vs. 62%).
  • In their interim report on a victim-offender conferencing program for young offenders in Washington County Minnesota, Umbreit and Fercello (1997a) found that all the victims were satisfied with the handling and the outcome of their cases.  Moreover, 100% of the victims felt that the restitution agreement was fair for them and 80% felt the agreement was fair for offenders.
  • An evaluation of victim satisfaction and perceptions of fairness at twelve different family-group conferencing sites in Minnesota found satisfaction levels with the process and outcome between 93% and 95% (Fercello & Umbreit, 1998).
  • Prenzler and Wortley (1998) found that almost all of the victims in a young offender project were satisfied with both the conference process and the agreements.
  • A study of family-group conferencing victims found that only 49% were satisfied with the program, although the reason most frequently reported for their dissatisfaction was a failure to receive the appropriate restitution (Morris & Maxwell, 1998).
  • A number of studies have found that the vast majority of victims would repeat the restorative justice process again and would recommend it to others (Chatterjee, 199 Coates & Gehm, 198 Umbreit, Coates, Kalanj, Lipkin, & Petros, 1995; Umbreit & Fercello, 1997a; Umbreit & Fercello, 1997b).

3.2.3 Offender satisfaction and perceptions of fairness

There is some indication that providing offenders with a more satisfying experience within the justice system may help to lower recidivism rates (Strang, Barnes, Braithwaite & Sherman, 1999). The traditional ‘professionalised’ justice system offers offenders very little opportunity to provide input into the court process and rarely allows them the chance to make amends through meaningful restitution.  Restorative programs, on the other hand, promote a setting in which all parties, including the offender, engage in meaningful participation and arrive at a collective solution for reparation.  A fair amount of research has been undertaken to evaluate offenders’ satisfaction levels and their perceptions of fairness after participating in restorative justice programs. The vast majority of these studies indicate that a high percentage of offenders are both satisfied with restorative programs and see these programs as being fair. Research also indicates that offenders find restorative programs to be more satisfying and fairer than the traditional criminal justice system.

  • Umbreit, Coates, and Kalanj (1994) found that those offenders who had participated in mediation were more likely to be satisfied by the process (87% vs. 78%) and see it as being fair (89% vs. 78%) than those who had not participated in mediation.  A high percentage of mediated offenders also expressed satisfaction with the outcome (90%) and perceived their restitution agreement as fair (88%).
  • A study conducted at two mediation program sites in England, by Umbreit, Warner, Kalanj, and Lipkin (1996), reported that the mediated group of offenders were more likely to be satisfied by the criminal justice system (79% vs. 55%) and perceive the system as being fair (89% vs. 56%) compared to non-mediated offenders.
  • A study conducted in New Zealand found that 84% of offenders were satisfied with the outcome of their family-group conference (Morris & Maxwell, 1998).  The researchers argued that the high level of satisfaction may have resulted from relief that the offenders did not receive a harsher penalty.  This conclusion was supported by empirical data that show offenders who received tougher penalties were three times more likely to express dissatisfaction with the outcomes than offenders who received less severe penalties.
  • Strang, Barnes, Braithwaite and Sherman (1999) found evidence which indicates that offenders in the Reintegrative Shaming Experiment in Australia find restorative programs to be fairer than going through the court system.  Seventy-two percent of offenders who were part of the conferencing process compared to 54% of offenders who went through the court system felt that the outcome of the process was fair.

3.2.4 Community effects

While restorative justice theory consistently describes a process of engaging the victim, the offender and the community, very little information exists regarding the community component of this triad. There were data in the literature on the positive effects of restorative justice for the parents of young offenders and for participating police officers and school officials.  There were also indications that the restorative justice experience facilitated closer relationships among participants and created a stronger sense of control and safety within communities. It is clear, theless, that this is a crucial gap in research.  Very little is actually known about the short- and long-term effects of restorative justice programs on community participants, including their sense of security, perceptions of the criminal justice system, or fear of crime.

  • Roberts (1995) found that victims improved their relationships with friends and family members following a victim-offender mediation.  The study also found that all participants reported positive changes in their views on crime and victimisation.
  • Following the Bethlehem Pennsylvania Police Family Group Conferencing project, police officers’ attitudes towards crime remained the same, but the officers did report a more community-oriented and problem solving focus in their work (McCold & Wachtel, 1998).
  • Cameron and Thorsborne (1998) reported that the majority of conference participants in an educational setting had closer relationships with other participants after the program. In addition, the school administrators felt that conferencing reinforced school values, created positive perceptions of the school among the participants’ families and transformed their disciplinary practices from a punitive to a more restorative approach.
  • Supporters of both victim and offender conference participants reported that the restorative justice programs created a greater sense of control and security in their community as well as restored harmony (Chaterjee, 1999).

3.2.5 Restitution agreements and completion rates

Restitution agreements are an important aspect of restorative programs.  These agreements, upon completion, act as a visible symbol that the offender accepts accountability for the offence and actively takes responsibility to repair the harm.  Typical restitution agreements involve financial compensation, community service, and/or service for the victim.  The available research indicated that there is a high rate of negotiated and completed restitution agreements for restorative justice programs.  Additionally, studies that compare restorative cases and court cases show a significantly higher rate of negotiating and completing restitution agreements in restorative programs. It appears then, that restorative justice programs achieve another one of their major goals - repairing the harm caused by crime.  It also appears that victim satisfaction is directly related to the extent to which these agreements are fulfilled.

  • Morris and Maxwell (1998) found that 85% of young offenders in family-group conferences in New Zealand agreed to carry out active penalties (such as community service).
  • At four different Canadian restorative program sites, Umbreit, Coates, Kalanj, Lipkin and Petros (1995) found that restitution agreements had been successfully negotiated in 93% of cases.
  • In their preliminary findings from an experimental comparison of conference and court cases, Strang, Barnes, Braithwaite and Sherman (1999) indicated that 83% of victims in conference cases versus 8% of victims in court cases, received both reparation and an apology from the offender.
  • In Coates and Gehm's (1989) empirical assessment of victim-offender reconciliation programs, they found that 98% of victims and offenders agreed upon restitution.  Of these agreements, there was a completion rate of 82% for financial restitution and 90% for service restitution.
  • In a national survey of 116 victim-offender mediation programs in the U.S., extensive phone interviews with program staff indicated that in 87% of cases a restitution agreement was reached, with a 99% completion rate (Umbreit, Fercello, & Umbreit, 1998).
  • In a comparison of two mediation programs matched with two court sites, Umbreit, Coates and Kalanj (1994) reported a restitution completion rate of 81% for mediated cases and 58% for court cases.  When examining the type of restitution agreements reached, the researchers found that 58% were financial in nature, 13% were for personal service to the victim and 29% had a community service component.
  • In a preliminary evaluation of victim mediation programs in California, researchers found that of all cases in which an agreement was reached, 97% of the contracts were completed or currently still active.  The small number of failures were exclusively from cases dealing with property offences rather than more serious crimes against the person (Niemeyer & Shichor, 1996).

3.3 Empirical Effects at the System Level

3.3.1 Cost-benefits and net-widening

The costs of restorative justice programming are predictably lower than the traditional system. Volunteers typically mediate sessions, cases can often be dealt with in a few hours and most offenders do not require legal representation.  Given the nature of a large proportion of restorative programs (i.e., accepting only minor and first time offenders), there is, however, the possibility of ‘net-widening’ whereby additional offenders are drawn into the criminal justice system.  Due to the high variability of criteria for referrals to a restorative justice program, and the four possible entry points, it is difficult to accurately determine if ‘net­widening’ is a problem. Would the inclusion of additional offenders into the system, who may have otherwise received an informal or formal cautioning, actually increase overall expenditures?  The answer is probably yes. Of those programs that are evaluated within the literature, however, there is some indication that the offence criteria being used for referrals are actually rather broad and include an increasing number of more serious offences, such as assault and impaired driving, that would not typically receive cautions.  But for the large majority of offences that are less serious, does it cost more for an offender to be tried and subsequently supervised in the community by a probation officer than a referral to a restorative justice program?  We could not locate a comprehensive comparison of the traditional system and a restorative justice program to adequately answer this question.  We found rather limited data on the costs of programs and on the issue of net-widening in the literature.

  • Umbreit, Coates and Kalanj’s (1994)  research into mediation programs in the U.S. found that a referral to the program cost approximately $230 and an actual mediation session cost less than $700.
  • In a survey of 116 mediation programs across the United States, program budgets ranged from $1 to over $400,000 with a mean cost of $55,077 (Umbreit, Fercello & Umbreit, 1998).
  • Research conducted on the Restorative Resolutions Project in Manitoba found that most offenders referred would have received a prison sentence through the traditional system (Bonta, Wallace-Capretta & Rooney, 1998).  The authors concluded, therefore, that net-widening was not an issue.

3.3.2 Criminal justice system

Unfortunately, we could not locate published research on the effects of restorative justice on the criminal justice system.  This is a significant gap in our current knowledge.  We do not know how the increasing number of restorative justice programs will affect the role of police, attorneys, or court and correctional officers.  The formal criminal justice system is, in all probability, experiencing significant changes as we move towards a secondary community-based stream of justice in Canada.

3.3.3 Data collection

This area of research also appears to be neglected in the literature.  Are the decreases we are witnessing in the Canadian crime rate a reflection of actual decreases in criminal behaviour or are we seeing the effects of pre-charge diversion programs, such as restorative justice practices?  Should we be developing additional methods of data collection to adequately reflect the actual crime rates?  Comparisons to previous years become rather difficult as more and more offenders are processed through non-traditional channels.

3.4 Summary of Findings

  • In general, empirical research into restorative justice is arguably still in its infancy. Numerous questions remain unanswered.  There are several issues, however, that do appear to be resolved.
  • Victims who experience a restorative justice program express high levels of satisfaction with the process and the outcomes.  Victims also believe that the process is fair.  There are strong indications that victims are much less satisfied within the traditional court system. In addition, victims’ satisfaction level appears to be related to the fulfilment of restitution agreements.
  • Offenders also express high levels of satisfaction with restorative justice programming and perceive the process to be fair.  In addition, research suggests that offenders processed by the traditional system are less satisfied.  There is evidence, though, that the severity of the restitution agreement is closely related to an offenders’ satisfaction level.  The harsher the restitution, the more likely an offender will express dissatisfaction with the program.
  • Most restorative justice program participants have a high level of success in negotiating restitution agreements.  There is also an indication that a high proportion of offenders referred to restorative justice programs follow through on their agreements and are more likely to comply than are offenders with court-ordered restitution.
Date modified: