Why restorative justice needs research

6 May 2009

 Paper to conference on ‘European best practices of restorative justice in criminal procedure’, Budapest, 27-29 April 2009


Before beginning research we need to be clear what we are trying to do, namely to reduce the crimes and other harms which people inflict on each other. We can start by introducing restorative processes in schools. If crimes are committed, we have to respond, and restorative processes are increasingly being used. The basic research question is. How well are we doing? This includes not only the outcome but the structure and the process itself: how well was the process carried out, did it involve victims, offenders and members of the community ? Did the community make the arrangements needed to enable the offender to make reparation? Is support available for victims whose offenders are not caught? Are we learning from what offenders and victims tell us, so that we can reduce the societal pressures that lead to crime?

Before we consider how restorative justice tries to make the criminal justice system work better, we can take a step back to consider how restorative practices can create a society in which people are less likely to harm each other; but when it happens, we would help the victim, and look for ways to prevent further trouble.

How to reduce the crimes and other harms which people inflict on each other?

 Hungary has followed this logic, by introducing a National Strategy for Community Crime Prevention in 2003. This includes non-violent conflict resolution, enhancing small-community integration and control and other social measures (Lévay 2007-8) Criminologists have suggested many ways of reducing the pressures towards crime; most of them are part of social policy, and have little to do with criminal justice policy.

 A comprehensive policy for reducing the amount of harm which citizens cause to each other would ideally start in schools, and the Zöld Kakas Liceum1 has shown how this can be done even with students who had not been successful in other schools. Among other things they were encouraged to make their own rules; but soon they found that they had so many rules that many of them were broken. They therefore concentrated on the essential rules, and at the end of the year they summed up their achievement: ‘We’ve learned punctuality. We’ve learned to respect our fellows. We’ve learned to cooperate. We’ve learned to be serious in serious situations’ (Kerényi 2006).

Schools in Hull, in northern England, have adopted restorative practices, with striking results for improving behaviour and the school’s performance generally; there are plans to give restorative training to everyone in the city who works with children, and to make Hull into a ‘restorative city’. (IIRP 2008; Mirsky, n.d.)

 Another version of this method is ‘discipline that restores’ (DTR). The principle is that the teacher remains in charge of the framework of the student-teacher relation­ship, but respects the student by offering choices at every stage. After analysing how attempts to control through punishment can make matters worse, Roxanne Claassen, the main author, invites each new class at the beginning of the school year to agree their own ground rules and set their own targets for the year. A ‘flowchart’ of increasingly serious but non-punitive interventions is explained. When a conflict arises, the first step is a ‘constructive reminder’. The next time, the teacher will ‘actively listen’, and talk to the student. If there is a further problem, the student can choose between ‘Four options’ for dealing with it (I impose on you, we go to an arbitrator, we go to a mediator 2, we agree between ourselves); usually they choose #3 or #4. For uncooperative students there may be a spell in a ‘thinkery’, a place where another teacher helps the student to think through what happened, who was affected, and a plan for working together. If the problem is still not resolved, a ‘family conference’ is held. Only then, if necessary, the school authority structure will be used (Claassen and Claassen 2008). Methods like these have the potential to teach children respect for each other, animals, and the environment.

 Research in schools is also reported by Sherman and Strang (2007: 53-4) in a wide-ranging review of published research, reporting reduction in anti-social behaviour and increased feelings of safety among students, though not all the findings were statistically significant.

 The next step towards a restorative society is to create a network of community mediation centres, as in Finland3, Norway, and parts of the United Kingdom. They can deal with civil disputes and those which can be privately prosecuted in continental legal systems; they could also extend their work to include victim-offender mediation.

How to respond when crimes are committed?

 The traditional justice system, as we know, is based on confirming that a crime was committed; that the accused is guilty of committing it; and imposing a punishment (or sometimes another sanction). This gives the accused an incentive to deny or minimize what he or she has done. Although restorative processes are only used when the accused accepts responsibility, it is claimed that they make offenders more likely to do so. They ask different questions:

 What happened?

  • Who was affected?

  • What is needed to put it right?

  • Who should do it?

  • How can members of the community be involved?

  • What would make it less likely to happen again?

 The task of fundamental research is to ask whether these are good questions; practical research asks whether they were asked in the right way and led to the repair of the harm.

How well are we doing?

Just as , when we were considering how to respond to crime, we began by considering how to prevent it, when we now consider the response itself, we begin by considering how the response is designed. So we have to ask, What are the qualities of a good justice system?’ We do not go straight to the outcome; we look first at the structure and the process. In this context, research could be compared to an audit.


Researchers, then, should be involved in the design of the system (in German this is called Begleitforschung, accompanying research), although this is not always possible for political reasons: it can be difficult to explain to senior lawyers and politicians the relationship of restorative justice to criminal justice. Mediation in criminal cases has only recently been introduced in Hungary (Act CXXIII of 2006, quoted by Lévay 2007-8), so there is still time to influence the direction in which it develops. Researchers may begin by looking at the preventive policies mentioned above, and how widely restorative practices are used in schools and communities. As regards criminal cases, if it is accepted that restorative justice should include participation of the community, as supporters of victims and offenders, as volunteer mediators, and managers of NGOs, researchers with their knowledge of the theory and practice in other countries can advise on legislation that enables this to happen; evaluate how well it is working, both numerically and qualitatively; and recommend changes later if necessary. They can assess whether there is full use of volunteers, and whether these represent all groups of society, including ethnic minorities: for example, how many Muslim mediators are there in the UK, how many Roma mediators in Hungary? Some programmes have used police officers as mediators; research has found that some do it very well, and the experience can broaden the outlook of the officers, but there are problems such as under-preparation, coerced participation and lapses in neutrality, ‘particularly in the case of the more experienced facilitators’ (Hoyle et al. 2002: 66).

 If we accept the principle of minimum state intervention (‘as much state as necessary, but as little state as possible’), researchers should look at the extent to which cases which do not need the full power of the state are ‘diverted’ (kept out of the system); for example, do prosecutors refer cases to be assessed for mediation rather than prosecution? Can people go straight to mediation, for civil or privately prosecutable cases? It is helpful if the legislation is designed so as to make this possible.

 The response to crime is a matter of public concern, and researchers would want to see what arrangements are made for public accountability. Is an annual report published? Are some resources of staff time allocated to explaining the restorative concept .to the public and to professionals?


Then researchers can see whether the process is being operated according to restorative principles. This is because in restorative justice the process is important, as well as the outcome. So researchers will look at how well it was carried out, and whether it involved victims, offenders and members of the community? Since restorative justice is concerned about victims as well as offenders, they will also ask if support is available for victims whose offenders are not caught? In Hungary, for example, this would mean examining the operation of the Act CXXXV of 2005 on Victim support and state compensation (Lévay 2007-8), but many victims need emotional support as much as, or more than, compensation, so this should also be part of a restorative system.

 Participation by victims will never reach 100 per cent, since it is a voluntary process, but if it is well explained and becomes well known, the level should rise. There is a presumption that the take-up will be higher if the process is explained to victims (and offenders) by mediators, who understand the process well. The way in which they are contacted also makes a difference: by letter, phone or visit. Mediators may also discuss with victims whether they would prefer one-to-one mediation, or indirect mediation, or a ‘conference’. One reason for low attendance can be that victims are not consulted about the time when the meeting will be held. Research in the early days of youth offender panels in England found that only 22 per cent of victims attended meetings (Crawford and Newburn 2003: 185), although (partly as a result of this research) efforts are being made to improve this.

 Research into the process will include questions such as:

% of victims contacted

% of victims agreeing to mediation

% of cases enabling victim and offender to meet in a mediation/conference

 There has been criticism of the conferencing process, especially for young offenders, on the grounds that they may be intimidated by ‘a roomful of adults’. With this in mind the English legislation allows a young person to be accompanied by an adult supporter (invited by the young person with the panel’s agreement) and anyone else whom the panel considers to be capable of having a good influence on the offender, in addition to parents or guardians. The early research found that only in 15 per cent of panels was the young person accompanied by more than one adult (Crawford and Newburn 2003: 122).

 The supporters need not be lawyers – some would say that lawyers should not take part in the mediation, because the restorative meeting is not a trial. It does not take place unless the accused has already accepted some responsibility for the harm caused. Advocates of restorative justice argue that the prospect of a restorative process, rather than a punitive one, encourages the admission of guilt; the presumption of innocence until guilt is proven ‘means no accountability, and it sets the conditions for re-offending’ (Sawatsky 2009: 120). It is common for defending lawyers to advise their clients to plead ‘Not guilty’ and say nothing, in the hope that in some way, perhaps a procedural technicality, they can escape punishment. The accused is of course entitled to legal advice, under Article 6 of the European Convention on Human Rights (Right to a fair trial), but the lawyer should be aware of restorative principles. When the outcome is a restorative one, the accused has an incentive to admit his or her involvement in causing harm, and ‘wipe the slate clean’. It will be interesting to see if researchers can find a way to explore this hypothesis. It is supported by the experience at Hollow Water, Manitoba, Canada where considerable sexual abuse was admitted, in two cases without a victim even coming forward (Sawatsky 2009: ch. 4). Of 107 cases, only 2 were found to have re-offended (p. 99). Further evidence is provided by research in England, where the use of restorative justice doubled (or more) the offences brought to justice as diversion from criminal justice. In an experiment in Brooklyn, a crime was twice as likely to be brought to justice where restorative justice was used, as compared with the court process (Sherman and Strang 2007: 4, 82-3)

 Any good system needs some form of follow-up and feedback, to assess its performance. Research is an investigation in depth, which can usually only be carried out every few years; monitoring is routine record-keeping, including asking the participants how well they thought the process was conducted. In both cases the results should be given to the mediators, and included in the training of future mediators. It may be possible to establish a practice review group, including practitioners, administrators and researchers, to consider issues that arise in day-to-day practice and consider whether changes are needed. These may be local arrangements, or may be passed to the national organization which supervises restorative work. It is suggested by Sherman and Strang that this organization should be an official ‘Restorative Justice Board’ (2007: 88), but there is also a case for an organization that is independent of government and can even press the government to make changes when necessary.

 Researchers will also want to look at the training of mediators, both for their skills in listening and leading the meeting, but also to ensure that they learn to recognise their own prejudices and treat everyone with respect, including ethnic minorities. They will also want to consider whether arrangements are in place to make sure that the process is conducted fairly. In addition to the basic skills, such as active listening, and condemning the act but not the person, mediators need to learn what to avoid, such as dominating the discussion and imposing opinions. There are also more complex issues: has the facilitator used subtle techniques to persuade the victim and offender to follow a ‘script’ of forgiving and apologising, which may not be what they really want (Zernova 2007)? Or is that the correct thing for the facilitator to do, in the interests of individual well-being and social harmony?

 Even a restorative process, however, can be conducted well or badly; in addition to the routine monitoring, researchers will want to discover whether there is a grievance procedure (a restorative one, of course!), and whether the principles of restorative justice are correctly explained to the participants – otherwise they may mistake bad practice for normal practice, and not realize that they have grounds for complaint.

 Research by Lawrence Sherman and Heather Strang (2007: 44-5) examines questions of this kind, and finds that restorative justice in general, and the programmes which they studied in particular, comply with legal principles and those of the United Nations (2006: annex II). Other standards which researchers may want to use as a basis for assessment include those of the Council of Europe (1999 ) and CEPEJ (European Commission for the Efficiency of Justice) (2007)

 Finally, a detail which may be unexpected: researchers may want to ask whether refreshments are offered to the participants after mediation? In some models of mediation and conferencing, this is normal practice; it may for example fill the time while an agreement is being written out, and sharing food and drink is a profound way of symbolizing reconciliation.

Outcome :

Having looked at the structure and the process, let us now turn to the outcome. At this point the research becomes more numerical. But with these numerical data, and the ones I mentioned earlier, it is important to remember the dangers. Firstly, numerical research, and randomized controlled trials (RCTs) in particular, are not necessarily the best method for all purposes. They need large numbers in order to achieve statistical significance, and therefore it is often not practicable to explore in depth the quality of the services being studied. Was it good restorative justice, indeed was it restorative at all? With smaller numbers, statistically conclusive findings are less likely to be achieved, leading to disappointment all round. There is a tendency to focus on a primary outcome of interest to the funder, which is often the reconviction rate. Aidan Wilcox and other researchers (2005) have pointed to several problems. Drop-out rates can be as high as 33 to 68 per cent, and there is then a danger that the remaining cases are no longer representative. RCTs originated in medical research, but in social research the important element of double-blind is not possible.

 Offenders may have received other treatments in addition to the victim-offender meeting, so the latter may not be responsible for the ‘success’. (Conversely, Wilcox and colleagues might have added, if additional measures which the offender needs are not provided, this lack rather than the restorative justice process may be responsible if he or she re-offends.) As for victims, their satisfaction may be simply because someone has listened to them, rather than resulting from the restorative meeting itself. (Wilcoxet al. 2005). In one case, when the random allocation method was used, satisfaction was lowest among the victims who were promised restorative justice but were then allocated to the control group and consequently did not receive it (Sherman and Strang 2007: 63-4).

 Qualitative research, despite its necessarily smaller samples, can complement the findings of RCTs. Action research or ‘accompanying research’ has already been mentioned., and was used by Carolyn Hoyle and colleagues (2002). It does not merely tell us about ‘restorative justice’ in general, but indicates whether this restorative programme is being well conducted; and it tells us not after the project but during it, and can (as these researchers did) propose improvements while it is still running. The relatively small numbers of such studies may be criticized; but research such as that of Hoyle and colleagues, and Zernova (2007), can at least draw attention to issues, suggesting that other projects should be on the look-out for them. If they turn out to be widespread, changes may need to be made in the practice – and even the theory – of restorative justice.

 Secondly, there can be undesirable side-effects of basing policy on statistics.. This has been a particular problem in Britain. People do things to make the statistics look better, which do not necessarily make people feel better. One example among many: the efficiency of the police is judged by the number of arrests they make, so they arrest people who are easy to arrest, or people for whom a warning would be quite sufficient. Some more serious offenders, whose cases are more difficult to investigate, remain free; others are not referred to mediation, although their cases might be suitable. There are other examples of the harmful effects of statistical targets, from the National Health Service and schools.

 A question that will naturally be asked is the percentage of agreements fulfilled fully or partly. An interesting finding is that an agreement to make reparation may be at least as effective as enforcement by the threat of punishment (Sherman and Strang 2007: 58-9). However, in those cases where reparation is not completed, some form of enforcement will be necessary, and researchers will want to see whether it is carried out in a restorative way, and how effective it is.


Reparation can take different forms. Some victims want no more than an apology, or ask that the offender should do some community work; for others the priority is that he or she should not offend again, and should undertake training, education, therapy, anger management, or other programmes that will help to avoid re-offending. Therefore researchers should ask whether there were adequate opportunities for community work, perhaps offered by NGOs, and appropriate rehabilitative programmes. If these are not available, and there is a high rate of re-offending, it cannot be said that ‘restorative justice’ has failed – it is the supporting services that were not provided. In one English prison a mediation was carried out between a burglar and three young women who shared a flat which he had broken into. The session was observed by a former Chief Inspector of Prisons, who was impressed. It went well, they expressed their feelings, and the offender told them how and why he became a burglar. He had had a typically disadvantaged upbringing, had missed school and had not learnt to read and write, he had other problems including drug addiction. He agreed to attend programmes including literacy classes and addiction therapy. Afterwards the former Chief Inspector spoke to the prison governor, and asked if those programmes were available in the prison; he was shocked to hear that none of them were (Lord Ramsbotham, personal communication). If that young man fails to keep his agreements, who is responsible?

 Researchers may also ask whether young people who have made reparation by community service receive thanks and perhaps a certificate; this symbolizes the fact that they have not been punished as outcasts, but have made a contribution to the community and are part of it. Even better, they may work alongside volunteers who are not offenders, and a plaque can be installed giving them credit for their work. Programmes in prisons, for example in England and Hungary4, show ways of developing the good qualities of offenders. The current practice in England of making offenders work in public places, stigmatized by wearing distinctive coloured jackets, is completely contrary to this principle.

 A common criterion is ‘victim satisfaction’, which in almost all research is found to be very high, and significantly higher than courts when this comparison is made. Victims who experience restorative justice are less likely to suffer post-traumatic stress, and return to work sooner. But research has its pitfalls: in one study there were so many restrictions on cases that could be included in the programme that the numbers were insufficient (Sherman and Strang 2007: 83).

 When people, especially politicians, ask ‘Does restorative justice “work”?’ they commonly mean ‘Does it reduce reconvictions?’ Sherman and Strang found (2007: 68-71, 88) that reconvictions were often reduced, not always significantly, but were almost never increased. In the controversial field of violence within families, they cite Canadian research finding a reduction by a half in emergency visits to the home, compared with an increase of 50 per cent in comparison families (citing Pennell and Burford 2000).

 A study of three groups of programmes in different parts of England looked at the question of reconvictions. Results varied, but in total offenders who participated in restorative justice were reconvicted statistically significantly fewer times than those in the control group. It may be relevant that the programme which did best was the one (in Northumbria) which used conferencing rather than one-to-one mediation. (Shapland et al.: 66-7)

 Restorative justice can save costs in courts, in prisons, and in health care for victims, according to Sherman and Strang (2007: 86). They could have added that at least some of the savings could be used to provide more restorative justice and other non-custodial measures, if there were a simple method of transferring the funds. The research by Shapland et al. (2008: 67) found significantly lower costs in one group of programmes, although in the other two the difference was not statistically significant.


This paper began by referring to the aim of reducing crime, for which social policy is more important than criminal policy. But can restorative justice contribute to it? When many offenders tell their stories, there are bound to be indications of societal pressures that lead to crime. This is not to deny that individuals have choices about resisting those pressures; but even a healthy plant cannot grow well in poor soil. New Zealand, once again, has shown the way: some facilitators, when they notice clusters of cases from a particular geographic location or school, gather a number of people from social services, police and so on to consider whether there can be a plan to tackle the pressures that affect they young people. Similarly in South Africa, the Zwelethemba programme links the ‘peacemaking’ with ‘peacebuilding’, and includes a system for transferring funds to it (Sawatsky 2009: 59). This does not address major problems, including inadequate funding of essential services such as education, or inequalities in society, but it is a step in the right direction. The involvement of volunteers in the process helps to spread public awareness of social needs. If similar schemes can be introduced in other places, they will need to be researched to assess their effects.

 I have tried to go back to first principles, and suggested that we should begin by thinking about prevention. If young people learn to resolve their differences and misunderstandings in a respectful way, we shall be building a society in which people respect each other’s humanity. Research on such programmes can show how well it is working and how it can be improved, and can inform others so that the pioneering examples can be followed. Similarly, the extent and quality of community mediation can be assessed.

 Some of these proposals are based on a particular view of restorative justice, believing in the value of involving members of the community, and where possible resolving conflicts by agreement, without the imposition of authority

 Of course some people will still harm each other, and the restorative movement proposes that we should respond with a different set of questions, based on putting right the harm and looking for ways to avoid more of it happening in the future. This response needs research into its structure, its process and its outcome, and I have suggested that researchers should be involved at the design stage, to assist the legislators (and to avoid excessively detailed legislation). I suggested some of the points which researchers could consider.

 May I end by stressing three particular points which researchers should look for: (1) when offenders agree to make reparation, are arrangements in place to enable them to do so? (2) Are the circumstances regularly discussed, to see how pressures towards crime can be reduced? (3) Can a system be introduced by which money saved on prisons could be transferred to non-custodial ways of dealing with offenders? If research is focused on these questions it will help us to achieve more effective restorative justice and to build a more restorative society.


 CEPEJ (European Commission for the Efficiency of Justice) (2007) , Draft guidelines for a better implementation of the existing recommendation concerning mediation in penal matters. Strasbourg: CEPEJ.

Claassen, R and R Claassen (2008) .Discipline that restores: strategies to create respect, cooperation, and responsibility in the classroom. South Carolina: BookSurge Publishing. (www.disciplinethatrestores.org )

Council of Europe. Committee of Ministers. (1999) Recommendation No. R (99) 19 … to member states concerning mediation in penal matters. Reprinted in I Aertsen et al., 2004..Rebuilding community connections: mediation and restorative justice in Europe. Strasbourg: Council of Europe.

Crawford, A, and T Newburn (2003) Youth offending and restorative justice: implementing reform in youth justice. Cullompton: Willan Publishing.

Hoyle, C, R Young and R Hill (2002) Proceed with caution: an evaluation of the Thames Valley Police initiative in restorative cautioning. York: York Publishing Services.

IIRP (International Institute for Restorative Practices), Hull Centre for Restorative Practices (2008) The City of Hull: Riverside Project.

Kerényi, M (2006) ‘About the Jumpstart Programme of the Zöld Kakas Líceum.’ Paper from “Improving Citizenship & Restoring Community,” 10th International Institute for Restorative Practices World Conference, Budapest. http://www.safersanerschools.org/library/hu07/hu07_Kerenyi.html (downloaded 11.4.2009

Lévay, M (2007-8) ‘Development of criminal policy in Hungary during the first decade of the 21st century.’ Archiwum Kryminologii (Warsaw) XXIX-XXX, 543-554.

Mirsky, L (n.d.) ‘Hull, UK: toward a restorative city.’

http://www.safersanerschools.org/library/hull09.html#top (downloaded 11.4.2009)

Pennell, J, and G Burford (2000) ‘Family group decision making: protecting women and children.’ Child Welfare 79(2), March/April.

Sawatsky, J (2009) The ethic of traditional communities and the spirit of healing justice: studies from Hollow Water, the Iona Community and Plum Village. London: Jessica Kingsley Publishers.

Shapland, J, et al (2008) Does restorative justice affect reconviction? The fourth report from the evaluation of three schemes. (Ministry of Justice Research Series 10/08). www.justice.gov.uk/publications/research.htm

Sherman, L W, and H Strang (2007) Restorative justice: the evidence. London: Smith Institute. www.smith-institute.org.uk

Wilcox, A, with C Hoyle and R Young (2005) ‘Are randomised controlled trials really the “gold standard” in restorative justice research?’ British Journal of Community Justice, 3(2), 39-49.

Zernova, M (2007) Restorative justice: ideals and realities. Aldershot: Ashgate.


conference papers/budapest

1 Described By Borbála Ivany in a workshop at this conference.

2 Terms such as ‘mediator’ and ‘facilitator’ will be used interchangeably in this paper.

3 described by Aarne Kinnunen in a workshop at this conference.

4 described at this conference in workshops by Andy Hudson and Melinda Gyökös.