In July 2008 criminologists at the University of Sheffield, UK, issued their fourth and final report on a major research initiative launched in 2001 by the British Home Office to examine the effects of restorative justice (RJ) for adults and youth. The report marks the culmination of more than seven years of planning and work involving the collaboration of government, academia, social service agencies, and police and criminal justice institutions, including probation, courts and prisons.

(The University of Sheffield has also published three previous reports on different aspects of these research studies: Implementing restorative justice schemes, 2004,; Restorative justice in practice, 2006,; and Restorative justice: the views of victims and offenders, 2007,

The findings of the fourth report, Does restorative justice affect reconviction?, 2008,, show that face-to-face RJ conferences both reduce crime and provide a cost saving to government. The report “focuses on one of the key original aims of the Home Office funding, whether restorative justice ‘works,’ in the sense of reducing the likelihood of re-offending and for whom it ‘works’ in this way. It also covers whether the schemes were value for money, measured as whether the cost of running the scheme was balanced or outweighed by the benefit of less re-offending” (Shapland et al., 2008, p. i).

Dr. Joanna Shapland, professor of criminal justice and director of the Centre for Criminological Research at the University of Sheffield, chief author of this and the three previous reports in the series, said, “One of the important things about this report is that [the evaluated RJ schemes] involved adult offenders as well as juveniles, and serious as well as less serious offenses, including serious violence and household burglary.”

The study examined RJ schemes run by three organizations funded under 2001 Home Office research grants: CONNECT, REMEDI and the Justice Research Consortium.

CONNECT worked with two London magistrate courts, offering three RJ processes, mainly post-conviction and pre-sentence, resulting in 50 studied cases. CONNECT provided direct mediation (mediated face-to-face meetings between offenders and victims) in 11 cases, indirect — or shuttle — mediation (where information is passed between victims and offenders, who do not meet face to face) in 37 cases and restorative conferencing (facilitated face-to-face meetings between victims, offenders and their respective supporters) in two cases.

REMEDI, based in Sheffield, offered indirect (97 cases) and direct (35 cases) mediation across South Yorkshire. A total of 132 adult and youth cases were included in this evaluation, from youth cases involving final warnings, referral orders and other youth justice sentences, to adults given community sentences during resettlement pre-release from prison or during a long prison sentence. Referrals came from the National Probation Service and from offenders and victims themselves.

A control/comparison group was established for the CONNECT and REMEDI schemes by matching each individual offender in the RJ group to offenders in a non-RJ group on variables that might affect offending, such as offense committed, age and gender.

The Justice Research Consortium (JRC — ran the largest scheme by far, utilizing only face-to-face restorative conferences in three sites: London, Thames Valley and Northumbria.

The JRC scheme employed a randomized research design whereby: “After both offender and victim in eligible cases had consented to a conference … approximately equal numbers of cases were randomly assigned either to a conference group, which proceeded to hold the conference, or to a control group, which had no further restorative input. The aim was to create two very similar groups of cases so that the effects of holding the conference could be studied. Some 728 cases reached the point of randomization, with 342 being assigned to a conference” (Shapland et al., 2008, p. 2). Shapland said that this research design was “the gold standard” for detecting the effectiveness of an intervention.

The seven separate studies included in the JRC scheme were designed and directed by Dr. Lawrence Sherman, Wolfson Professor of Criminology at Cambridge University and director of the Jerry Lee Center for Criminology at the University of Pennsylvania, and Dr. Heather Strang, director of the Centre for Restorative Justice at the Australian National University. The studies were carried out by their research team, led by Dr. Sarah Bennett, Dr. Nova Inkpen and Dr. Dorothy Newbury-Birch. The JRC, chaired by Sir Charles Pollard, former chief constable of Thames Valley Police and former member of the Youth Justice Board for England and Wales, brought together police and criminal justice agencies as operational partners to run the conferences.

The two London studies involved adult offenders who were pleading guilty in a London Crown Court to serious burglary or robbery. Those randomly assigned to RJ took part in an RJ conference after plea and before sentence.

Both of the Thames Valley studies involved adult offenders and offenses of serious violence. One involved offenders serving custodial sentences, the other involved offenders sentenced to community supervision for their crimes.

In Northumbria, two studies involved adult offenders pleading guilty in magistrates’ court to assault or property crime with a personal victim. Again, those assigned to RJ took part in conferences held between plea and sentencing. Another study involved youth offenders who had been given a final police warning for violent or property offenses with a personal victim. For all the schemes — CONNECT, REMEDI and JRC — the Sheffield report examined the prevalence of reconvictions among offenders who had experienced RJ and a comparison group during a two-year period. It also measured the frequency of reconvictions over that two-year period, the seriousness of reconvictions over the period and the cost of offending, which combines elements of both seriousness and frequency into one monetary estimate of the criminal justice costs of processing the offense.

The Sheffield report states that “the only site for which there was a [statistically] significant difference between experimental and control groups in the likelihood of a conviction, or in the likelihood of a conviction, caution, reprimand or final warning was the JRC Northumbria magistrates’court property RCT, where the experimental group was significantly less likely to be reconvicted” (Shapland et al., 2008, p. 21).

However, the report concluded, “Summed over all three restorative justice schemes, those offenders who participated in restorative justice committed statistically significantly fewer offences (in terms of reconvictions) in the subsequent two years than offenders in the control group” (Shapland et al., 2008, p. 66). In other words, although only one study showed a statistically significant difference in prevalence of reconviction between the RJ group and the non-RJ group, frequency of reconviction was much lower in the RJ group – 27 percent lower on average across the seven JRC studies and even lower among the post-sentence violent offenders serving sentences in custody or in the community.

The report also pointed out that “There were no statistically significant results pointing towards any criminogenic effects of restorative justice (making people worse) in any scheme” (Shapland et al., 2008, p. iii). In other words, there was no evidence that restorative justice led to reoffending.

The report also explored the JRC data to see what kinds of offenders, what kinds of offenses and what elements of restorative justice lead to less re-offending. It appeared that the effects of RJ were unrelated to offenders’ gender, ethnicity or age.

In addition the report found:

    • In the REMEDI and CONNECT schemes, offenders who said they wanted to meet their victim at a conference — i.e., were prepared to engage in direct mediation — were less likely to be reconvicted and had lower frequency of reconviction.
    • Offenders observed (by researchers) to be participating actively in the conference had a significantly lower cost of subsequent convictions than those who participated less actively.
    • Offenders who said the conference was useful to them were significantly less likely to be reconvicted and had a lower frequency of reconviction (Shapland et al., 2008, p. 68).

Cost savings as a measure of the success of RJ is of special interest to governments and lawmakers. A program that can reduce re-offending can lower the future costs of trying cases, incarcerating and administering other legal sanctions for offenders and providing victim services.

The JRC programs produced statistically significant savings: “Nine pounds saved for every one pound spent,” said Strang. The report concludes that “JRC produced a net benefit in terms of reconviction (the sums saved in decreased reconviction were greater than the cost of running the scheme), whilst CONNECT and REMEDI produced a net cost. Hence JRC had produced value for money” (Shapland et al., 2008, p. vi).

Commented Shapland: “Conferencing seems to have somewhat more positive results than other RJ schemes, and there are good reasons for that. Because [the conference] is pulling together support from those close to an offender, as well as the supporters of victims, it gives a boost to those who are trying to stop offending. Whereas mediation seems to be less future focused; they tended to talk more about the offense and the effect on victims, rather than on what’s going to happen now.”

Shapland thinks future research will focus more on what types of RJ should be offered for different types of offenders and victims, and at what point in the criminal justice process various interventions might prove most beneficial.

There is hope among RJ advocates that it may be rolled out nationally across the UK. Shapland would like to see that happen, but she wants to be sure the programs are conducted well: “From a human rights standpoint, we need good practice regulations, which specify the types of reports judges will get, and we need assurances that there will be no undue pressure to either victims or offenders. The other thing that’s really important, if you do this kind of RJ, is that whatever is said during the event does not become evidence in any subsequent criminal or civil proceeding — that it’s not a fishing expedition. It was agreed in the schemes we reported on that this would not happen, but if it’s rolled out or done more widely, and I hope it will be, one has to set these safeguards up — as it has been in other countries [such as New Zealand and Northern Ireland].”


Shapland, J., Atkinson, A., Atkinson, H., Dignan, J., Edwards, L., Hibbert, J., et al. (2008). Does restorative justice affect reconviction?: The fourth report from the evaluation of three schemes. UK Ministry of Justice Research Series. Retrieved November 30, 2008, from

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

All our donors are acknowledged annually in Restorative Works.