by John Braithwaite
Special Issue:
Evidence for Restorative Justice
Rethinking Effectiveness
Restorative justice is a way of selecting
strategies to respond to challenges like
healing the hurts of crime. Empathic empowerment of stakeholders who take turns
to speak in a circle are at the heart of its
strategy. The evidence is encouraging that
restorative justice works better than less
flexible top-down state decision making.
The effectiveness of restorative justice depends mainly, however, on the efficacy of
the intervention strategies that are chosen.
It is time to redirect R&D efforts to improving the quality of restorative strategy selection.
Asking “Does restorative justice work?”
is like asking whether any meta-strategy (a
strategy about selecting strategies) works.
Consider problem-oriented policing as an
example of a meta-strategy. Problem-oriented policing is an approach developed
by University of Wisconsin professor Herman Goldstein for improving police effectiveness through examining and acting on
the underlying conditions that give rise to
community problems. Responses emphasise prevention, go beyond the criminal justice system alone, and engage with other
state, community, and private sector actors.1 The evaluation literature is modestly
encouraging that when police are trained
to use problem-oriented policing their average effectiveness in preventing crime improves.2
Yet the effectiveness of problem-oriented policing in practice is highly variable.
Consider a local police unit’s diagnosis of
the crime problem in its locality as caused
by young black men who sell drugs. They
conclude that a good way of solving this
problem is to nab a few young black men
and beat them senseless in a publicly visible way. This would be a transparently ineffective strategy not only in the sense that
it could increase rather than reduce crime,
could even trigger city-wide race riots, but
also because it would set back other policy objectives like reducing racism in the society. The fact that quite often local police
are bound to choose counterproductive local solutions might leave us amazed that
the evaluation literature shows modest effectiveness overall.
Restorative justice is likewise a metastrategy for selecting strategies. Restorative justice is a relational form of justice for
selecting problem-prevention strategies. It
18
empowers stakeholders affected by putting
the problem in the center of a circle of deliberation, rather than putting the person
alleged to be responsible for it in the dock.
As with problem-oriented policing, there
is encouraging enough evidence that restorative justice “works” cost-effectively in
preventing a variety of injustice problems
that include crime prevention. However,
the really important evaluation questions
around restorative justice are not at the level of meta-strategy, but at the level of the
particular strategies that are chosen.
So the argument of this paper is that it
may now be time to redirect evaluation research attention onto how to improve the
quality of strategy selection when we do restorative justice. First, in the next section
we consider the latest evidence on the effectiveness of restorative justice in crime
prevention. Then we consider its effectiveness in enriching democracy and improving
justice in other ways beyond crime prevention, like helping child victims of violence to
be safe, secure and empowered with voice
within their families.
The Latest Evidence on Restorative
Justice Effectiveness
My book Restorative Justice and Responsive Regulation3 summarizes the evidence
on the effectiveness of restorative justice
in realizing various justice values, including
crime prevention. It is cautiously optimistic.
The latest important addition to that literature is a meta-analysis for the Campbell
Collaboration on the impact of restorative
justice on crime by Heather Strang et al.4 Its
conclusions are fundamentally similar to the
previous meta-analyses of over thirty tests
of the effectiveness of restorative justice by
both Latimer, Dowden and Muise5 and Bonta et al.,6 each conducted for the Canadian
Department of Justice. All three meta-analyses found a statistically significant effect
across combined studies in lower reoffending for restorative justice cases (compared
to controls). The difference in the Strang et
al. study is greater selectivity, more exacting methodological standards for inclusion
in the meta-analysis. Only ten studies were
included, all randomized controlled trials.
The overall result was the same—a modest
but statistically significant crime reduction
effect.
None of those most intimately involved
in the development of restorative justice
THE VERMONT BAR JOURNAL • SUMMER 2014
ever predicted huge crime reduction effects because we all saw badly managed
conferences that made things worse rather than better. A banal kind of counterproductive restorative justice, for example, is
where either the victim or the offender did
not turn up, pulling out at the last moment,
leaving the other side angrier than they
would have been had reconciliation never
been attempted. We were disappointed in
the extreme weakness of the effectiveness
of restorative justice in preventing property
crime in the Strang et al. evaluation as those
results started to come in, with one Canberra experiment actually finding slightly more
crime for the property offenders who went
to restorative justice (though not a statistically significant difference). At the same
time we were amazed at more than a 40%
reduction in reoffending (compared to controls randomly assigned to court) in the first
year outcomes of the RISE youth violence
experiment in Canberra (which reduced in
year two), and even more surprised when a
reduction in reoffending in one of the British violence experiments also achieved a
45% reduction in offending over two years.
The reductions in the other violence and
mixed violence and property experiments
in the Strang et al. review are still very substantial, but at about half this level.
What we have is some studies (mainly
with property crimes) showing disappointingly inconsequential ef