Killias and colleagues (2006) carried out a systematic review of research findings comparing custodial sentencing with non-custodial alternatives. The review specifically compares the effectiveness of custodial sanctions (imprisonment) versus non-custodial sentences (‘alternative’ or ‘community’) sanctions on post-sanction re-offending rates. The study has two elements: 1) a systematic review and 2) a meta-analysis. From nearly 300 studies, the systematic review identified 23 that met the minimal conditions of the review. These allowed in total 27 comparisons of custody and community sentences. Five studies were based on controlled or natural experiments and these were included in a meta-analysis. The meta-analysis is the focus of this summary.
What is it?
There are currently approximately 82,000 people incarcerated in prisons in England and Wales (Sturge, 2018). This represents an increase of nearly 50% since the early 1990s and there are significant concerns about overcrowding and violence. At the same time, there has been a substantial growth in alternative options, including the use of electronic monitoring or ‘tagging’, community sentences and restorative justice.
Whilst custodial sentencing has formed the historical basis of prison, over the last century or so, the relative merits of non-custodial forms of sanction have been the subject of considerable debate. This includes the century old notion that short sentences are, to paraphrase the review summarised here, too short to have any rehabilitative effect, but long enough to expose inmates to a criminogenic environment. This sentiment was recently echoed by the UK’s Ministry of Justice when it announced in 2019 that it is considering banning prison sentences of under six months.
The summarised review defines ‘custodial’ as ‘any sanction where offenders are deprived of freedom of movement’. This is a necessarily broad definition that encompasses prison as it is conventionally known, but also closed residential settings (where or not residents have some freedom to move outside of the domain for periods of time). The alternative is equally broad, and ‘non-custodial’ is defined as ‘any form of sanction that does not involve the deprivation of liberty’, such as community work, electronic monitoring, financial sanctions or a suspended custodial sentence.
Should it work?
Prison represents one of the main instruments of contemporary and historical crime control. It has a long history, and can be said to comprise a number of competing objectives that have come to the fore at different periods. Each of these has been shaped by historical context and evolving ideas about crime, criminals and the best ways of responding to, or preventing, crime (Garland, 1990). These objectives can be broadly defined as retribution, incapacitation, deterrence, and rehabilitation (Howard League of Penal Reform, 2015) and each has a different criteria of success that are both complex and not necessarily comparable.
Does it work?
Across the five studies included in the meta-analysis, the review found that rates of recidivism did not vary significantly between prisoners subject to custodial or non- custodial sanctions. As the reviewers state: ‘according to the results, non-custodial sentences are not beneficial in terms of lower rates of re-offending beyond random effects’ (Killias et al., 2006: 1).
The reviewers are clear however that there are not enough studies of sufficient quality/rigour or appropriate design to make any conclusions about the relative effects of prison versus its alternatives beyond the findings of the five studies.
How strong is the evidence?
The evidence is not strong. The data were assessed in line with the Campbell Systematic Review guidelines. Of the 23 studies included in the systematic review, only five were suitable for meta-analysis. The criteria for meta-analysis was initially to include only studies that were controlled experiments, quasi-experimental with four or more control variables, or natural experiments. In practice, the quasi-experimental studies were excluded on the basis of quality. In practice then, four of the studies were randomized control trials, and one was a natural experiment. All studies were assessed as either Level 5 on the Maryland Scale or at the high end of Level 4.
Is it worth it?
In England and Wales, the total average cost per prisoner in 2016/17 was £22,933, a small decrease from the 2015/2016 cost of £24,249 in 2015/16. These figures are for direct costs only and when total expenditure of the prison system is added, the cost per prisoner in 2016/17 was £35,371. The costs of non-custodial sentences vary significantly and are harder to calculate as a comparable aggregate.
Can it be implemented?
The review did not consider implementation.
What's missing from the evidence?
The main focus of this review was the effectiveness of prison versus alternatives to prison in reducing reoffending rates for those subject to these sanctions, and within a relatively short time-scale. By comparison, other benefits, including potentially longer-term benefits, such as the personal benefits derived from non-custodial sentences, were not considered.
Of the five studies included in the meta-analysis, four were conducted in the US and one in Switzerland. None were conducted in the UK. The UK government currently funds non-custodial alternatives to prison, and recently re-engaged in an old debate about the potential damage caused by short-term prison sentences. This debate is referenced in the review considered here, but is not reflected in high quality research conducted in the UK.