What Works to Reduce Reoffending: A Summary of the Evidence

CHAPTER TWO: HOW DO INDIVIDUALS DESIST FROM OFFENDING?

Individual and social influences

Researchers have found that a number of individual factors, such as age, gender and the strength of social bonds are associated with reoffending. This section of the review describes the findings of the research into the impact of these factors upon reoffending.

Age and gender

The majority of offenders will have desisted from crime by the time they reach their mid 20s or early 30s. A highly consistent finding of longitudinal studies, both in the UK and internationally, is that offending begins in early adolescence, peaks during the late teens and tapers off in young adulthood. In the Cambridge Study in Delinquent Development which followed a cohort of 411 men born in a working class neighbourhood in South London from ages 8 to 56, the majority of offenders had desisted from crime by the age of 28, with a peak decrease in offending shown at the age of 23^[65]. Findings from the Edinburgh Study of Youth Transitions and Crime^[66] found that 14 was the peak age for self-reported offending, with a sharp decrease after that. At age 14, 52% of boys had engaged in four or more delinquent acts in the previous 12 months. By age 17, nearly half of these had stopped or sharply reduced their offending. Some longitudinal studies have documented that a small minority of offenders (about 5% of the offender population) continue to offend throughout adulthood and are responsible for a disproportionately large number of offences^[67]. However, even persistent offenders have been shown to desist or reduce their frequency of offending over time.^[68] The relationship between age and offending is interpreted as reflecting a number of underlying changes in biology, social contexts, attitudes and life circumstances that influence offenders' motivation to desist from crime rather than simply as a result of maturing as a person gets older^[69].

There are gender differences in the process of desistance from crime. The process of desistance may be similar in some respects for young men and women being driven by maturation, transitions, changed lifestyles and relationships^[70]. However, some gender differences have been found in the rationales given for desisting from crime. Young women tend to offer moral as opposed to instrumental rationales for stopping offending and were more likely to emphasise the importance of relational aspects of the process including parental attitudes, experiences of victimisation, the assumption of parental responsibilities and disassociation from offending peers^[71]. Some young women link their decisions to desist to the assumption of parental responsibilities. In general, young men focus more on personal choice and agency. Amongst persisters, girls and young women were more often keen to be seen as desisters, perhaps reflecting societal disapproval of female offending^[72].

In their study to explore the routes into and out of offending for young people in Scotland, Jamieson et al. (1999)^[73] interviewed 75 young people (aged 14-25 years) categorised into desisters (those who had not offended with the last year), resisters (young people who had never offended) and persisters (young people who had recently offended and were going on to criminal careers). They concluded that whilst younger desisters (like resisters) are inclined to fear the consequences of crime and view offending as 'futile' and morally wrong, older desisters are more likely to associate their abstinence with becoming more mature and moving on with their lives such as pursuing training or education. Males were more likely to say that their abstinence was 'personal choice', whilst females were more inclined to explain their desistance in terms of 'relational aspects' such as having gained parental responsibilities, not wanting to let their families down or having become more aware of the consequences of crime on their victims. In contrast, young people who offend classed as persisters were found to be less committed to education and employment and were most likely to have family members or peers also involved in crime. Persistent offending was often linked to drug addiction (particularly the need to fund a drug addiction) and in the case of females, was usually linked to involvement in relationships with male partners also involved in crime. Female persisters however, were more likely than their male counterparts to say they were trying to desist from crime and were more likely than young men to have adopted avoidance techniques to facilitate desistance. The literature suggests that girls mature (physically and emotionally) at an earlier age than boys and therefore will "reach and pass through the turbulent period associated with offending at a younger age"^[74].

Research evidence also points to differences in moral reasoning between the genders to explain why females have a stronger inclination than males to desist from offending. Underpinning women's moral reasoning is a general ethic of care and responsibility to others. In their 1999 study exploring young people's pathways into and out of crime, Jamieson et al.^[75] found that boys were much more likely than girls to have been the victims of physical assaults outside their own homes and as a result of their own experiences were more likely to adopt an individualistic approach to moral reasoning with a specific tendency towards 'victim blame'. Girls on the other hand were found to have a more 'relational' approach to moral reasoning, their accounts of offending were much more likely to include the effects of their actions on others.

Research has also shown that there can be gender differences in accessing some avenues which produce the social ties linked to desistance. For example, Huebner, DeJong and Cobbina suggest that in America women faced particular problems in finding employment following release from prison due to lack of childcare, discrimination and conflict with employers^[76].

Social ties

Quality social ties formed through employment, marriage and education can promote desistance. It is a reasonably consistent finding in the literature that the occurrence of key life events, such as obtaining and remaining in suitable employment, acquiring a stable partner and completing educational qualifications, increase the likelihood of desistance from offending by adding structure to offenders' lives and acting as a source of informal monitoring and emotional support^[77]. The same effect has been observed when offenders move away from criminal peers^[78]. More recently, researchers have stressed that the perceived strength, stability and quality of social attachments matter more than the events per se^[79]. However, the effects of marriage, employment and education on reoffending are complex and not independent of each other^[80]. Much of the literature covered in this review which investigated the impact of social bonds focused on employment and marriage, the findings of which are summarized below.

Being employed has been shown to be associated with reduced reoffending^[81]. However, there is evidence that just having any job does not encourage desistance, but that the stability and quality of the job are also important factors^[82]. The type of employment available and where it is available may also impact on the effect of employment on reoffending. Bellair and Kowalski found that the availability of jobs which were likely to hire low-skilled, former offenders (such as manufacturing and retail jobs) in particular areas was predictive of recidivism in those areas^[83]. Qualitative research from the Teesside Studies of Youth Transitions and Social Exclusion showed that many of its participants who were involved in offending experienced 'economic marginality'; "churning, non-progressive movement around low-level jobs, training places and time on 'the dole'"^[84]. Work was available but it was not stable and so did not necessarily lead to the formation of strong social bonds found to be important in desisting from crime. Similar to the findings of Bellair and Kowalski, the authors attribute this in part to the decline of manufacturing jobs in the area in which the study was conducted^[85].

Studies have found that employment can be associated with reduced reoffending for both men and women. For example, a US longitudinal study found that, among women, those who were homemakers and those who worked in the domestic sector had increased chances of desisting from offending^[86]. Whilst another US study found that employment was not predictive of desistance for women, the authors suggest that this may be due to a lack of available employment for those participating in the study^[87].

Research has found that marriage^[88] can be an important factor in an individual stopping offending^[89]. However, it is possible that the effects of marriage are not the same for everyone. For example, a number of studies have considered differences in the impact of marriage for men and women. Research in the Netherlands found that for men the impact of marriage varied based on whether their spouse also had a criminal conviction, but not for women^[90]. The study found that marriage led to reduced offending for women regardless of their spouse's criminal record, but for men marriage only led to reduced reoffending if their spouse did not have a criminal record. However, others have found that the impact of marriage on women's offending is less clear than it is on men's offending, finding no impact of marriage on subsequent offending^[91].

Imprisonment and community disposals

This section examines the impact of different forms of processing by the criminal justice system and their different impacts on rates of reoffending. The section covers the impact of imprisonment, community disposals, suspended sentences, imprisonment on remand^[92], bail supervision and the speed of punishment.

Imprisonment

Prison can represent value for money in the short-term when it is used for high-risk serious and/or certain types of prolific offenders, although overall the evidence about the effectiveness of prison in reducing reoffending is mixed. One argument for the imprisonment of offenders is that prison may prevent reoffending in the short term through incapacitation effects^[93]. However, evidence to support incapacitation effects is mixed. A recent study in the Netherlands found that the incapacitation effect of first time imprisonment (that is, the number of crimes prevented by imprisoning an offender rather than the offender serving a sentence in the community) was small, preventing between two and two-and-a-half recorded offences per year^[94]. The authors suggest that this is in line with other recent findings into the incapacitation effect of incarceration, although not all researchers have been able to identify an incapacitation effect of imprisonment. In England and Wales researchers were unable to find evidence of a reduction in the length of criminal careers following short-term imprisonment when compared to community sentences^[95]. As a result, the authors suggest that there was no observable incapacitation effect of short-term imprisonment, and instead offending was postponed during imprisonment rather than prevented. Consequently they conclude that imprisonment should be reserved for the most serious offenders^[96].

In addition to incapacitation imprisonment may also reduce reoffending if the prospect of returning to prison provides a deterrent effect. The evidence for a deterrent effect of imprisonment is again mixed. Some studies have found that prison can deter some individuals from committing further offences^[97], especially those with stable jobs or relationships who have more to lose from imprisonment^[98]. However, other studies have found no discernible impact of incarceration on future re-arrest, and as a result the researchers contend that we cannot conclude that imprisonment reduces recidivism^[99], and that it may increase the likelihood of reoffending^[100].

Researchers have also investigated whether serving a longer prison sentence leads to reductions in reoffending. A systematic review of studies comparing offenders who spent more time (an average of 30 months) versus less time (an average of 12.9 months) in prison found that offenders serving longer prison sentences were more likely to reoffend following release^[101]. These analyses controlled for offenders' level of risk. However, the results should be interpreted with caution as the studies did not control for other differences between groups, and the results were mainly based on US studies conducted during the 1970s. A more recent Dutch study also examined the relationship between the length of sentence received and subsequent rates of offending between people with similar characteristics who were imprisoned for similar offences^[102]. The authors found that longer periods spent in prison did not lead to changes in the proportion of offenders receiving future convictions, or the rate of future conviction. However, the authors raise questions about the generalizability of their findings outside of the Netherlands. The researchers suggest that the accumulation of recent studies suggest that prison may not have a strong deterrent effect^[103], although they stress that the evidence base is "nascent" and so further research is required. A similar conclusion is reached by Durlauf and Nagin who, in summarizing the existing literature, suggest that the length of existing prison sentences has at best only a marginal affect in reducing reoffending^[104]. Taken together these studies suggest that there is little evidence that increasing the length of sentence served for a particular offence would lead to reductions in reoffending.

It is also possible that, rather than reducing reoffending, imprisonment can increase long term reoffending by weakening social bonds and decreasing job stability^[105]. Reoffending may also be increased by experiences of victimization in prison^[106]. It is possible that the effects of imprisonment are not the same for all those who are imprisoned, and some authors suggest that the imprisonment may be especially criminogenic for 'low-risk' offenders^[107], although there is only limited evidence to support this. In addition, research shows that prison regimes may differ, particularly in terms of the quality of interpersonal relationships between prisoners and staff^[108]. This shapes prisoners' relative experiences of prison as painful, fair or degrading and may subsequently impact on levels of reoffending.

There is evidence that, when tangible and intangible costs of crime are included, imprisonment of high-risk serious and/or prolific offenders can represent value for money in the short-term, however costs are more likely to outweigh benefits when less serious, non-repeat offenders are imprisoned^[109]. These analyses do not take account of possible negative long-term effects of prison on reoffending, and should, therefore, be interpreted with caution.

Community disposals

Community sentences are more effective in reducing reoffending than short-term prison sentences and may provide greater opportunity for rehabilitation.

A number of studies have compared the effects of short-term imprisonment with those of community disposals. Scottish and English data suggest that community sentences are more effective in reducing recidivism than short-term prison sentences (of less than 12 months). In Scotland, reconviction rates are lower for those given community sentences compared to those released from short custodial sentences. 44% of those released from custody in 2011-12 were reconvicted within the following year, and the reconviction rate for those given short custodial sentences was 53% for a sentence length of between 3 and 6 months and 59% for less than 3 months. Whilst not directly comparable, due to the potentially different characteristics of offenders given each sentence type, 33% of those given community sentences (which in 2011-12 included the Community Payback Orders, Community Service Orders, and Probation Orders), were reconvicted within a year of being sentenced. In 2011-12, among females, 41% of those discharged from custody, and 28% of those given a community sentence were reconvicted within a year, but the same caveat of non-comparability of groups applies^[110]. It should be noted, however, that these figures do not control for potentially different characteristics of offenders receiving different disposals. In England and Wales, studies by the Ministry of Justice control for the differences in the offender characteristics by using both matching-by-variable and propensity scoring methods, to match offenders with similar characteristics. They show that - under matching-by-variable - the proven reoffending rate of offenders commencing probation supervision (either Community Order or Suspended Sentence Order) in 2007 was 46%, which was seven percentage points lower than the 53% for those who had served short-term custodial sentences of 12 months and under^[111]. Using propensity score matching for the 2007 data, the study again found a difference of seven percentage points. The updated version of this study, using 2010 data, is not directly comparable but finds a similar difference using propensity score matching^[112].

Cullen et al. present a review of a number of international studies which examine the effects of imprisonment and community sentences on reoffending. In sum, they suggest that the evidence consistently shows that prisons do not reduce reoffending more than non-custodial sentences^[113]. Other authors have found that those released from prison had higher reoffending than those serving community sentences using a number of different methods^[114]. Finally, American researchers found that prison was associated with higher likelihood of property and drug recidivism when compared with custodial sentences for both men and women^[115].

Scottish and international evidence suggests the greater effect of community sentences in reducing reoffending may be due to the fact that offenders on community sentences have more opportunities to access rehabilitation services compared to offenders on short-term prison sentences that have limited access to rehabilitation programmes in the short period of time they are in prison^[116]. There is evidence from meta-analyses that the quality of the service that is provided within a sanction rather than the sanction in itself can impact on recidivism^[117]. In Scotland, McIvor found that, in the context of drug courts, judicial review - and, in particular, continuity of sentencer review - was associated with increased compliance and reductions in recidivism^[118]. Also in Scotland, Weaver and Armstrong compared experiences of those serving short-term prison sentences with those serving community disposals. The study found that short prison sentences were seen by some as meaningless^[119], putting people's lives on hold but not helping them overcome their problems. Most of the negative experiences of prison were its unintended consequences in losing employment, housing or contact with family. In contrast to short-term prison sentences, community punishments were more often seen as positive and constructive^[120], allowing offenders to get help for their immediate problems such as drug and alcohol use. Such support services were often unavailable for those on short prison sentences.

Community sentences in Scotland

As part of the Criminal Justice and Licensing Act 2010, Scotland implemented a presumption against short prison sentences of three months or less. This is in accordance with the findings of a number of research studies which have compared the reoffending rates of those serving community sentences against those serving short prison sentences. An evaluation of the implementation of presumption against short sentences, as well as the use of Community Payback Orders and Criminal Justice Social Work Reports which were also implemented as part of the Criminal Justice and Licensing Act 2010, is currently being undertaken for the Scottish Government. The results of the evaluation will be published upon completion.

Cost-benefit analysis of community disposals and prison

There is limited cost-benefit analysis evidence comparing community-based sanctions with prison. Matrix Knowledge Group found some evidence that surveillance using either an Intensive Supervision Programme or Home Detention Curfew (HDC) represents value for money compared to prison^[121]. However, they also found that that there was no statistically significant difference in savings to society between community service and prison, or between community supervision with a cognitive behavioural element and prison. However these results should be interpreted with caution as they were based on a small number of studies.

Community sentences for women who offend

Despite the increasing numbers of women given community sentences in the UK and in other jurisdictions in recent years, there has been little research into whether they reduce women's reoffending rates or into women's experiences of these disposals. Women are proportionately more likely than men to be placed on a probation order; however the risk of breach for those with more chaotic lifestyles means that the intervention may ultimately result in a custodial sentence. In Scotland, while women are more likely to complete probation and community service orders than men, where breach proceedings are pursued, women are slightly more likely than men to have their orders breached as a result of non-compliance, while men's orders are more likely than women's to be revoked as a result of a further offence ^[122]. It has been argued on theoretical grounds that this higher risk of breach for women may have negative consequences for the process of desistance and disrupt interventions in the community designed to help women stop offending^[123], although this has not been tested empirically. Women are also more likely to breach a Drug Treatment and Testing Order (DTTO) than men^[124].

Interviews with women on probation in Scotland indicate that they are often dealing with a wide range of social, financial and emotional issues which they raise with workers to seek help with dealing with them. This finding raises important questions about whether community disposals should take these contributory factors into account in the design and provision of community penalties^[125]. The study concluded that community disposals can provide opportunities to access practical and emotional help but that they are not being used to their full potential. If community disposals were designed to provide more structured help to women, this would clearly have consequences for workers involved in supervising and supporting women - in terms of skills, focus of interventions, criteria for measuring 'success' and time as a resource^[126].

Other disposals

Evidence is limited, but those serving suspended sentences may also have reduced reoffending when compared to those serving short-term prison sentences. In a review of international evidence on suspended sentences, Armstrong et al. suggest that the evidence on the use of suspended sentences and recidivism is mixed, with many studies plagued by methodological problems^[127]. However, Armstrong et al. conclude that there is some limited evidence to suggest that those serving suspended sentences have lower reconviction rates than those on prison sentences of twelve months or less, and slightly lower reconviction rates than those on community orders.

Remand can prevent some individuals from reoffending in the short-term through incapacitation; however it can also be associated with negative effects that may hinder longer-term desistance. Remand prevents reoffending in the short term through incapacitation effects. However, alongside this incapacitation effect, international and Scottish research has consistently documented the negative effects associated with remand including an increased risk of suicide and mental distress, disintegration of social supports and family ties, and disruption to employment that increase the likelihood of reoffending upon release^[128].

More research is required into the impact of bail supervision on reoffending. A Scottish qualitative study found that supervised bail can provide prosocial modelling and help with practical problems if the relationship between bailee and supervisor is positive^[129]. For some people, supervised bail was seen as helping to change their behaviour in the long term and helped to support family relationships, in contrast with remand and curfews which were seen to damage family relationships^[130]. However, it is possible that for some people experiencing bail supervision may be stigmatising, demonstrating that experiences of supervised bail are not uniform. Of the bail orders studied, three-quarters were successfully completed (that is, the bail order did not end because of breach or remand). However, the small number of cases involved in the study mean it is difficult to generalize from these results.

Research has not demonstrated the effectiveness of swift sentencing in reducing reoffending. As far as we are aware, there are extremely few studies that have tested the effects of celerity (or swiftness) of punishment on reoffending. Although there is some recent evidence of weaker quality that increasing the celerity of punishment may contribute to reductions in high-risk driving behaviours^[131], its effect on other types of crime is under-investigated, making the drawing of any useful conclusions impossible. In relation to young people, there is some argument that a swift response (not necessarily a punitive one) is important as it relates the response to the behaviour^[132].

Early release measures

This section presents the findings of research into early-release measures and the assessment of their impacts upon reoffending. The early-release measures covered in this review are electronic monitoring and parole supervision.

Electronic monitoring

Offenders released under electronic monitoring are no more likely to engage in criminal behaviour when released from prison compared to those who are not eligible for early release. However there is considerable variability in the experiences of electronic monitoring^[133]. There is clear evidence from both Scotland and England that only a small proportion of offenders released on Home Detention Curfew (HDC) reoffend whilst on curfew. An evaluation of HDC by the Ministry of Justice found that offenders who receive HDC under the current provision are no more likely to engage in criminal behaviour when released from prison, when compared to offenders with similar characteristics, who are not eligible for HDC^[134]. Previous research into HDC suggests that it can have variable effects, reduce both positive and negative peer association, as well as making it more difficult to find employment^[135].

There is some positive international evidence on the effectiveness of electronic monitoring, however its generalizability to Scotland is questionable. A study in Argentina found that use of pre-trial electronic monitoring reduced recidivism by between 11 and 16% when compared to prison. However, these findings may stem from idiosyncratic use of pre-trial electronic monitoring in the Argentine justice system, and so these results have questionable validity to inform as to the use of post-release tagging in Scotland^[136]. Electronic monitoring was also recently evaluated in Sweden using a quasi-experimental design. The evaluation found that offenders who participated in an early release programme that included electronic monitoring in the home, a job placement and a treatment programme were less likely to be reconvicted in the 3-year period following completion of their prison sentence compared to the control group^[137]. However, it was not possible to ascertain to what extent this positive effect on reoffending was a result of the electronic monitoring in the home or of the other elements included in the programme.

Qualitative research suggests that electronic monitoring can be associated with a complex set of emotions, from gratitude about avoiding prison to psychological stress at the constraints tagging places on the capacity to live a normal life^[138]. A qualitative study in Scotland interviewed 20 former gang members about their experiences of police enforced curfews and electronic monitoring. The participants' experiences were mixed; some found that electronic monitoring can help to cut off ties to antisocial situations, people and places, but in some cases curfews led to increase strain on family relationships and to coping with this strain via alcohol and drugs^[139].

Parole

The majority of offenders released on parole successfully complete their licence period but evidence on the impact of parole on reoffending is mixed, and there is a lack of evidence about its longer-term effects. A Scottish study of release outcomes of prisoners sentenced to 4 years or more on or after 1 October 1993 and whose full sentence expired on or before 31 March 2001 found that 79% of those released on parole successfully completed their full licence period, and among those, 82% did not attract any convictions while they were on licence^[140]. A recent study in England and Wales found that people released from prison on license had a one-year reoffending rate between 14 and 17 percentage points lower than those not on license. The two year reoffending rate was lower by 16 to 20 percentage points. However, after three years, although reoffending rates remained lower for those who had been released on license, the results were not statistically significant^[141].

However, other researchers suggest that parole supervision may lead to increased reconviction due to higher rates of detection for those under supervision^[142], and higher rates of violation for minor infringements^[143]. In a study of parole violation in California, Grattet et al. demonstrate that the increased risk of reconviction for those on parole in areas with higher levels of supervision was in part a function of increased supervision over and above the individual characteristics of the parolee^[144]. It should be noted that, given the importance of the US system of supervision to the findings of this study it is uncertain to what extent the findings can be generalized to supervision arrangements in Scotland.

Few studies covered in this review examined the long-term effects of parole. In an American study Osterman found that there was little difference in reoffending between those who received parole supervision for a short time and those who did not receive supervision at all, whilst those under parole supervision for the duration of their study were less likely to reoffend than those who did not receive parole supervision^[145]. As a result, Osterman concludes that parole supervision had little long-term effect on reoffending, and attributes these findings to the 'reactive' type of supervision present in the area of the study. This suggests that these findings may not easily generalize to Scotland.

Diversion

This section outlines the findings of research into diversion from regular criminal justice system processing and its observed impact on subsequent reoffending for young and adult offenders. In this section the term "diversion" refers to alternatives to court disposals including diversion to social work, direct measures, and other forms of diversion.

Young offenders

Diverting young people away from the criminal justice system can be effective in reducing their reoffending and can have positive long-term impacts in people's lives. Findings from the Edinburgh Study of Youth Transitions and Crime (hereafter, the Edinburgh Study) indicate that the deeper a youth is carried into the formal processing system, the less likely he/she is to stop offending. The authors argue that the most significant factor in reducing offending is minimal formal intervention and maximum diversion to programming that does not have the trappings of criminal processing^[146]. This finding has been echoed by a number of other studies.

A recent systematic review of 29 studies found that young people with a prior criminal record who were diverted from the criminal justice system to social work were less likely to reoffend compared to those who went to court. Diversion to social work produced bigger reductions in reoffending compared to simple release that was not combined with some form of intervention^[147]. Another meta-analysis also found that, on average, diversion by either intervention or caution was more effective in reducing reoffending than 'traditional justice system processing' such as probation or imprisonment^[148]. Taking all studies together, support is found for the idea that the more processing a person receives the more criminogenic the effect. However, it is possible that this finding only holds for 'low-risk' youth who had lower levels of reoffending when diverted before being charged, rather than being diverted after being charged. Furthermore, the studies included in the review are predominantly from the United States, and the authors identify Scotland as a youth justice system with quite different characteristics. This means that the generalizability of these findings to Scotland is questionable. In England and Wales, positive effects on reoffending have also been reported in the process evaluation of Triage^[149]. Triage diverts young people who have offended for the first time under police custody to support services provided by a youth worker and, where appropriate, restorative justice informed interventions. However, a further report was unable to evaluate whether Triage had led to reduced reoffending among its participants due to a lack of available data^[150].

Throughout the literature, there is the recurring concept that both 'needs' as well as 'deeds' are important to understanding youth offending and desistance from it. As Fraser et al. highlight^[151], findings from the Edinburgh Study indicate a strong relationship between involvement in violent offending and a range of vulnerabilities, including self-harm. The literature argues that there are strong and consistent links between needs and deeds within the youth justice context; links which provide strong support for the Kilbrandon ethos underpinning the Children's Hearings System. Up to age 17 years and 6 months Sheriffs can request the advice and disposal of a case at the Children's Hearing System. It is argued that increasing the number of under 18s diverted to this childcare system, where their offence and criminogenic needs can be addressed together, reduces the risk of them reoffending and entering into the adult system. These findings demonstrate the negative effects of labelling by the justice system are long lasting, and the authors therefore recommend non-intervention wherever possible. This longevity of the effects of diversion from the justice system are also shown in the findings of a study in Rochester, New York, which found contact with the police in adolescence was associated with a number of negative outcomes later in life^[152]. These include increased probability of arrest and involvement in crime in the early 20s, greater likelihood of dropping out of high school, and increased involvement in crime and drug use. In turn, there were indirect effects of these negative outcomes in the early 20s to those measured in the late 20s/early 30s, such as increased drug use and welfare dependence and unemployment.

Adult offenders

There is less evidence on the effectiveness of diversion in reducing reoffending among adults, though some UK studies are currently underway. To the best of our knowledge, there is no systematic review of the effectiveness of diversion among adult offenders. There is some international evidence that diversion to drug or mental health treatment can reduce reoffending among offenders that experience such problems^[153], although other researchers have suggested that the evidence for diversion schemes to mental health services is limited^[154]. In Scotland, an evaluation of diversion to social work schemes found that the majority of accused had completed their period on diversion successfully and the majority of the objectives set were recorded as having been fully or mostly achieved by the time diversion ended. For the 111 accused for whom information about further charges was available, ten (out of 46) on social work diversion programmes and 17 (out of 65) from mediation and reparation schemes had further charges or convictions recorded against them (57% of those referred)^[155]. In England and Wales, positive results have been reported in the process evaluation of the Intensive Alternatives to Custody (IAC) diversion programme that offers an intensive community order as an alternative to short-term custody. Initial results suggest that whilst the IAC group had lower levels of reoffending than those who served short court order, this result was not statistically significant^[156]. Similarly, no significant difference was observed between the IAC group and a matched group serving other court orders. However, this result may be due to the numbers of offenders who have been through the diversion programme and further analysis of IAC will be undertaken with subsequent cohorts as data becomes available.

Diversion has also traditionally been used with female offenders, and some researchers recommend (on theoretical grounds) early interventions and diversion to social work for women who offend, due to the nature of many women's offending^[157]. In England and Wales, women can be diverted to community-based centres that aim to provide support to tackle underlying causes of offending. An evaluation of six Women's Community Services found that feedback from service users has been positive, but identified no impact on reoffending due to data collection and monitoring issues^[158].

Rehabilitation

This section examines evidence on the effects of a number rehabilitative programmes on reoffending. Included in this section are summaries of research regarding:

• Risk, needs and responsivity (RNR) assessment and cognitive behavioural therapy (CBT).
• Risk assessment and treatment for sexual offenders and domestic abuse.
• Holistic interventions.
• The impact of motivation to change on rehabilitation and strengths-based approaches to reoffending.
• Interventions to develop social bonds.
• Interventions to improve agency, self-efficacy and good problem-solving skills.
• Programmes for employment, education, drug treatment, alcohol misuse, and mental health interventions.

Risk, Needs and Responsivity and Cognitive Behaviour Therapy

Interventions are more effective when they are based on a sound assessment of risk, need and responsivity. A significant body of research emphasises the centrality of risk, needs and responsivity (RNR) assessment to effective interventions and improved outcomes in reduced reoffending^[159]. To take one example, a meta-analysis of the effectiveness of young offender programmes in Europe showed that programmes adhering to RNR principles had around 18% less reoffending than control groups^[160]. The risk principle states that the level of intervention should be matched to risk of reoffending, with higher risk offenders receiving more treatment. The need principle asserts that only factors directly associated with reoffending should be targeted in interventions and that crime-prevention may be overlooked if too much focus is paid to other social needs^[161]. Finally, the responsivity principle recommends that intervention programmes should be matched to characteristics of the offender.^[162] Important responsivity characteristics include cognitive functioning, mental health issues, personality issues and trauma^[163]. RNR principles are based on general personality and cognitive social learning theory^[164].

Factors important to establish risk and need

The RNR model describes the "central eight" domains which predict reoffending and outline the areas which treatment should target^[165]. Whilst the central eight are not an exhaustive list of all possible combinations of risk and need, they are considered the "best established"^[166] risk and need factors to predict reoffending. These eight domains are split into two groups of four based on their association with reoffending; a "big four", which are most predictive of reoffending, and a "modest four", which are less predictive of recidivism. The big four comprise: a history of criminal behaviour; antisocial personality pattern; antisocial attitudes, values, beliefs and cognitive-emotional states; and antisocial associates. The modest four is made up of: low levels of rewards in the home (family/marital), school/work, and leisure/recreation; and substance abuse. Substance abuse is strongly predictive for women, and so the authors suggest that there may be a "big five" for women who offend^[167].

Assessment of these factors is used to identify those most suited for greater supervision and treatment, as well as the factors that intervention programmes should target^[168]. These are based on 'dynamic risk factors'; that is those which can be changed. Those which cannot be changed are known as 'static risk factors'. In the "central eight", for example, a history of offending is a static risk factor which cannot be changed by an intervention, whereas substance abuse is a dynamic factor.

RNR principles have been converted into inventories such as the Level of Service/Case Management Inventory (LS/CMI)^[169]. Scottish developments, such as the introduction of the LS/CMI and the development of a shared approach to risk practice, are based on this evidence. In addition, the Multi Agency Public Protection Arrangements in Scotland and England provide an opportunity to test the impact of the collaborative approach to risk practice.

Recent developments in risk assessment

There is increasing interest in incorporating strengths and protective factors in assessment, and instruments that support the structured assessment of risk with attention to protective factors are emerging^[170]. Developers of risk assessment instruments also highlight the need for greater measurement of non-offending identity^[171]. The development, application and rigorous testing of such instruments will allow for greater understanding of the relative contribution of strengths and protective factors to risk assessment.

A number of studies have found that RNR assessment tools have greater predictive validity in demonstration projects than when used in practice. The authors of LS/CMI suggest that this finding is due to greater rigour and integrity of the evaluation process in their own studies, and the availability of high quality data in demonstration studies^[172]. However, the authors also state that "there may be some loss in the true predictive validity of a risk assessment scale as it transverses national, and hence legal, boundaries"^[173]. As such, more research is needed about implementation and use of risk-assessment tools in practice^[174].

In addition, some researchers have suggested that there can be a tension in practice between responsivity (that is, tailoring interventions to individuals' needs) and delivering the programme as dictated by the programme manual^[175]. This may explain the reduced effectiveness of programme roll-out when compared to demonstration programmes^[176]. Furthermore, when investigating the use of risk-assessment tools in practice, qualitative research in Ireland and Northern Ireland found that practitioners often resisted using risk assessment tools in favour of clinical judgement^[177].

Applying RNR principles to different groups of offenders

The LSI tools originated from a sample that was predominantly male.^[178] A range of studies demonstrates the applicability of the factors included in the LSI and other tools across age, gender and race^[179], and a number of studies have been conducted to assess their validity with particular populations, such as women, people from ethnic minority backgrounds, those with a mental disorder^[180], as well as across specific offence types.

Researchers have found that, in the LS/CMI, gender-neutral needs (including attitudes, peers, behavioural pattern and history, employment and education, and in particular substance use) are better predictors of reoffending in women than gender specific factors (such as parenting responsibility and stress, victimisation history, and self-harm)^[181]. This is not to suggest that there are not gender differences: there is a generally reported higher prevalence of victimisation, poverty, low self-esteem and low self-efficacy women offenders than males; Van Voorhis et al. suggest that, whilst the Level of Service-Revised inventory (LSI-R) is valid with women, adding gender-specific factors to LSI-R increased its predictive validity^[182]; and Andrews et al. found that drug use was a stronger predictive factor for women than for men^[183]. Andrews et al. also found that women with low levels of assessed risk offended at lower levels than low risk men, meaning that RNR-based tools may be "over-predicting" reoffending in such women: this suggests that practitioners should ensure that women offenders who are assessed as low risk only receive a low intensity of intervention, in accordance with the risk principle. Taking the findings of several studies overall, they showed little gender difference in the predictors of recidivism suggesting that, in sum, the factors are likely behave in a gender-neutral manner. However, there is increasing consensus that regardless of whether gender-specific concerns are predictive of recidivism, are criminogenic needs, or are indicators of gender-specific pathways into offending, they are responsivity issues that must be addressed in the delivery of services to enhance effectiveness^[184].

There has also been increasing research interest in examining the use of risk assessment tools for people from different ethnic groups. Research into the validity of risk assessment measures with aboriginal populations in Australia, New Zealand and Canada found that on the whole risk assessment tools validly classified Aboriginal offenders. However, these tools displayed less accuracy in some domains^[185], differences in the magnitude of the predictive effects of the central eight between aboriginal and non-aboriginal offenders^[186], and potentially under-classified low-scoring aboriginal offenders. Based on the risk principle, this may possibly reduce access to needed treatment. The few evaluations which have been conducted on studies which incorporate items specifically for different cultural groups, show some effectiveness in reducing reoffending, however many of these studies have methodological limitations^[187]. Furthermore, given that much the research into offenders from different ethnic or cultural groups comes from Canada, Australia, New Zealand or America, it has - until recently - been uncertain to what extent these findings can be generalised to Scotland.

Emerging research is reporting on the early use of the LS/CMI in Scotland, from the first validation study conducted outside North America^[188]. This finds that it is generally performing well in Scotland, and these findings apply irrespective of gender; the author contrasts this to the findings of Olver et al^[189], that application of the LSI instruments outside Canada showed poorer predictive validity. Furthermore, the gender-responsive items embedded in the LS/CMI distinguish between males and females, and among females. These findings not only validate the LS/CMI and its general applicability, but also reinforce the success of its early implementation in Scotland.

Research findings are tentative, but social conditions may alter the effectiveness of RNR tools in predicting reoffending. In recent years a number of studies have investigated the impact of social context (features of where people live) on the ability of risk assessment tools to predict reoffending. Onifade et al. found that the capacity of risk assessment tools to predict reoffending varied based on characteristics of neighbourhoods in which people lived^[190]. Put another way, risk assessment tools were accurate for people in some neighbourhoods but not others. As a result they suggest that only using individual-level assessment of risk does not give a full picture of the risk of reoffending. However, more research is required to investigate these effects before firm conclusions are drawn as the research findings at present are equivocal. For example, Wang et al. found that on the whole individual factors such as criminal history were more predictive of reoffending than features of the counties in which people lived^[191]; the authors also state that more research is required into the impact of social context on reoffending. One explanation for these divergent findings is that social scientists are currently less able to accurately measure and model social factors which are thought to affect reoffending^[192] than they can measure individual factors. Indeed, each of the studies cited above suggests that their findings may be due to the way in which they have measured social influences on offending^[193].

The advantage of assessing dynamic factors or criminogenic needs is that it adds to the currency and relevance of the assessment. It is therefore important to remember that, just as dynamic elements associated with reoffending may change for an individual, so may her or his likelihood of re-offending - and so an assessment made at one point may not be valid if that person's circumstances change considerably (see, for example, section on Social ties above). This relationship between change in risk/needs and change in re-offending is not yet well understood on the individual level, although some studies do point to its predictive validity. However, the importance of recognising the dynamic nature of risk and the associated need to regularly review and update assessments is underlined in the manuals of LS/CMI and other instruments, and identified as standard practice in Scotland^[194].

Great care must be taken in applying a risk assessment approach to young people who offend. Fraser et al. highlight that over the past 15 or so years, the risk factors and assessment approach to devising preventative strategies has become a dominant discourse in youth justice and that something of a consensus has been built around the precipitating factors of family conflict, truancy, drug use, lack of/ irresponsible parenting, low intelligence, delinquent peers and community organisation^[195]. One of the dangers of looking at risk factors for offending is the potential to pre-emptively stigmatise young people based on assumptions about what they might do in the future, not what they have done, and may lead to "net-widening" of services. In addition, whilst many risk factors have been identified, less is known about how to robustly establish which risk factors are causes and which are merely correlations.

Cognitive-behavioural programmes can lead to modest reductions in reoffending especially when they are rigorously implemented and combined with support in solving practical problems. Antisocial attitudes are among the strongest predictors of reoffending^[196]. There is good evidence from experiments conducted in the United States that cognitive-behavioural programmes that aim to change offenders' thinking styles and attitudes can result in modest reductions in reoffending when rigorously implemented^[197]. Evidence from the UK is more mixed, with some studies reporting modest reductions in reconviction rates and frequency of reoffending among programme participants (e.g. the evaluation of the Enhanced Thinking Skills programme) and others show no significant effects^[198]. A recent evidence review by the Ministry of Justice suggests that CBT can reduce reoffending by between eight^[199] to ten^[200] percentage points, and between six^[201] and eight percentage points^[202] in custody settings. Cognitive behavioural programmes are often part of treatment based on the RNR principles outlined above.

Implementation of CBT programmes

Differences in results of American and UK studies may reflect variations in the quality and rigour of programme implementation rather than genuine differences in effectiveness. Programmes may work better in the US simply because they are implemented better, though differences in the characteristics of programme participants may also account for some of the variation in outcomes. Research into the factors affecting outcomes of CBT programmes includes the quality of implementation^[203] and organizational factors of the agency implementing the intervention^[204] such as job satisfaction, training and supervision^[205]. In a recent study Wright et al. found that cognitive behavioural programmes were less effective in reducing reoffending in disadvantaged areas^[206]. The authors acknowledge that this association could be explained through reference to decreased access to resources and networks which would support desistance. However, from their own study, they conclude that the association may be due to the lower quality of programme implementation in disadvantaged areas. Their argument is based on a regression analysis which demonstrates that the correlation between markers of disadvantage and reoffending in their own sample is not statistically significant, once programme quality (indicated by the Correctional Programme Assessment Inventory (CPAI)) is controlled for. They theorise that a lack of resources in disadvantaged areas may limit the ability of programme organisers to run effectively in these locations. In either case, these findings are especially important because many offenders return to areas of high socio-economic deprivation upon leaving prison^[207].

Process evaluations of cognitive-behavioural programmes delivered in England and Wales have reported a range of problems and shortfalls in implementation including high attrition rates, long waiting lists, lack of booster work prior to release and ineffective targeting^[208]. High attrition rates can substantially alter the observed effects of intervention evaluations^[209]. In Scotland, no outcome evaluations of accredited programmes have been conducted as yet but process evaluations have highlighted similar problems to those in England^[210]. A recent UK review of the quality of offender supervision highlighted that accredited programmes cannot operate effectively in isolation, without addressing the broader context in which offending takes place and the multiplicity of offenders' needs^[211].

Cognitive-behavioural therapy for women who offend

Significantly fewer women than men are assessed as having considerable attitude problems requiring intervention. Although prevalence rates are low, there is preliminary evidence to suggest that the evaluation of anti-social attitudes is an important factor for assessment of risk for women^[212]. For example, results of prediction studies on US samples show statistically significant relationships between particular anti-social attitudes and recidivism in female offenders^[213]. However, in addressing anti-social attitudes, there is disagreement in the literature as to whether cognitive-behavioural approaches are as effective for women as they are for men.

Some feminist theorists criticise CBT for not adopting a holistic approach. Research from Australia has shown that female offenders were more likely to rate strength-based, holistic programmes which were collaborative and understood women's perspectives as having helped them to reduce their offending, although the authors raise some possible concerns about the sampling and outcome measures used^[214]. Other criticisms include that CBT programmes ignore contextual factors such as partners, family and friends, ignore the 'woman's voice' in relying on quantitative data, do not focus on strengths and do not recognise women's pathway into crime^[215]. These criticisms are essentially theory-driven and there is little robust evidence on how effective cognitive-behavioural programmes are on women's offending behaviour.

There is, however, general agreement that positive outcomes for women may be enhanced if responsivity factors (such as rewarding strengths including prosocial thinking and ensuring empathic staff attitudes) are incorporated into CBT programmes. One study found that empathic probation officers who actively challenge criminal thinking while simultaneously rewarding prosocial thinking can reduce recidivism by almost 80%^[216]. While some US evaluations have found positive results for women, in the UK, there is a paucity of reliable evidence on effectiveness of CBT programmes for women. One of the only UK evaluations to consider the impact of CBT on female prisoners was undertaken in 2006 but found no significant differences in the one- and two-year reconviction rates for male or female participants on the Enhanced Thinking Skills (ETS) Programme^[217]. The ETS's replacement, the Thinking Skills Programme (TSP) was introduced in 2009 and designed with the specific purpose of incorporating more gender-specific elements into cognitive skills programmes. Through interviews with women on the programme it identifies areas for improvement such as the use of mixed gender groups and relating the programme more explicitly to relationships outside prison. An evaluation of TSP is yet to be undertaken using reconvictions data, however an evaluation using psychometric tests found that those who completed TSP showed improvements in attitudes and thinking styles compared to those who had completed ETS^[218].

Cognitive behavioural therapy and young people who offend

CBT interventions have been found to be the most effective interventions in reducing reoffending in young people^[219]. However, a Scottish qualitative study into the use of CBT with young people in secure facilities suggests that CBT may misconceive the nature of youth offending^[220]. Rather than being based on improper cognition, the interviews suggested that offending was associated with peer pressure, substance abuse, and boredom. In addition, the interventions delivered in secure settings were not considered by the young people interviewed to have much relevance to their lives outside of the secure facility^[221].

Cost-effectiveness of cognitive behavioural therapy programmes

Limited work appears to have been undertaken on the value for money of CBT programmes. Matrix Knowledge Group^[222] found some evidence that prison with behavioural treatment represents value for money compared to 'standard' prison.

Risk assessment and interventions for specific types of offender

Interventions and risk-assessment tools for specific types of offending have had limited success. Limited research has been conducted in predicting and reducing certain types of reoffending, including domestic violence, sexual offending and knife crime.

Sex Offenders

Whilst RNR tools can predict general recidivism for sex offenders, more research is required in predicting sexual recidivism^[223]. A number of unique tools have been developed to assess sexual recidivism but there has been significant debate about their use, however, due to potential misclassification of individuals to risk groups^[224], and uncertainty as to how risk groups should be interpreted and used in practice^[225]. A number of evaluation studies have been undertaken which, taken together, show mixed results as to which type of assessment is most effective, with some evidence that both actuarial and structured clinical assessment tools can help in assessing risk^[226].

Evidence is mixed, but there is some evidence that those who receive treatment have lower rates of sexual reconviction^[227]. CBT is typically^[228], but not universally^[229], found to be the most effective intervention for reducing sexual recidivism. Programmes for sexual offenders should be matched to the risk level of the participants, and using inappropriate participants can skew findings as to programme efficacy.^[230] Interventions with sex offenders have been found to work best with medium and high-risk offenders^[231]. Research findings suggest that low and high risk offenders should be kept separate during treatment^[232] and researchers have suggested that female sexual offenders are qualitatively different from male sexual offenders and so should not be involved in group treatment with male sexual offenders^[233]. There are presently no validated risk assessment techniques for female sexual offenders, and so researchers assert that clinical judgement must be used^[234]. Risk assessment methods validated with women who offend can be used to assess general risk of reoffending among female sexual offenders, but not risk of sexual reoffending^[235].

Domestic abuse

There are two main types of existing treatment for people who commit domestic abuse^[236]. The first is based on CBT, built on the belief that domestic abuse is based on 'cognitive distortions' and inability to appropriately process feelings. The second is the Duluth model, designed around feminist psychoeducation, which aims to re-educate violent men on their beliefs about domestic abuse and women, as well as providing anger and stress management and relationship skills training. The Duluth model emphasises that domestic abuse is used by men as a tool to control women. However, both types of interventions have only shown small capacity to reduce domestic abuse, and further research and development of programmes is required^[237]. An evaluation of the Caledonian System, which an integrated approach to address men's domestic abuse in Scotland, will be commissioned by the Scottish Government in 2014 and will report in due course.

A review of research literature on effective interventions and practices to deal with perpetrators of violence against women includes stalking and rape in addition to domestic abuse^[238]. It found that the effectiveness of Intervention Orders (IOs) - commonly used to try curtail stalking - is not yet fully established. Regarding rape, it reported that the effectiveness of Sexual Offender Treatment Programmes (SOTPs) can be difficult to determine for rape offenders as not all sexual offenders are offered treatment, relatively few rapists complete treatment programmes, and even fewer programmes are designed specifically for rapists.

Holistic interventions

Holistic interventions that address multiple criminogenic needs are more likely to be effective in reducing reoffending. The evidence suggests that offenders often experience multiple problems, many of which are considered "criminogenic" in the sense that they contribute directly towards offending.^[239] It has, therefore, been argued that multi-modal, holistic and sequenced interventions, which address a range of problems, are more likely to be effective in reducing reoffending^[240]. In 2002, a report by the Social Exclusion Unit^[241] found that:

• prisoners are 13 times more likely to have been in care as a child;
• 63% of young people have substance misuse issues on admission to prison;
• of all prisoners 80% have writing, 65% have numeracy; and 50% have reading skills of an 11 year old;
• 25% of these young people have clinically significant communication impairment.

Data from 10,000 assessments of offenders' needs in England and Wales using the Offender Assessment System (OASys) show that over half of offenders had needs related to education, employment and thinking styles. Additionally, just over half of offenders in custody were assessed as having a need related to their lifestyle and associates. Drug problems were more common among offenders in custody (39% of those assessed) than in the community (27% of those assessed). Overall, offenders in custody were found to have a greater number of needs. Among adult reception prisoners that took part in the Surveying Prisoner Crime Reduction (SPCR) study conducted in England and Wales, 68% reported that having a job would help them desist from offending, followed by having a place to live (60%)^[242]. A recent study of reoffending amongst those in England and Wales on Community Orders found offenders often had multiple, complex needs, and the rate of reoffending increased with the number of criminogenic needs.^[243] Almost nine out of ten offenders with a drug misuse need, for example, had three or more other needs. The following factors were independently associated with the likelihood of reoffending: having previous criminal offending (as shown by the OGRS score), committing an acquisitive (rather than violent) offence, having a drug misuse need in the early months of a Community Order, having an unstable accommodation need, having a pro-criminal attitude, and elements of their supervision (see below).

These findings are congruent with desistance studies in which offenders report they value practical support more than any other type of intervention^[244] even though they are not necessarily accustomed to actively seeking help from outside agencies to solve their problems^[245]. This suggests offender managers might need to adopt a more proactive approach to solving offenders' practical needs while, at the same time, trying to enhance their problem-solving skills and empower them to search out suitable help when needed.

A holistic approach to addressing offenders' needs further means that ongoing support should be available as required. For example, there is strong evidence that provision of practical support in prison is unlikely to have a lasting impact on the risk of reoffending unless it continues upon release^[246]. Aftercare should, therefore, form part of a comprehensive intervention package. It is also important that the services provided are appropriately sequenced: for example, employment, while critical in the longer term, is often not a realistic short-term goal until other issues and needs have been addressed.

Holistic interventions for women who offend

Qualitative research in England has stressed the importance of practical factors in some women's offending, especially by those who have high levels of need^[247]. McDermott contends that a strict focus on cognitive processes can obscure the importance of these multiple needs, including structural factors such as unemployment. The report also recommends that holistic services have self-referral or drop-in facilities, rather than those that can only be accessed through court order^[248].

It has also been suggested that single-sex services should be made available to women who offend^[249]. In Scotland, the 218 centre in Glasgow offered an innovative, women-only holistic service designed to divert vulnerable women away from custody. An evaluation of 218 was published in 2006 but could not provide clear evidence on whether residence at the centre reduced reoffending. However, the evaluation suggested that the effectiveness of a holistic programme like 218 is often difficult to measure in quantifiable terms which may miss the benefits of service such as 218 in long-term crime prevention^[250]. The evaluation also revealed that there are numerous perceived benefits associated with the range and level of services provided at 218 which are not offered over the course of short-term custodial sentences. The evaluation found that women who used the services available at 218 identified significant decreases in drug and/or alcohol use (83%), improvements in their health and well-being (67%), access to stable accommodation and referrals to longer-term support services. Although the quantifiable effects of the programme on reoffending could not be demonstrated, the feedback on 218 from service users was almost universally positive.

There is some evidence to suggest that the sequencing of interventions in holistic approaches is important. For example, a study in the US found that even women offenders who have experienced victimisation said they found services that offered 'long term tangible support' as more 'helpful' than therapeutic or support services - the most helpful service being welfare benefits^[251]. This accords with results from several studies which conclude that while victimisation experiences possibly play a role in the onset of criminal offending, they are not associated with recidivism^[252]. As a result, longer term and more complex needs such as dealing with stress and mental health might be better dealt with after basic, practical needs are addressed first.

Holistic interventions for women who offend in Scotland

In response to the Commission on Women Offenders report and its recommendations on service redesign in Scotland, the Scottish Government is developing 'one-stop' Community Justice Centres (CJCs) for women who offend. These centres will involve workers from multiple agencies providing assistance with addiction, mental health, housing, debt, education and employment for women who offend^[253]. Given the lack of control groups involved in the evaluation it will not be possible to evaluate the direct effects of CJCs in reducing reoffending. However, the evaluation will consider to what extend CJCs have been able to undertake activities found to support desistance from crime. The evaluation is yet to report its results.

Holistic interventions for young offenders

Young people who offend require holistic interventions. The international research literature shows that the through-care strategies with the most favourable results in relation to reoffending rates are 'holistic'; that is, focused on the whole range of an individuals' needs and integrated with support in the prison and in the community. This support is necessary not only in the early weeks of readjustment on release but also in the long term.^[254] Indispensable processes for successful 'habilitation' or 'integration' include teaching basic skills, helping young people to develop the capacity to cope with their 'survival' needs in the outside world and establishing meaningful links whilst in prison with a range of community services that can offer continuing support^[255].

Fraser et al.^[256] point to similar evidence based on systematic reviews of programmes and interventions in the US. In terms of the reintegration of young people who had offended, early intervention with those starting to offend and reducing reoffending through community programmes the following types of programmes had success or were found to be 'promising':

• Education and health home visits and programmes for pre-school intervention;
• capacity building in schools; awareness raising campaigns in schools with clear messages and prosocial norms;
• training in 'social competency' e.g. managing stress, self-control, problem solving, emotional intelligence.
• The use of civil and criminal responses as situational management to reduce reoffending (e.g. responding quickly to breaches);
• specific rehabilitation programmes for juvenile (and adult) re-offenders 'using treatment appropriate to their risk factors';
• proactive arrests for carrying weapons intensive supervision and aftercare for more serious offenders;
• proactive police strategies focusing on specific offences delivered in a respectful manner e.g. polite field interrogation of suspicious people;
• community based mentoring;
• after-school prosocial activities;
• residential employment focussed interventions for youths;
• thinking skills intervention for high risk youth;
• situational risk management e.g. metal detectors in schools;
• 'gang' monitoring by community workers, probation and police.

In addition to these promising strategies as discussed by Fraser et al., a systematic review undertaken in 1998 of over 200 experimental or quasi-experimental studies of interventions with young people who offend (mainly males aged between 10 and 21 years) found that three intervention types showed the strongest and most consistent evidence of reducing re-offending. These were interpersonal skills training, individual structured counselling and behavioural programmes. The review found that these interventions reduced re-offending by about 40 percent^[257]. A recent meta-analysis also found that aftercare has proven to be effective in reducing reoffending in young people, but the study authors reinforce that its effectiveness varies by the type of aftercare received, the quality of implementation and the age and assessed risk level of the participant^[258].

A final holistic intervention which may help to reduce offending in young people is problem-solving courts. These courts involve workers from social and healthcare services as well as legal professionals to help provide support for young. A systematic review found that evidence for their effectiveness is mixed, with some studies showed positive effects and others showing few benefits compared to controls. Many evaluation studies displayed methodological problems, such as small samples and lack of appropriate comparison groups. As such more research is required into their effectiveness, including qualitative research to understand why specific outcomes were observed^[259].

Fraser et al.^[260] suggest that the following strategies are ineffective in reducing reoffending for young people:

• short term non-residential employment interventions,
• summer work programmes,
• diversion from court to job training for young people, arrest for minor offences,
• increased arrests on drug dealing locations,
• 'boot' camps or 'scared straight' programmes (taking young people who offend to adult prisons),
• 'shock' probation, parole or sentencing,
• home detention and electronic monitoring,
• vague, unstructured rehabilitation programmes.

Motivation and strengths-based approaches

Interventions that are appropriately matched to the offenders' level of motivation are more likely to be effective in reducing reoffending. It is a consistent finding in the desistance literature that only those offenders who are sufficiently motivated to change and are optimistic about the future will manage to desist from offending. Therefore, interventions are more likely to be successful if they target motivational factors and provide a sense of hope^[261]. Research suggests that only a minority of offenders are prepared for change at the start of an intervention^[262], and so in most cases some motivational work would be required to increase participation and retention in services. Motivation should, therefore, be seen not simply as a selection criterion but a treatment need. Especially for those at the start of the journey towards desistance, providing a sense of hope for the future can help promote and sustain their motivation to change. Offenders who are contemplating change need to believe that an alternative future is possible and, therefore, it is worth changing to accomplish future goals^[263]. However research from America suggests that, whilst motivation to change may be a necessary condition of reducing reoffending, it may not be sufficient in itself to reduce reoffending if it is not coupled with tangible resources to support change^[264]. Similarly, the Sheffield Pathways out of Crime study found that despite wanting to desist, many members of the study still re-offended^[265]. This was attributed in part to a lack of financial resources and leisure opportunities.

Strategies to increase motivation to change include setting realistic goals appropriately matched to the offenders' stage of readiness to change, reinforcing positive behaviours on a one-to-one basis and within a group and building helping relationships (e.g. buddy systems, self-help groups). It is also important that professionals help offenders recognise the positive changes that desistance from offending can bring to themselves and their environment. Offenders will be motivated to change only when the pros of changing outweigh the cons and change is more likely to be sustained if it is chosen freely rather than imposed^[266]. There is some evidence that motivational interviewing can help offenders recognise their problems as well as initiate and sustain motivation to change throughout treatment^[267].

Focusing on offenders' personal strengths rather than over-emphasising risks is advocated in the literature as an effective way to increase motivation^[268]. This strengths-based approach to treatment forms the basis of the Good Lives Model (GLM) which has been used with some success with sex offenders^[269].

The Good Lives Model

GLM aims to "equip clients with internal and external resources to live a good or better life"^[270]. GLM is based on the idea that all people attempt to attain a number of "primary human goods"^[271]. The configuration of these goods varies between individuals, but they are considered by each individual as "intrinsically beneficial" and represent a person's life values and priorities. Human activity is directed towards obtaining primary goods of:

• "life (including healthy living and functioning);
• knowledge;
• excellence in play;
• excellence in work (including mastery experiences);
• excellence in agency (i.e., autonomy and self-directedness);
• inner peace (i.e., freedom from emotional turmoil and stress);
• friendship (including intimate, romantic and family relationships);
• community;
• spirituality (in the broad sense of finding meaning and purpose in life);
• happiness; and
• creativity."^[272]

Secondary or instrumental goods are the methods by which these primary goods are attained. To achieve these goods all people, including those who offend, have a good life plan (whether explicit or implicit). In the GLM offending "results from flaws in an individual's life plan"^[273]. Primary goods can be sought directly via offending, or indirectly, wherein problems in the pursuit of primary goods by socially acceptable means leads to offending. The core of GLM treatment is attempting to achieve primary goods through socially acceptable means.

Links between the GLM and RNR

The proponents of GLM state that it maps on to each facet of the RNR model^[274]. The risk principle can be incorporated by varying the intensity of supervision based on a client's level of risk. The responsivity principle is incorporated through the targeting of goods that are identified as important by individual clients. It should be noted that this is a somewhat different interpretation of the responsivity principle than in RNR, as this is client directed, rather than treatment being matched to the characteristics of the individual. There is the most divergence between the approaches in their conceptualization and treatment of need. In GLM criminogenic needs are considered "internal or external barriers towards living a good life"^[275]. In GLM non-criminogenic needs - that is, needs which are not correlated with reoffending - are considered important for "client engagement". Factors not directly related to recidivism (any of the areas outlined above which do not fit in with the central eight, such as inner peace, creativity or spirituality) may still be important parts of living a good life, and addressing these issues can help clients to stay engaged with treatment programmes. Indeed, development of GLM is in part motivated on the high attrition rates for RNR programmes^[276]. In contrast, the RNR approach considers these factors either as being a waste of resources^[277] or as being outside the responsibility of the CJS and best served by other agencies^[278].

The value of the GLM has been much debated by those who contend that only criminogenic needs should be addressed in offender treatment, as in the RNR approach. Proponents of RNR suggest that GLM does not add significantly to the effectiveness of interventions based on RNR^[279], citing a lack of studies supporting increased effectiveness for interventions using GLM. However, others have contended that in practice the two approaches are very similar^[280]. As such, further research is required into integrating the two perspectives, especially focusing on offenders' adoption of a 'reformed' identity'^[281].

Interventions to develop social bonds

Interventions that help offenders develop prosocial social networks have significantly higher chances of success in reducing reoffending. Desistance studies have found that rebuilding ties with family, friends and the wider community and developing new prosocial relationships through work or marriage are important aspects of desisting from crime^[282]. Furthermore, research suggests that offenders who feel a welcomed part of society are less likely to reoffend compared to those who feel stigmatised^[283]. It is therefore important that criminal justice professionals work not only with offenders but also with their families, friends and the wider community (e.g. employers, community groups, the voluntary sector) to ensure prosocial and positive relationships can be developed and sustained^[284]. This is particularly true for offenders who have spent long periods of time in prison and may not have access to an active network of contacts. Interviews with women offenders raise the importance of successful reintegration and indicate that rehabilitation will depend on the active support provided by family and close friends. Positive support is likely to have a significant impact on their desistance from crime after release from custody^[285].

Family-based interventions for young people

Family-based interventions encompass programmes that focus on improving parenting skills and relationships within the family. Parenting interventions have traditionally been used to prevent the onset and continuation of offending among juvenile offenders, as there is evidence that poor parenting skills are associated with an increased risk of offending among young people^[286]. Systematic reviews of parenting programmes have consistently found small but statistically significant effects on juvenile recidivism. The most effective programmes are reported to be multi-systemic therapy which involves work with the young person, his or her family and school staff, school-based child and parent training programmes, parent training plus day-care provision and home visiting^[287]. Positive results have also been reported for functional or behavioural family therapy, family empowerment and allied therapeutic approaches, especially when used with young people who have committed more serious offences^[288]. However, a recent review stated that multi-systemic therapy as well as other family interventions such as functional family therapy and multidimensional treatment foster care, is promising but limited^[289]. The authors suggest that programmes must be well implemented and that flexible mental health services should be made available for young people. It should be noted that as many of these studies were undertaken in the US more work needs to be undertaken on replicating these findings elsewhere^[290]. The review also stressed that successful interventions were found to work at multiple levels (such youth, family, peer, school and neighbourhood) rather than just focusing on the individual.

Despite these positive findings for some young people who offend, Fraser et al. caution that the research literature identifies that the family should not be the sole focus of any intervention work^[291]. Those young people with the highest level of need are often those who are no longer part of any family unit and who, for various reasons, may not have any contact with parents. Furthermore, for those young people who remain with their families, it has been highlighted that there is a need to look beyond the family to the wider community context that influences and impacts on parents' ability to parent effectively. They highlight that there are a number of different programmes of support and intervention, appropriate to a range of need and age and stage of child/young person development, that have been demonstrated to have some degree of success in addressing risk factors within families^[292]. MacQueen et al. also caution that evidence around 'what works' in a Scottish or UK context is limited and much of the evaluative research had been based on American populations^[293].

Family interventions for adults who offend

Despite the success of family-based interventions with young people, their use with adults has not been evaluated. As a consequence, there is not sufficient evidence that family interventions to improve adults' relationships with their families can reduce reoffending. Potential exceptions are that of home leave and family visits in prison^[294]. Mears et al. found that receiving visits in prison was associated with reduced reoffending^[295]. For those receiving eight or more visits the effects of visitation were comparable to well-implemented cognitive behavioural programmes, associated with a reduction in reoffending of around eight percentage points. However, these are the results of a single study and more research is required into how visitation is intended to reduce recidivism. These findings echo qualitative research which suggests that one of the most significant triggers of change and sustained abstinence from offending is the formation and strengthening of family relationships. For example, Healy in her comparative study of desisters and non-desisters in Ireland found that the desire to live up to family responsibilities and expectations was one of the biggest triggers of the decision to abstain from offending^[296].

Family-based interventions might be particularly beneficial for women offenders as reviews suggest interpersonal needs related to the family is one of the strongest predictors of positive outcomes among this group. Some research provides an insight into what type of family interventions would be most effective with women offenders. Dowden and Andrew's meta-analysis of several family-based interventions^[297] found that programmes treating family processes yielded strongest reductions in reoffending for samples of women. This finding has been confirmed by more recent studies that found that programmes targeting family relationships for female offenders yielded the greatest treatment effects. The meta-analysis also identified effective targets for family intervention (i.e. 'needs') in terms of which aspects of family interventions yielded the best results in terms of reduced re-offending, and which targets did not seem promising. The strongest positive association with reduced re-offending came from intervention programmes which focused on interpersonal criminogenic needs (family processes and anti-social associates), followed by those which focused on personal criminogenic needs (anti-social cognition and self-control). 'Family process' needs were defined as those around 'attachment', 'affection' and 'supervision'. Family interventions had a statistically significant association with reduced re-offending when they were clearly focused on these three family-related areas of need. Less focused forms of family intervention, or family interventions which had different targets (not specified in the paper), were statistically significantly associated with higher rates of re-offending^[298]. Other studies have found that for women positive friendships and bonding with their children are protective factors^[299]. In contrast desistence in men is more closely linked with the break-up of a pro criminal peer group, and establishing a stable intimate relationship^[300]. Moreover, research suggests that the protective effect of intimate relationships in male offenders is age related^[301].

Relationships with anti-social peers

Relationships with anti-social associates has been described as 'one of the most potent predictors of reoffending' and is therefore recommended as a priority treatment target^[302]. Meta-analytic research has confirmed that this area is an effective treatment target as there is a strong positive association between correctional programming in the area of 'associates' and reduced reoffending for studies with predominantly or entirely female samples. Other studies have found that a composite of anti-social peers/attitudes comprised the greatest risk factor for young girls. In a qualitative study of offending and desistance conducted in Scotland women often attributed their initiation into problematic drug use to their relationship with partners who were involved in drug use and associated offending^[303]. The initiation of women into drug use was also identified as a pathway to women's offending by professionals (such as police officers and social workers) who observed that women often committed offences (such as shoplifting) or became involved in prostitution to supply both themselves and their partners with drugs. However, in some cases the influence of male partners on women's offending (and substance misuse) was believed by workers to be more diffuse through experiences of physical and emotional abuse and financial control or exploitation^[304]. In sum, while there is some disagreement between research findings, the greater and more robust evidence suggests that family relationships and associate issues present a valuable treatment target for girls and women.

Agency, self-efficacy and problem-solving skills

Interventions that aim to increase offenders' sense of agency, self-efficacy and good problem-solving skills are more likely to be effective in reducing reoffending. Offenders are more likely to eventually desist from offending if they manage to acquire a sense of agency and control over their lives and a more positive outlook on their future prospects. Therefore, interventions that aim to enhance perceived levels of self-efficacy and problem-solving skills are more likely to be successful in reducing reoffending. This was also found by McIvor et al. specifically in relation to women^[305].

Employment programmes

There is mixed evidence, mainly from the US, on the effectiveness of employment programmes in reducing reoffending. There is strong evidence that offenders with stable and quality employment are less likely to reoffend^[306]. However, there is mixed evidence, mainly from the US, on the effectiveness of interventions designed to improve employment prospects of offenders. The first published US systematic review of educational, vocational and employment programmes for adult offenders in prison and community settings found lower reconviction rates for participants compared to non-participants^[307]. In contrast, a more recent US systematic review of community-based employment programmes reached different conclusions, finding no significant difference in the likelihood of re-arrest between participants and non-participants. This has led researchers to conclude that stand-alone employment programmes are unlikely to be effective unless they are combined with motivational, social, health and educational support services to help address other criminogenic needs of offenders that may act as barriers to finding employment. These barriers can include, learning difficulties, mental illness and substance abuse^[308]. It is possible that these divergent findings may be due to differences between specific employment programmes. For example, Yahner and Zweig suggest that transitional jobs programmes are more effective than employment programmes which focus on offenders sending out a set number of job applications per week^[309]. However, a recent evaluation of a transitional job programme in the US showed that, whilst high-risk offenders who completed the jobs programme were less likely to reoffend than a comparison group, those who completed the programme were no more likely to find employment^[310]. This suggests that more research is required into the mechanisms by which employment programmes are intended to reduce reoffending.

Evidence from the UK about the effectiveness of employment programmes is more uncertain and tends to come from process evaluations of probation-led programmes. These evaluations have showed that the most successful elements of effective employment programmes are: strong local partnership; training related to local employment needs and opportunities; long-term funding and generous lead-in times^[311]. In addition, the outcome evaluation of the probation-led ASSET programme, that offered employment-related advice, training and work placements to offenders aged 16-25 years, found that participants were slower to reoffend and had a lower one-year reconviction rate (43%) compared to those who were referred but did not attend (56%). However, the authors acknowledged the limitations of their research design noting that their positive results might be attributed to selection effects, that is, that participants did better because they were more motivated to change. The ASSET programme was less successful in terms of improving employment outcomes, with only 13% of participants managing to secure employment over the lifetime of the project^[312]. Many factors may have contributed to the limited success of the ASSET programme in securing employment including unwillingness on behalf of companies to employ ex-offenders and lack of sufficient motivation from offenders to follow-up job opportunities. To sustain motivation, offenders should be instructed to view the attainment of a good job as the end result of a gradual process rather than as a single event^[313]. Further work by the Department of Work and Pensions suggests that sharing of information between agencies can aid offenders' employment prospects^[314]. Finally, research suggests that the most successful programmes for getting prisoners back into employment are those which coordinate work before and after release from prison^[315].

There is evidence that Black British offenders are the least resourced to find suitable employment compared to other ethnic minority groups such as Indians or Bangladeshis who are more likely to receive some support from family members^[316]. Therefore, interventions directed at improving employment prospects would be particularly beneficial for those of Black British origin.

Employment programmes for women who offend

For women, offending has also been shown to be associated with a lack of education, accommodation and employment, although the level of need appears to be lower among female than male offenders^[317]. Unfortunately, there is a dearth of information on the outcomes for women who engage in employment programmes. One of the few evaluated employment programmes for women offenders seems to suggest that approaches should offer long term, holistic approach and that the effectiveness of the programme is moderated by the motivation of the offender to obtain employment^[318]. A frequently cited employment programme run in Victoria, Australia commenced in prisons (six months prior to release) and offered links to employment services local to where the women lived. The programme also confronted the challenges of finding work and offered life-skills preparation, placement in employment and skills in retaining employment. Lawrence et al.^[319] found that participation in prison based treatment programmes and community based treatment programmes was positive; it led to lower recidivism rates for women who had previously been in custody - in June 2005 there was a 41% reduction in return to custody by women in Victoria. Within the first 2 years of the CSEPP pilot programme there was a 27% reduction in reoffending rate by women registered with the programme. Interviews with female ex-prisoners affirmed that a critical element to success in reducing reoffending was the individual readiness to change^[320]. It should be noted that the evaluation did not use a non-treatment comparison group and that the women in the programme were motivated to find employment so it is not known if the programme would have been as successful with a less motivated group of women.

Education programmes

Stand-alone education programmes are unlikely to reduce reoffending. There is evidence to suggest that the association between lack of basic skills education and reoffending is indirect, meaning that poor educational skills can increase the risk of reoffending only to the extent they impact negatively on other criminogenic needs such as employment prospects^[321]. McGuire^[322] in his review of offender rehabilitation programmes concluded that vocational training activities without associated links to tangible employment prospects are unlikely to lead to reductions in reoffending. Another UK review of prison-based educational programmes found mixed evidence of effectiveness, with greater benefits reported among high-risk offenders^[323]. Matrix Knowledge Group^[324] found some evidence that a prison sentence combined with vocational or educational interventions represents value for money compared to imprisonment without rehabilitation.

An analysis of two US studies found that overall there is some evidence to support the view that general education has some beneficial effects for female offenders and could be a fruitful area for further work^[325]. Similarly, in a small-scale study of young people in New Zealand, reading comprehension was found to predict recidivism, even when controlling for other risk factors. As a result the authors suggest that targeted educational programmes should be provided for young people in prison^[326]. However, the small size and composition of the sample, with a very high proportion of people with learning disabilities, raise questions about the generalizability of these findings. Conversely, there is some evidence that prison-based work and apprenticeship schemes are not of use and may even be detrimental^[327]. However, these conclusions are based on just two studies, both from the US, and findings are complicated by the lack of detail on the differences between groups.

Drugs programmes

Drug treatment programmes have, on average, a positive impact on reoffending and offer value for money. Drug abuse is a risk factor for reoffending and a significant proportion of offenders are assessed as having this particular criminogenic need^[328]. A recent meta-analysis of drug-treatment programmes in Europe found that treatment reduced recidivism in drug-using offenders by around 30%, from roughly 40% in the treatment group and around 59% in the non-treatment group^[329]. A review by Holloway et al. found that more intensive interventions that focus on the multiple problems of medium-to-high risk problem drug users are more likely to bring about reductions in reoffending than less intensive programmes and that men benefit more compared to women and young people who offend compared to old^[330]. Offenders that enter treatment quickly, stay in treatment for as long as required and are provided with wider support are more likely to desist from offending^[331]. There is strong evidence that prison-based treatment programmes are most effective when followed-up with community aftercare supports^[332].

A number of different drug treatment programmes have been used for offenders with drug problems, including therapeutic communities (TCs), drug courts, cognitive-behavioural programmes and pharmacological substitution.

Therapeutic communities

A meta-analysis by Holloway et al. found that TCs were one of the most effective interventions to reduce drug-related offending^[333]. A recent systematic review by Mitchell et al. also found that TCs were consistently associated with moderate reductions in both reoffending and drug use, and were the form of drug intervention most supported by the available data^[334]. Other reviews of TCs have suggested that there is some evidence from US studies that TCs can reduce risk of reoffending for some offenders, but that success may depend on readiness for treatment^[335]. In an investigation of the long-term effects of participation in a TC was associated with "persistent, significant and quite strong" negative effect on future reconviction over a 12-18 year follow-up period^[336], although the size of the effect was variable. However, a different study found no treatment effect of TC participation after five years^[337]. In general there is less evidence from UK studies about the effectiveness of TCs^[338].

Drug courts

Holloway et al.'s meta-analysis also found that drug courts were effective interventions in reducing drug-related offending^[339]. A separate review of the effectiveness of drug courts in reducing reoffending found that the vast majority of studies reported a reduction in offending for drug court participants^[340]. Adult drug courts were found to be more effective than youth drug courts, although both showed reductions in recidivism. However, there was significant variability between studies, suggesting that more research is required into establishing what features of drug courts help to reduce reoffending. Once more, evidence for the effectiveness of drug courts varies between the US and the UK^[341]. Researchers believe that quality of, and access to, treatment is a mediating factor for drug courts in the UK, as well as continuity of staff^[342].

Due to the complex nature of drug courts it is not known exactly what features of drug courts are effective in reducing recidivism^[343]. However, factors suggested as being associated with reduced reoffending include the judge's level of experience, the amount of time a person spends in front of the judge during the status review hearing, collaboration between different agencies, and a programme length of at least one year^[344].

Cognitive-behavioural programmes and drugs

Another systematic review of drug treatment programmes for offenders found that programmes with a cognitive-behavioural component had a small but statistically significant positive effect on reducing drug use relapse when compared to standard correctional treatment^[345]. This finding was echoed by Bahr et al., who found that those who completed an intensive, CBT-based drugs treatment programme in prison had lower recidivism than a matched comparison group^[346]. However, these results have not yet been replicated elsewhere.

Pharmacological substitution

Pharmacological substitution (that is, providing drug misusing offenders with alternative drugs such as methadone or proscribed heroin) was found to be the most effective treatment for drug using offenders in a meta-analysis of interventions in Europe^[347]. However, there were only a small-number of non-pharmacological studies (e.g. therapeutic communities, RNR programmes) available for inclusion in this review, and the review focused mostly on opiate misuse. The review concluded that more research is needed into treatments for other drug types.

UK evidence on drugs programmes

Positive results have been reported in Scotland from evaluations of DTTO orders, drug court pilots, targeted intelligence-led arrest referral schemes, like the Persistent Offenders Project (POP) in Glasgow, and some prison-based drug-treatment programmes such as the Saughton Drug Reduction Programme^[348]. In England, prisoners who completed the 12-step Rehabilitation of Addicted Prisoners Trust (RAPt) programme achieved greater reductions in self-reported drug use and offending compared to dropouts and non-starters, although it was not possible to separate out self-selection effects^[349]. As noted above, the evidence for the effectiveness of TCs and drug courts is weaker in the UK than it is in the US^[350].

Drugs and women who offend

In Scotland, a significant number of women in prison are drug users, with a high proportion imprisoned for offences directly related to problem drug use^[351]. An international study has found drug use to be especially predictive of reoffending in women^[352]. Whist there is speculation about the characteristics likely to lead to effective services that meet the needs of female drug users there are very few studies that have tested the effectiveness of drug interventions (or of gender-specific responsivity factors) on reducing re-offending in women.

A Rapid Evidence Assessment undertaken by the Home Office in 2008 is encouraging about the efficacy for women of some forms of treatment^[353]. There was evidence that aftercare, in particular residential treatment provision, enhanced the effects of prison-based treatment in the short term. There was no evidence, however, that the positive effects persisted beyond two years post-release: one study that followed participants up for this long found that initially statistically significantly positive effects became non-significant at two years. In a recent systematic review Tripodi et al. found that substance abuse programmes in prison can reduce reoffending in women. However, only six studies were included in this review, and all data came from the US. There are therefore some questions about the generalizability of these findings to Scotland.^[354] Similarly, in a review of the effectiveness of drugs intervention in reducing reoffending in women, Perry et al. found that there was some evidence that interventions can be effective, but that there was large variation between the seven studies reviewed^[355]. Studies have also found that parental drug abuse has a more profoundly negative effect on women than men, which is consistent with research findings cited in this paper that dysfunctional family dynamics influence recidivism for girls and women^[356].

Research on women drug users suggests that not all drug use is criminogenic and that recreational and occasional drug use are not strong predictors of reoffending^[357]. This study also found that the type of classification used to define 'substance abuse' can affect prediction strength for reoffending - if drugs had be consumed prior to the commission of the original offence then substance abuse was predictive of reoffending, but that the generic DSM-III diagnostic criteria was not a good predictor of reoffending. If the aim of drug interventions is to reduce reoffending, then this may suggest that intensive interventions should be targeted at only those with criminogenic, as opposed to recreational, drug use.

Cost-effectiveness of drug treatment programmes

There is evidence that drug treatment represents value for money. A recent Home Office study (DTORS) estimated that for each £1 spent on structured drug treatment, on average society saves £2.50 in terms of reduced crime, costs to the criminal justice system and health and social care services^[358]. Also, a recent Scottish review of interventions for drug-using offenders found that the costs of crime are reduced significantly for individuals in treatment (£1,536 costs per year for those in treatment for more than one year compared to £12,713 per year for individuals with no intervention in place)^[359].

Young people, substance misuse and offending

There is a well-established link between substance misuse and offending behaviour. In their review of youth violence in Scotland, Fraser et al. highlight that research with young people in custody points to the significant role of substance misuse, especially excessive drinking, in the backgrounds of convicted violent offenders, both male and female. Some studies have reported that young people who have offended state that they have been under the influence of alcohol and/or drugs when committing offences and that violent offences are commonly perpetrated alongside offences with a financial motivation (e.g. shoplifting and robbery are often committed to finance a drug habit)^[360]. Similarly, the Edinburgh Study found that those young people who reported being multiple substance users reported higher levels of delinquency, both in volume and variety of offences, than single substance users and non-users ^[361]. However, none of the research covered in this review explicitly examined the impact of drug treatment programmes on young people who offend, although many of the holistic programmes described above include a drugs treatment component.

Alcohol programmes

There is emerging evidence that alcohol-brief interventions can reduce alcohol misuse, however their effect on reoffending has not been widely investigated. Alcohol misuse increases the risk of reoffending and there is evidence to suggest its prevalence among offenders is increasing^[362]. However there is as yet no evidence to show a direct effect of alcohol treatment on reduced reoffending^[363], although alcohol interventions can reduce alcohol problems more generally^[364].

There is emerging evidence from the health literature that alcohol-brief interventions - short, evidence-based, structured conversations about alcohol consumption^[365] - based on motivational interviewing techniques can be effective in reducing low to moderate alcohol misuse^[366]. NHS Scotland completed a study of ABI implementation in 2011 which showed that ABIs were useful in assessing levels of alcohol issues, but the pilot was unable to ascertain the impact on offending behaviour^[367]. A review of interventions for the treatment of alcohol problems among the wider population found that cognitive behavioural and mutual support approaches such as 12-step were the most successful in reducing alcohol misuse^[368].

Mental health programmes

Little evidence is available on the effectiveness of mental health interventions in prison and community justice settings. Mental health problems are disproportionately prevalent in the prison population, and especially among women prisoners^[369]. A large scale survey published in 1998 found that around three quarters of sentenced prisoners suffer from two or more mental disorders, compared to less than one-twentieth (4%) of the general population^[370]. The Surveying Prisoner Crime Reduction (SPCR) survey of 1,435 adult reception prisoners in England and Wales found that more than a quarter (26%) of women reported having been treated and/or counselled for a mental health and/or emotional problem in the year before custody, compared with 16% of men.

Morgan et al. conducted a meta-analysis of available studies which failed to show any significant association between treatment for offenders with mental illness and reduced recidivism, although some studies did show positive results^[371]. However, only four studies were identified which investigated this outcome and so further research is required into the impact of treatment of mental illness and its impact upon reoffending. Whilst the body of evidence is limited, there is some suggestion that speciality probation caseloads may help to reduce reoffending for those with mental health issues^[372]. Researchers have also noted there are also gaps in service provision for young people aged 16-18^[373].

Community supervision and through-care

This section summarizes research into the supervision of offenders in the community, mentoring, through-care services, accommodation services and the public recognition of desistance. Research findings in these areas are presented and their potential to help reduce reoffending is discussed.

Supervision

A respectful, participatory and flexible relationship with a supervisor can trigger the motivation to change and promote desistance. Supervision should place adequate emphasis on helping offenders overcome practical obstacles to desistance such as unemployment and drug misuse. Probation can serve both as a deterrent and as a vehicle for change. In England and Wales, Rex found that for some probationers simply being on probation served as a deterrent whereas for others getting help on to how to solve practical problems was more important^[374]. Other research from Scotland and England confirms that offenders particularly value getting help from their supervisor on practical problems such as unemployment and lack of accommodation^[375]. However, a recent English study found that probation officers were found to be less able to help with personal and social problems. For example, when faced with significant practical difficulties, such as unemployment, probationers were often referred to external agencies. These experiences were often found to be frustrating and unhelpful, and the experience of being referred to another agency made some probationers feel undervalued^[376]. Another English study that followed-up a larger sample of 199 male and female probationers concluded that an individual's level of motivation to change and his or her social circumstances largely determined whether they would succeed in desisting from crime, with probation exerting a smaller influence^[377]. However, when probationers were interviewed four years later they were more inclined to see the value of what they had taken from probation^[378] and interviews with a group of the same probationers more than ten years after finishing probation supervision suggest that probation can impact people's lives long after it is finished^[379].

For men and women, qualitative research suggests that a good working relationship between the offender and his or her supervisor can act as a catalyst for change, especially when the offender has already taken the decision to give up crime, but it is unlikely to produce large reductions in reoffending on its own right^[380]. However, new quantitative research on offenders on Community Orders in England and Wales found that offenders who felt their Offender Manager understood their needs were significantly less likely to reoffend.^[381] Qualitative research in Scotland has shown that women often have different relationships with probation officers than men^[382]: of those interviewed, women tended to expect a certain level of support from social workers and were disappointed when they were treated by their support workers in an uninterested way.

Overall, research suggests that desistance is more likely to be achieved when a "working alliance" with the supervisor is developed^[383]. When asked about effective supervision, offenders often say they value being listened to and recognised as individuals^[384] and cite empathy, respect, flexibility, the ability to listen and professionalism as the defining characteristics of an effective working relationship with the supervisor that triggered change^[385]. It is also important to help offenders develop a sense of personal agency and higher levels of self-efficacy that will empower them to change. For these reasons, it has been argued that service users should be involved in co-designing the interventions that are meant to support them in desisting from crime^[386]. This suggestion is backed up by some evidence from evaluations of mentoring services that show mentoring is more likely to work when its goals are defined in agreement with the service user^[387] and when the amount of contact is proportionate to the offenders' level of needs^[388]. However, more research is required to understand what might be the most effective ways of involving service users in the design of interventions and how effective such approaches would be in reducing reoffending.

Features of successful probation supervisors

Overall, studies report more benefits in cases where the supervisor respects and fosters the offender's personal agency, focuses on strengths as well as criminogenic needs and risk and draws up an action plan in consultation with the offender^[389]. Utilizing prosocial modelling and reinforcement, problem solving techniques and cognitive techniques have also shown to be effective in supervision meetings^[390]. Keeping the same officer has also been associated with successful outcomes in probation^[391]. The use of prosocial modelling (where the case manager acts as a positive role model and encourages prosocial actions) has also been associated with higher rates of compliance and lower rates of reoffending^[392]. Other important features of supervision include dealing with relapse (e.g. breach, reoffending) in a proportionate and fair manner, rewarding progress towards change and involving users in the design of interventions. A recent Australian study suggested that using a wide range of skills in probation is likely to lead to better outcomes (in terms of reoffending) than those who only use a smaller range of skills^[393]. These findings point to the need to invest in interpersonal skills training for offender managers, and research has suggested that successful supervision requires appropriate staff training^[394].

Intensive supervision programmes, which emphasise control and sanctions over support, are ineffective in reducing reoffending. Petersilia and Turner^[395] evaluated intensive supervision programmes (ISPs) in the USA, in which parolees or probationers are placed in small caseloads, face regular and unannounced visits by supervising officers, and are threatened with revocation and incarceration if they misbehave. They found no reductions in recidivism and, in fact, the overall one-year recidivism rate for offenders in the ISPs was higher than for those in the probation-as-usual control groups (37% versus 33%).

Following on from this, Bonta et al (2008) found that supervision practice could be improved if less time was devoted to issues of compliance and more time spent focusing on criminogenic needs in particular, criminal peers and thinking styles^[396]. The authors suggested that training practitioners in such skills and techniques might improve the effectiveness of routine supervision. They followed up this hypothesis by designing and delivering such training, which involved structuring supervision sessions to adhere to RNR principles and include cognitive behavioural techniques. They demonstrated that by so doing, recidivism could be reduced by 15%, an outcome that was further improved when practitioners availed themselves of post-training support, in the forms of booster sessions and clinical supervision^[397]. This innovative model of community supervision, the Strategic Training Initiative in Community Supervision (STICS), has subsequently been implemented on a large scale in Canada, but acknowledging the well documented 'drift' that occurs the authors have built in important implementation and quality assurance strategies^[398]. The outcomes of this large scale implementation are some years away, but it already offers learning about how to improve the effectiveness of community supervision.

In contrast to the ISPs, the UK's Integrated Offender Management Schemes (IOMS) have been evaluated more positively, although the impact on reduced reoffending is unclear^[399]. Whilst implementation of schemes is locally variable, the IOMS emphasise coordinated working between offender management services, including the police, voluntary services and social workers. The schemes combine support, interventions appropriate to the individual and disruption visits. Disruption visits, usually carried out by police, aimed to re-engage and/or catch and control those who were disengaging from the IOMS and who were perceived to be at risk of reoffending. Although the evaluation does not attempt to directly measure impact on reoffending, reports from stakeholders and offenders were largely positive.

Mentoring

There is some promising evidence that mentoring can have positive effects on reduced reoffending, employability and motivation to change though more studies are needed to reach a safe conclusion. Relatively few UK studies have evaluated the effectiveness of mentoring schemes in reducing reoffending and addressing criminogenic needs, none of which have used a robust design with appropriate control groups. In Scotland, the evaluation of the Routes out of Prison project found that contact with the life coaches helped the majority of interviewed offenders to access services and increased their motivation to desist from offending^[400]. There is also an indication from studies in England and Wales that mentoring can lead to reduced reconviction rates among participants, increase chances of employability and contribute to positive changes in thinking styles when motivational interviewing techniques are used by mentors^[401]. Mentoring is especially likely to work with young people under 19 years of age who are still at risk^[402]. Mentoring is advocated in the literature as a potentially effective way of helping offenders build new social networks that can support the desistance process, and to the extent it can help extend social bonds, offer emotional support and encourage uptake of services is supported by desistance theory^[403].

There are even fewer studies that can determine the impact of mentoring on female reoffending. A rapid assessment of 18 studies which included mentoring and control groups) found that the research on impact on reoffending was limited but that overall mentoring reduced reoffending by four to eleven percent (although they point out that the more robust studies found no significant impact)^[404]. They found that mentoring was more successful if the mentor and mentee met at least once per week and for considerable periods. The programmes were also more successful if they targeted medium-high risk offenders, adhered to 'best practice' principles and if they were one of a number of interventions - a finding consistent with other studies suggesting that multi-modal interventions are generally more effective^[405]. A recent review into effective throughcare suggested that mentoring may be useful for "building upon inter-agency co-operation, supporting individuals with practical issues while also fostering self-reliance and individual responsibility", but there is little evidence about its impact upon outcomes^[406]. Other reviews of 'what works' have also found that transitional support programmes were generally effective in reducing recidivism^[407].

Mentoring in Scotland

Research is currently being undertaken in Scotland to evaluate whether Public Social Partnerships are delivering effective mentoring services. A report on the evaluation will be published on the Scottish Government's website after the evaluation has been concluded in 2015.

Through-care

Through-care may contribute to reducing reoffending by providing practical support to offenders leaving prison, although at present there is insufficient evidence to draw firm conclusions about the impact of through-care on reoffending. Through care is intended to reduce reoffending by addressing the needs of prisoners as they re-enter the community. An international review into the elements of effective through-care suggested that successful transitions involve contact with offenders while they are still in prison, continuity of contact in the community and for prisoners to be able to have input into the services that they receive^[408]. Consistency of contact can also help to build trust between service users and providers. However, it may be useful to separate the monitoring and support functions of through-care, as monitoring can reduce openness between service users and providers. The review stated that on the whole there is little robust evidence available to assess of the outcomes of through-care projects.

In Scotland, statutory through-care is provided for all those serving sentences of four years or longer, and voluntary through-care is available for those serving sentences of less than four years. A review of through-care in Scotland found that third-sector provision of through-care is important, but that funding for third sector providers is erratic^[409]. It was also noted that for many prisoners, especially those serving short sentences, basic welfare provisions such as opportunity to apply for homelessness benefits are not available until release from prison, which can take up to six weeks to process^[410]. This can leave some people without support at the point that they are released. Moreover, navigating bureaucratic issues involving access to services is especially difficult for those who may not have access to identification documents, telephones or contact addresses. In response to these issues, a Community Reintegration Pilot (CRP) has been undertaken in Scotland to improve service provision for short-term offenders. The CRP involves assessment of offenders' needs and interviews with offenders, prison officers and other interested partners, such as health care workers, to determine how needs will be met in the community. The evaluation of the CRP^[411] assesses to what extent the CRP functioned as planned, but is not able to assess long-term impacts of the pilot on reducing reoffending.

Accommodation

There is some promising evidence that holistic resettlement programmes can help to reduce reoffending though more studies are needed to reach a firm conclusion. One of the few holistic resettlement interventions that have been evaluated in the UK is the Pathfinders resettlement programme for prisoners on short-term sentences. The evaluation of the programme found that the offenders who completed a cognitive skills and attitudes training programme in prison and maintained post-release contact with a mentor who provided emotional support and help with practical problems showed lower reconviction rates and were more likely to be employed post release. Furthermore, the vast majority (80%) of the 51 offenders who were interviewed as part of the second phase of the evaluation said that the programme had helped them to control their substance misuse problem to some extent^[412].

The plethora of multiple and complex needs often faced by women offenders also signal a need for holistic approach to services. There is some strong international evidence that discharge planning and aftercare could lower recidivism rates for women^[413]. Studies have shown that holistic discharge planning with primary health care, peer support and social work input which started in prison and continued in the community can lower group risk of recidivism^[414].

Whilst no direct impact on reoffending has been observed, accommodation is considered important for desistance. There is also increasing consensus that it is more effective to re-house ex-offenders into mainstream rather than hostel accommodation. Having stable accommodation is known to support desistance from offending as it increases the chances of finding employment^[415] and, accommodation is considered a necessary condition for reducing reoffending^[416]. The study of transitional care in Scotland identified housing as one of the main problems encountered by short-term prisoners with drug problems on release from prison, and experiencing housing problems made it more likely that they would resume drug misuse^[417]. However, there is relatively little evidence on the effectiveness of different forms of help in securing accommodation for offenders. Resettlement might help reducing reoffending, but it is difficult to separate out the effect of accommodation on this^[418] and the evidence is limited as to whether resettlement leads to the formation of positive social bonds rather than reducing negative bonds^[419]. There is also mixed evidence on the effectiveness of hostel accommodation in reducing reoffending, with some evaluations reporting cases where this type of accommodation fostered the development of networks between offenders, thus sustaining a criminal lifestyle. This has led researchers in both Europe and North America to conclude that it is more effective to re-house offenders into mainstream accommodation with security of tenure, rather than into hostel accommodation^[420]. A review of through-care in Scotland suggested that the quality of accommodation available to offenders, particularly women, is poor, and the Commission on Women Offenders recommended a system of supported hostels or "scatter flats" to help reintegration into the community^[421]. The Scottish Government accepted recommendations by the Commission on Women Offenders to increase availability of supported accommodation, sustain tenancies for women when in custody and secure access to safe accommodation for women upon release from custody^[422]. This is to be achieved via working with local authorities, social landlords and third sector organizations.

There is also some research to suggest that early intervention may help prevent people losing their accommodation as they are taken into custody^[423]. A recent study in England and Wales found that having accommodation before imprisonment was found to be negatively associated with reoffending. This shows the value of preventing people from losing their accommodation while in custody. The study also found that people with accommodation problems were more likely to offend than others with similar criminal histories^[424].

A recent review of the quality of probation supervision noted that offenders are not necessarily accustomed to seeking help from outside agencies to solve accommodation problems; therefore a more proactive approach to supervision is required^[425]. To be able to sustain accommodation, offenders will also need advice in managing money and debt^[426]. There is evidence that accommodation is a particular issue for female prisoners who are more likely than men to lose accommodation when in custody. For those young people who do not or cannot return home, or where their home situation breaks down, they are severely disadvantaged by the lack of appropriate supported accommodation which can lead to re-offending, being placed in risky situations or further trauma-related harm. This is especially the case for young people involved in offending who are leaving secure care or custody^[427].

Public recognition of desistance

Publicly recognizing desistance may help reduce reoffending, but this proposal has not been empirically tested. Some studies have found that public recognition of offenders' progress towards desistance can help them develop a new, non-criminal identity and lead to improved self-esteem^[428]. This discovering of a new self is closely associated to sustained abstinence from offending^[429]. As a consequence, researchers have recommended that the criminal justice system should find ways to formally mark and reward desistance markers such as for example the successful completion of a prison or community sentence^[430]. Calverley and Farrall^[431] report examples of offenders who felt particularly good about themselves when invited by local drug agencies to give a talk about their experiences of coming off drugs. Such opportunities provide ex-offenders with a sense of reward and achievement and remind them of the benefits of staying away from crime^[432]. Other ways to reward desistance might include sealing of criminal justice records earlier in the offenders' criminal career than usual, restoration of civil rights, awarding certificates or pardons and using a system of graduated rewards and sanctions to reward compliance and support motivation as implemented in the context of problem-solving courts^[433]. However, this is an area in which it is difficult to provide 'evidence' in the same form as some other types of interventions (such as randomized controlled trials) and so the support for this policy is theoretical rather than empirical^[434].

Reparation and restoration

This section explores evidence on the impact of unpaid work and restorative justice on reoffending.

Reparation through unpaid work

The effectiveness of unpaid work in reducing reoffending has not been widely investigated but some qualitative evidence suggests that generative activities involving contact with the beneficiaries are more likely to be effective than menial tasks. In the time available, we were not able to find any studies that have measured the effect of unpaid work in reducing reoffending using a robust control group design. A recent study in England and Wales found that 25% of offenders subject to a stand-alone unpaid work requirement (community payback) were reconvicted^[435]; however it is possible that these lower reconviction rates reflect a lower risk of recidivism among offenders sentenced to unpaid work rather than a genuine positive effect. An earlier evaluation of seven "pathfinder" community service projects in England and Wales also produced promising findings.^[436] The study analysed staff and offender views, accessed via interviews and questionnaires, as well as administrative data and repeated measures of attitudes and self-reported problems, assessed via the Crime Pics II questionnaire. At the point of evaluation, a total of 1,250 offenders had been allocated to these projects. The evaluation found that those who completed their community service showed highly significant reductions in pro-criminal attitudes and self-perceived problems. Staff reported that two-thirds of project participants were seen as having undergone positive change and having good future prospects. Of those offenders who completed the questionnaire prior to completing their community service, 76% thought that community service had made them less likely to offend in the future. As the authors acknowledge, it is difficult to determine to what extent the relatively low-risk profile of offenders allocated to these projects influenced these positive outcomes in statistical terms. For this reason, they also explore the processes by which benefits might arise from community service; Analysis of associations between questionnaire responses suggests that community service delivers greatest impact when offenders perceive the work to be of value to themselves or others.

In support of this finding, in Scotland, qualitative evidence from the evaluation of the Community Reparation Order scheme pilot^[437] showed that placements that provided opportunities for direct contact with the beneficiaries and led to the acquisition of new skills were more valued by offenders compared to placements involving menial tasks with no obvious benefit to others^[438]. Offenders also noted the positive effect that praise of their work had and those that were in more regular contact with a supervisor reported more positive experiences. It has been reported that unpaid work of a generative nature can trigger the motivation to change as it provides offenders with the opportunity to enjoy reciprocal relationships, gain trust and appreciation of other people and give something back to the community^[439]. There is some evidence that "making amends" can help offenders develop a prosocial identity that is conducive to change^[440]. In contrast, some types of unpaid work programmes can be perceived as stigmatizing^[441].

Further evidence into the impacts of unpaid work will be provided by an ongoing process evaluation of a number of interventions, including unpaid work, that were introduced in Scotland by the Community Payback Order in February 2011. This evaluation will provide qualitative feedback from practitioners and offenders and should offer insights into the perceived value of unpaid work in the context of Community Payback Orders in Scotland. However, the study design is not intended to assess the impact of unpaid work on reoffending rates.

With regard to work in prison, there is some, less robust, evidence from the US that work in prison is associated with higher employment rates upon release though this effect could be attributed to factors that caused offenders to apply for work in prison rather than the experience itself^[442]. As reported in previous sections, work in prison is more likely to be of benefit to offenders if it is linked to real prospects of employment outside of prison^[443].

Restorative justice

There is mixed, though mostly positive, evidence on the effectiveness of restorative justice in reducing reoffending. Although approaches may differ, restorative justice has been broadly defined as "a process whereby parties with a stake in a specific offence resolve collectively how to deal with the aftermath of the offence and its implications for the future"^[444]. Restorative justice practices, in Scotland and internationally, most commonly consist of face-to-face conferencing between victims, offenders and their family or supporters. However, mediation between victim and offender (without supporters) may also be used, and can be face-to-face or conducted indirectly, such as via letter. Conferences generally allow all affected to talk about the circumstances and impact of the offence and work towards an apology from the offender to the victim(s) and a shared agreement about what ought to happen next. Conferences or mediation may take place prior to sentencing or afterwards.

A review of international interventions provides some promising evidence that restorative justice processes can reduce reoffending for some (but not all) offenders.^[445] Analysing only studies where some kind of control or comparison group was used, the authors find statistically significant reductions in reoffending where face-to-face approaches were used with four groups:

• Violent offenders under 30 in Canberra;
• Violent female offenders under 18 in Northumbria;
• Male property offenders under 18 in Northumbria;
• Property and violent offenders aged 7-14 in Indianopolis.

A later meta-analysis of restorative justice in Canada found that programmes had, on average, a positive impact on reoffending rates^[446]. In contrast, a 2005 evaluation of the court-referred Restorative Justice Pilot in New Zealand found no statistically significant effect of restorative conferencing on reoffending rates, although 92% of the victims reported satisfaction with the process^[447]. Similarly, a recent systematic review of the effectiveness of restorative justice conferencing in reducing reoffending in young offenders was unable to find evidence of its effectiveness^[448], though the authors state that this finding must be interpreted with caution given the small number of studies eligible for their review. Given the increase in offending with age in adolescence (the age-crime curve), it is likely that impacts on reoffending with young offenders may vary over time: Hipple et al. (2014) in the United States, for example, find significant effects at 6 months but not by 24 months.

Looking specifically at UK interventions, a study using randomised controlled trials (RCTs) with predominantly adult offenders in England and Wales found that those who completed restorative justice conferencing had a 14 percentage point reduction in the frequency of offending^[449]. The conferencing programmes were considered to offer 'value for money' because the estimated cost savings associated with reduced reoffending were greater than the cost of running the scheme. The study found no statistically significant impact of age, gender, ethnicity or offence type on the impact of restorative interventions in terms of reoffending, though the way the conference was experienced by offenders did produce significant effects^[450].

Given the variation in these findings, criminologists have called for further research into the process through which restorative justice works to reduce reoffending.^[451] The limited evidence available highlights the importance of the offender's active involvement in the conference, an acknowledgement of harm done^[452], the development of a conversational rhythm, and the expression of emotions^[453]. Robinson and Shapland (2008) suggest that conferencing and other restorative approaches may help to reduce reoffending through the contribution of the victim and the offender supporters in supporting the decision to desist, perhaps by providing an avenue to manage feelings of shame; by building social capital which could support change; and by the conference as a whole suggesting individualised paths to overcoming practical obstacles to desistance, through the items in the outcome agreement (which may include taking part in substance abuse programmes or training, for example)^[454].

Positive effects of restorative justice interventions may be more likely to be detected if more sophisticated measures of recidivism are used, such as the frequency and severity of reoffending^[455]. However, caution is warranted in the way in which restorative justice is conceptualised in relation to reducing reoffending: because offenders must usually volunteer to participate and admit to the offence, restorative justice is most likely to attract those who wish to desist^[456].

Deterrence

Deterrence-based interventions such as "Scared Straight" do not reduce reoffending. Deterrence can be either general or specific in nature. General deterrence refers to the effects of punishment on the general public (i.e. potential offenders), whereas specific deterrence refers to the potential inhibiting effect of punishment on the individual made subject to it. As the focus of this paper is on reoffending, we only review here the evidence on specific deterrence. Studies which have evaluated deterrence-based programmes such as "Scared Straight" or boot camps have been found them to be ineffective in reducing reoffending or, in the worst of cases, can even lead to increases in offending^[457]. No studies were found in this review which presented positive impacts of deterrence-based interventions.

Conclusion

From the evidence reviewed above, it appears that criminal justice interventions can have a positive impact on reoffending. However, it is important to note that very different effects are evidenced from different programmes within broadly similar approaches, and a single programme can impact differently for different individuals. Almost all of the reviewed studies have found substantial variability in outcomes depending on a range of factors, involving the person, the intervention, the quality of implementation and the research design^[458].

One principal implication of this is that there is no single solution to the problem of reoffending and how it can be reduced. Interventions that work well in one context may work less well in others. It is therefore important to consider a number of factors before deciding on an intervention approach for a given group of offenders, including level of motivation, needs and strengths, and diversity.