What Works to Reduce Crime?: A Summary of the Evidence

PART ONE: ADDRESSING THE UNDERLYING CAUSES OF CRIME

Biological and chemical factors

There is a growing body of evidence that suggests that predisposition to offend can be a natural biological function associated with genetic, hormonal or neurological factors. A psychological theory of crime was, for example, developed in the 1960s by Hans Eysenck^[5] who suggested that individual differences in the functioning of the cortical and autonomic nervous systems (as a result of genetic inheritance) could account for an 'antisocial' personality. He argued that there were three main dimensions of personality: extraversion; neuroticism; and psychoticism and that those at the extreme of these continuum were most likely to be offenders. There are also a number of studies on the effect of hormones, such as testosterone and cortisol; neurological deficits; neurochemical imbalances (e.g. reduced levels of serotonin and higher levels of dopamine).^[6] However, the evidence suggests that the impact of biological factors diminishes as young people are exposed to environmental factors that shape behaviour. For example, genetic research that has focused on studies of identical twins (some reared together, others reared apart), found that heritability accounted for about 41% of childhood conduct disorder, but by adulthood accounts for only 28% of adult antisocial personality disorder.^[7] The effect of biological factors is therefore highly likely to be mediated by other situational or environmental conditions - many of which are explored in this paper.

Recently there has been a growing interest in the role that lead exposure has on crime rates. The negative impact of lead exposure on physical and cognitive development has been well established in the medical and neuroscientific literature. Research has linked early years lead exposure to behavioural problems such as aggressive behaviour, impulsivity and attention deficit and hyperactivity disorder (ADHD), all of which have been identified as risk factors for offending behaviour. Marcus et al^[8] conducted a meta-analysis of 19 studies examining the association between environmental lead exposure and conduct problems in children and adolescents. Conduct problems were broadly defined and included aggressive and violent behaviour and delinquent, antisocial and criminal behaviour. Despite methodological variations between the studies, the researchers found that the results were remarkably consistent. They also found that the strength of the association between lead and conduct problems was not significantly affected by controlling for potentially confounding variables such as home environment and social class.

A recent article on the American political news website Mother Jones drew attention to the body of evidence which supports the theory that patterns in childhood exposure to lead are linked to the decline and fall of crime rates in the US and elsewhere^[9]. It cites various studies which have found that the rise and fall of environmental lead levels is followed by a strikingly similar rise and fall in violent crime, but roughly 20 years later, suggesting a link between childhood exposure to lead and later violent offending. Nevin^[10] (2007), for example, found associations between patterns of preschool lead exposure with crime rates around the world, in: the USA, Canada, Britain, France, Australia, Finland, Italy, New Zealand and West Germany (as was). Replicating the pattern found by Nevin at the national level, Reyes (2007) compared US state level emissions data with crime rates. Although she found only weak evidence to support a relationship between lead and murder rates and no evidence to support a link between lead and property crime, she did find a significant association between childhood lead exposure and violent crime. Similarly, Stretesky and Lynch^[11] looked at the relationship between air lead levels and crime rates at the county level in the US. Again, they found a link between lead and crime rates, but unlike Reyes, they found that the effect applied to both property crime and violent crime. Stretesky and Lynch also found that there was an interaction effect between lead exposure and deprivation: the association between air lead levels and crime rates was strongest in counties with higher levels of resource deprivation. They speculate that this is likely to be because people in deprived areas are less likely to receive adequate screening and treatment for lead exposure and poisoning. People living in poorer areas are also more likely to live in older and un-renovated housing, and to live near industrial plants where lead is handled.

Individual-level studies also support the association between lead and crime. For example, Wright et al.^[12] carried out a prospective study examining the relationship between prenatal and postnatal blood lead concentrations and later arrests in early adulthood. Between 1979 and 1984 the researchers recruited pregnant women from four prenatal clinics into the Cincinnati Lead Study. The women recruited came from areas of Cincinnati with a high concentration of older, lead-contaminated housing. Prenatal blood lead concentrations were measured, and then childhood blood lead concentrations were measured regularly until the children were six and a half years old. The arrest records of 250 of the 376 children originally recruited were then examined for the period from when they turned 18 until the end of 2005. The researchers found that pre- and postnatal blood lead concentrations are associated with higher rates of total arrests and arrests for violent offences.

Overall, the link between early years lead exposure and cognitive and behavioural impairments which are predictors of later offending, such as impulsivity, aggressive behaviour and ADHD, is well supported by the evidence. The link is most clearly established in relation to violent crime, but there is less consensus on the relationship between lead and property crime. The magnitude of the effect of lead on crime rates is less easily established and the interaction between lead exposure and other variables, such as ethnicity, socioeconomic status, needs further research. None of the existing research goes as far as to claim that lead exposure is the main cause of violent crime, but it is argued that lead is a major explanatory factor for variation in crime rates.

Self-control and effective parenting

Lack of self-control has long been identified in the criminological literature as being a key explanation of crime. It was central, for example, in Gottfredson and Hirschi's^[13] general theory of crime. They explain that those who have low self-control have difficulty deferring gratification, lack diligence and tenacity, tend to be thrill-seeking and have difficulty taking a long-term perspective. They also argue that those who have low levels of self-control not only have a higher propensity to offend but are also likely to suffer other negative life outcomes such as substance misuse, failure to maintain relationships and hold down a steady job. Indeed, elsewhere Gottfredson and others^[14] argue that conduct problems such as low-self control are so highly correlated to delinquent behaviour that they could be considered proxies for it.

There is clear evidence from longitudinal studies that low levels of self-control in childhood are correlated with offending in later life. The Edinburgh Study of Youth Transitions and Crime (a longitudinal study of 4,300 young people who started secondary school in Edinburgh in 1998) has found a relationship between lack of impulse control and delinquency at ages 12/13^[15]. Another longer running longitudinal study (the Cambridge Study) has shown that serious disruptive behaviour in early life often leads to frequent and serious delinquency and crime that can endure well into adulthood. This study found that among the 23 'chronic' offenders in the sample who continued to offend into adulthood, most of them were rated as troublesome, restless and impulsive by teachers at primary school.^[16] Furthermore, the Dunedin Study in New Zealand which followed a cohort of 1,000 children from birth well into adulthood has shown that the impact of childhood self-control extends beyond offending to have negative consequences for educational attainment, physical health, substance dependence and personal finances and those with low levels of self-control as children were also more likely to become unplanned teenage parents.^[17] There is also sociological research that links low self-control to unemployment.^[18] Dunedin study researchers have even demonstrated that 'Differences between children in self-control predicted their adult outcomes approximately as well as low intelligence and low social class origins'.^[19] Promoting self-control in childhood therefore seems central not only to reducing offending but also to improving wider life-chances.

Gottredson and Hirschi believe that the major cause of low self-control is ineffective child-rearing. A key role of parenting, they suggest, is to teach children to recognise and consider the long-term consequences of their behaviour and to reflect on the needs and feelings of others. The role of parents in maintaining antisocial behaviour (by giving it attention and by ignoring desirable behaviour) was first established by Patterson in 1982.^[20] Since then, many US studies have noted that problem behaviour often starts early with the combination of temperamentally difficult toddlers and ineffective monitoring and discipline.^[21] The evidence suggest that good parenting, particularly in the very early years, helps children to learn how to control impulsive, oppositional, and aggressive behaviour and this learned control reduces disruptive behaviour in the long-term.

A 2010 Campbell Collaboration review of interventions aimed specifically at improving self-control in children under age 10^[22] found that self-control interventions are effective at improving self-control as well as reducing delinquency and problem behaviours. The researchers conducted a meta-analysis of 34 studies, all of which had a randomised controlled evaluation design and took post-test measures of self-control and/or delinquency and problem behaviours. The vast majority of the studies included in the review were carried out in the United States (91%), and the samples were largely male (56%), white (68%) and from high-risk/low income populations (65%). The interventions were largely group-based and took place in schools. The types of intervention included: social skills development programmes; programmes designed to develop participants' cognitive coping strategies; immediate/delayed rewards interventions; video-based/role-playing programmes; and relaxation training.

The review concluded that overall, these programmes do serve both to improve children's/adolescents' self-control and to reduce delinquency and problem behaviour. The success of these programmes indicates that self-control is not a fixed attribute, but is something which can be worked on and improved in the early years, which in turn can lead to reductions in delinquent behaviour. However, the researchers were not able to identify which type of intervention, or aspects of the interventions were particularly effective, or to look at whether the positive outcomes persisted into late adolescence/early adulthood.

There have been many reviews of the literature on the impact of interventions aimed at improving the quality of parenting as a means of improving control over impulsive, oppositional and aggressive behaviour. Farrington and Welsh's^[23] meta-analysis reviewed interventions delivered in a range of different forms (home visits; parent education plus day care; clinic-based training; school-based training; home/community-based training programmes and; those that used a combination of these). The review concluded that, based on the evidence from studies of only the highest quality methodologically, family-based prevention programmes were effective in preventing child behaviour problems, delinquency and other related problems. What is more, they also produced other beneficial outcomes such as improved life prospects. By comparison, Bernazzani and Tremblay's 2006 systematic review^[24] of training programmes with parents of children under the age of 3 (at the start of the programme) found only 7 studies that met the criteria for inclusion and reported mixed results. Four studies reported no evidence of impact, two reported beneficial impacts and one reported beneficial effects with mainly beneficial effects with some very minor harmful effects.

However, the comprehensive 2008 Campbell Collaboration review^[25] - found robust evidence of impact and drew some confident conclusions for policy. Studies were included in this review only if they had a randomized controlled design that provided before-and-after measures of child behaviour problems among experimental and control subjects. The focus was also limited only to interventions targeted at children under or around the age of 5 at the start of the intervention. Parent training was to be the main element of the intervention, though not necessarily the only element, and the intervention could be targeted at high risk groups or offer a universal provision. Parent training included, for example, training to strengthen competencies at monitoring and appropriately disciplining their children's behaviour and may also have included increasing parent's involvement in the school education of their child. The review focused on two general categories of interventions: those that centred around educating parents to improve their children's life chances, often beginning at birth and some in the later stages of pregnancy; and those that combined parent training with day-care for the child. In total, 55 studies were included, most of which were conducted in the US.

The findings from the studies were combined using meta-analysis. In the most cases the outcome measure was the parent's response to a checklist of questions about the child's behaviour. The findings suggested that, on average, the interventions were having a small to moderate impact on behavioural problems and that there were no statistically significant differences in outcomes across type of programme. However, the authors point out that it was difficult to identify which particular features of the interventions were responsible for the positive effect when they involved a number of elements.

The authors concluded that early family/parent training is an effective intervention for reducing behaviour problems among young children and that these effects appear robust across various weighting procedures, context, time period and outcome measure. Additional descriptive evidence indicated that early family/parent training was also effective in reducing delinquency and crime in later adolescence and adulthood. Overall, the findings lend support for the continued use of early family/parent training to prevent behaviour problems such as antisocial behaviour and delinquency. The reviewers conclude 'early family/parent training should continue to be used to prevent child behaviour problems such as conduct problems, antisocial behaviour, and delinquency among young persons in the first five years of life. Such programs appear to have few negative effects and some clear benefits for its subjects' . They go on to say that their findings offer further support for a number of large scale programmes implemented in Western nations to improve parenting skills.

A randomised controlled trial of a parenting intervention in the UK adds further evidence of impact and offers more insight into the focus and nature of these types of interventions.^[26] The trial was carried out in England between 1995 and 1999. Eligible children were aged 3 to 8 years and referred for aggressive and antisocial behaviour to their local NHS multidisciplinary child and adolescent mental health service. Most lived in deprived circumstances and were at high risk of later juvenile delinquency and social exclusion. The trial randomly allocated participants to intervention or control (in this case being retained in the waiting list). There was no evidence that groups differed in any important respects. The parents of six to eight children were seen as a group for two hours each week over 13-16 weeks (the children did not take part). The programme covered play, praise and rewards, limit setting, and handling misbehaviour. In each session, parents were shown videotaped scenes of parents and children together, which depict "right" and "wrong" ways of handling children. Parents discussed their own child's behaviour and were supported while they practised alternative ways of managing it. Each week tasks were set for parents to practise at home and telephone calls made to encourage progress. Mothers were interviewed on entry to the trial and after completion of the intervention (or waiting list period) five to seven months later and their accounts of their child's behaviour were used as the primary outcome measure. However, the validity of these accounts were tested by randomly selecting 20 cases to be directly observed in a structured play task. This confirmed that the mother's accounts of their child's behaviour was a reliable outcome measure.

The results revealed that children of parents allocated to parenting groups demonstrated a large reduction in antisocial behaviour, while the behaviour of those in the control group did not improve. After the intervention, parents were giving their children far more praise to encourage desirable behaviour and more effective commands to obtain compliance. The study concluded that these type of parenting programmes show promise as a cost effective way to reduce the personal and economic burden of antisocial behaviour.

Further examples of parenting programmes that have been implemented and evaluated in the UK are provided in MacQueen et al's review.^[27] They cite a review of the evidence on parent training that concludes that effective responses are likely to be staged and varied across the life course with: parent training and telephone support for parents of under 8s; parent training with support for social skills and problem-solving for parents of 8-12 year olds; and family therapy for parents of adolescents. Commentators also argue that, because children exhibiting early-life behaviour problems become increasingly resistant to change over the life course, it is important to begin intervention as early in the life course as possible.^[28]

The role of the school and wider community in setting behavioural boundaries

Longitudinal research on offending has identified not only that parental attitudes and parenting style are associated with offending but also that features of the wider social community are important. Lack of discipline and disorganisation in school; community disorganisation and physical deterioration/neglect and; attitudes among peers that condone problem behaviour have also been shown to be associated with offending.^[29] The theory that the community is a source of informal social control largely developed out of work by Shaw and McKay who explored the relationship between delinquency and urban neighbourhoods in Chicago from the 1900s to 1950s. Their work suggested that male juvenile delinquency was a result of the disruption of community social order which is maintained by institutions such as the family, church and school as well as community centres and organised clubs.

There is some evidence that attachment to religious institutions can help reduce the risk of offending. In a review of literature for the Youth Justice Board for England and Wales^[30], researchers from the University of Oxford reported that a number of studies, mainly from the USA, suggest that religion or religiosity may act as a protective factor. Religion can, they explain, act as a regulatory mechanism through the promotion of pro-social moral values and normative beliefs.

In terms of policy intervention, promoting the role of the school in setting appropriate boundaries is perhaps the most practical route to reducing delinquency and crime. When we consider the proportion of time young people spend in school, especially during a period in their lives in which they are most disposed to offending (the mid to late teenage years), the role of the school in shaping attitudes and behaviours seems obvious. In their report on the role that education can play in the prevention of offending, Holden and Lloyd^[31] note that schools can provide social conditioning in a variety of levels, all of which play a core protective role. Not only does school have an obvious role in providing qualifications that will enhance employment prospects but it also enhances 'soft skills', such as the ability to communicate with others effectively, and it helps establish and reinforce behavioural norms. This will include strict adherence to attendance and punctuality, behavioural standards and respect for others.

Gottfredson et al^[32] review a range of school-based interventions designed to prevent many of the precursors of delinquent behaviour (including school-drop out and truancy as well as problem-behaviour). They examined a total of 178 studies that measured the impact of these interventions on either criminal behaviour, substance misuse or other forms of antisocial behaviour. Overall the authors conclude that several strategies are effective in reducing problem behaviours and that among the most effective strategies are those designed to set and enforce clear boundaries on behaviour. These involved building school capacity to manage discipline by establishing, clarifying and consistently enforcing school rules or policies and establishing norms or expectations for behaviour by using, for example, campaigns, ceremonies or similar techniques.

Another way in which the community can provide a source of informal social control is by appointing mentors as role models. Mentoring programmes typically involve non-professional volunteers spending time with young people and acting as role models in a supportive, non-judgemental manner. Welsh and Hoshi^[33] reviewed six mentoring programmes, all of which were implemented in the USA and measured outcomes related to offending. Treatment lasted from 12 weeks to 4 years and the length of follow-up was from immediate to 6 months. The methodological robustness of the evaluation was high, with 4 using randomised controlled experimental designs. While only two of the evaluations measured direct crime outcomes, the other programmes were included in the review because they measured disruptive and aggressive behaviour and antisocial acts - research has revealed that there is considerable continuity between these behaviours and offending.

The findings revealed that the effects on crime were mixed - while one study found desirable effects on crime overall (this was the programme that lasted 4 years), the other evaluation found desirable effects on crime for those with prior offences but undesirable effects for those who had no prior criminal activity.^[34] Among the studies that measured impact on disruptive and aggressive behaviour: one study revealed that the intervention had no impact; one that there was a positive impact on substance use, antisocial behaviour and academic performance 6 months after the intervention ended; one that there was an improvement in parent-rated behaviour but not teacher-rated behaviour; and one that there was a reduction in truancy and behavioural problems (but the small sample size and short follow-up period limited the strength of the findings from this evaluation). The authors therefore concluded that no clear assessment could be made of the crime prevention effectiveness of community-based mentoring.

Empathy and compassion

The ability to see things from the other's point of view and having the emotional capacity to feel for the other person is also referred to as 'emotional intelligence'. For research on the relationship between emotional intelligence and offending see Aleixo and Norris.^[35]

Losel et al^[36] discuss the effectiveness of interventions to improve emotional intelligence and explain how one approach to reducing the risk of offending among young people is to provide social skills training. This training is frequently based on cognitive- behavioural concepts of social learning and problem solving and tends to include sessions designed to teach social perception, identification of emotions, perspective taking and empathy, evaluating consequences, self-control and interpersonal problem-solving. This can be focused either on those identified as being at risk of offending or on general populations.

While various reviews have suggested that social skills training is a promising approach to the prevention of antisocial behaviour and crime, Losel et al conducted their own review of these type of interventions. They focused only on studies that: examined programmes designed to reduce antisocial behaviour in children and young people (up to the age of 18) and had a randomised design. Their review included published and unpublished studies written in English and German. It did not include evaluations of interventions targeted on young people who were already known offenders - the focus was instead on preventative interventions (although it examined some universal interventions most were focused in young people who were already exhibiting antisocial behaviour). In total 55 evaluations met the criteria, which together yielded 89 treatment-control group comparisons involving over 9,000 subjects. Most of these were published in the 80's and 90's in the US, two-thirds measured impact only up to 2 months after intervention. Most of the programmes that were reviewed were relatively short (with almost half lasting no more that 2 months). The typical format was group training carried out in a school setting.

In nearly half of the studies, impact was measured on the basis of aggressive behaviour and in most cases was assessed by teacher or behavioural expert. Nearly half of the studies revealed a positive impact, 40% showed no impact and 10% suggested there had been a negative impact. Results were combined in a meta-analysis and the overall the effect^[37] was rated as being positive and significant. Although the overall effect size was rated as 'small', the authors argue that, given the short duration of the interventions and the typical group-based setting, even this small impact meant that these interventions were likely to be cost-effective.

When the authors compared the impact of different types of interventions, they found that cognitive-behavioural programmes were most successful and were the only type to have sustained a significant impact three months or more after the intervention ended. (Raynor^[38] also shows how programmes that have a cognitive-behavioural basis have been shown to be particularly effective in reducing reoffending.) Losel et al also found that programmes targeted on young people who had already developed some behavioural problems had the largest effect and those that were applied universally had non-significant effects. The authors suggest that this is because few of those subject to universal programmes would have gone on to exhibit serious problematic behaviour (and hence significant differences between the control and treatment groups were never likely to have been generated). They therefore suggest that targeted provision is likely to be most cost-effective.

The effectiveness of cognitive-behavioural techniques in reducing the risk of offending was also demonstrated in a review of school-based interventions by Gottfredson et al.^[39] These interventions were designed to prevent many of the precursors of delinquent behaviour including school-drop out, truancy and antisocial/aggressive behaviour. Of the nine types of intervention reviewed in this study, those that focused on cognitive-behavioural training were among the most rigorously evaluated and provided effect sizes that were among the highest of those reviewed. They demonstrated clear positive effects on school attendance and persistence and aggressive/antisocial behaviour and some studies also demonstrated reductions in criminal behaviour.

The impact of trauma

There is evidence that offending is related to the experience of abuse and neglect. A Scottish Children's Reporter Administration (SCRA) study of children and young people referred to the Children's Reporter for offending in Glasgow^[40] found that among the 15 young people who had 20 or more offence referrals: one third had been initially referred in relation to physical abuse by parents or carers and in eight cases background reports disclosed the existence of domestic abuse. Parental alcohol or drug misuse was also apparent in the majority of cases. Batchelor's 2005 study^[41] of 21 young women imprisoned for violent offending also showed that two-fifths of them had been sexually abused. The same proportion described witnessing regular incidents of serious physical violence between their parents (most of which were attributed to parental alcohol misuse).

There is also some evidence from the Edinburgh Study of Youth Transitions and Crime on the relationship between self-reported involvement in violence/offending at age 15 and a range of vulnerabilities. Regression analysis revealed that violent offenders were (among other things) significantly more likely than non-violent youths to be engaged in self-harming and to have been a victim of crime by the age of 15. For girls, another significant variable was having experienced family crisis in the last year.^[42]

While this underlines the importance of strong social relationships it also underlines the importance of a child protection system that identifies and addresses abuse and neglect as early as possible. However, even with the most effective child protection system, we have to accept that there are some forms of trauma that we cannot predict or avoid. Bereavement of a close relative or friend is an obvious example. Research on resilience to adversity by Mastern et al^[43] provides some important clues to how we might better equip young people to deal with these types of trauma. According to this research, IQ, strong supportive family relationships and good relationships with someone outside the family are commonly cited as the most important protective factors. While IQ may not be amenable to change, the quality of care that a child receives is frequently cited as the most important protective factor. A child's belief in their own agency and self-efficacy, their belief in their own ability to do something about the situation, rather than perceiving themselves as a passive victim is also frequently cited as an important protective factor. This is linked to optimism and the ability to re-frame a situation so that the positives as well as the negatives can be seen. If children are able to master a difficult situation this will in turn have a positive effect on their self-esteem. A similar point is made in MacQueen et al^[44] who explain that having the opportunity to take part and succeed in challenging tasks, can help promote self-esteem and self-efficacy, which in turn can help build resilience.

Strong social relationships

Among the many factors that have been shown, in longitudinal studies, to be correlated with offending are: alienation, lack of social commitment and lack of neighbourhood/community attachment^[45]. In the longest longitudinal study in criminology, Sampson and Laub's reanalysed an archive of data from an already influential prospective study of juvenile and adult criminal behaviour. The dataset contained a sample of 500 male juvenile delinquents and a comparison sample of 500 non-delinquent juveniles that had been matched, case by case, on the basis of race/ethnicity, IQ and household income. Extensive data had been collected on the samples at ages 14, 25 and 32 during the 1940s, 50s and 60s. As well as reanalysing the existing data set, Sampson and Laub used criminal records to track the sample and interviewed 52 men selected to represent different patterns of criminal behaviour. With a 50 year window on the lives of these men, Sampson and Laub's key argument is that crime is more likely to occur when an individual's bond to society is weak. They found that the strongest and most consistent effects on both official and self-reported delinquency in childhood and adolescence can be explained by lack of informal family, school and peer social controls. Three family factors stood out as strongly predictive of delinquency: low levels of parental supervision; the combination of erratic, threatening and harsh discipline; and weak parental attachment. In addition, school attachment had large negative associations with delinquency independent of family process. Peer influences were weaker. In adulthood, the stronger the ties to work and family (particularly marital attachments), the less an individual was likely to be involved in crime and deviance. This held true for those who had displayed previous delinquency as well as those who had not. Sampson and Laub suggest that involvement in institutions such as marriage and work introduce important incentives to conformity - they represented investments that the men did not want to risk losing.

What is particularly striking about Sampson and Laub's findings is that structural background factors such as family poverty had little direct effect on delinquency but were instead mediated by informal social controls. Another important finding, for our purposes, is that while there was a strong relationship between early antisocial tendencies (such as violent temperament) and delinquency and offending in later life, the most powerful explanation of delinquency in adolescence is informal social control. This means that even those predisposed to offending in their early years can avoid or desist from offending given the right social circumstances. While there is no denying the central importance of parenting in the early years, there are nonetheless, what Sampson and Laub call, 'turning points' in adulthood - experiences and circumstances that can redirect criminal trajectories in either a positive or negative direction.

Consistent with findings from the Edinburgh Study of Youth Transitions and Crime, Sampson and Laub argue that trajectories of desistence cannot therefore be predicted using information collected during childhood - "While childhood prognoses are reasonably accurate in terms of predicting levels of crime between individuals up to their 20s, they do not yield distinct groupings that are valid prospectively over the life course"^[46]. What happens during the adult life course also matters and those who were interviewed described the major turning points in their adult lives as including: marriage/partners; the military; reform school; work; and neighbourhood change. What appeared to be important about these changes was that they involved new situations that: helped draw a line under the past and provided the opportunity for identity transformation; provided supervision and monitoring as well as new opportunities for social support and growth; and provided structured routine activities.

While Sampson and Laub do not believe that early life experiences set in place a trajectory from which individual cannot deviate they do, nonetheless, recognise that delinquency in early life is very likely to undermine the social bonds that would enhance the ability to avoid offending. They refer to early delinquency as setting the scene for 'cumulative disadvantage'. Imprisonment as a juvenile will, for example, have a negative impact of future job prospects.

Sampson and Laub's argument that school attachment can serve as an important 'turning point' in the lives of young people resonates with findings from the Edinburgh Study of Youth Transitions and Crime. Analysis of the trajectories of criminal convictions among the young people in the Edinburgh Study of Youth Transitions and Crime reveals that while offending tends to decline after the age of 20, during the teenage years there were 3 distinct groups of offenders: a 'desister group' who also began attracting convictions around the age of 9/10 but whose probability of conviction stabilised between age 14-16 before declining and stopping completely by about age 20; a 'later onset' group who attracted their first conviction around the age of 15/16 and peaked at around age 20 and; a 'chronic group' who began attracting convictions around the age of 9/10 and whose probability of conviction rose steeply in the early to mid teenage years and remained high until their early 20s. (At the peak age of conviction, the probability of conviction for this group was also significantly higher than for the other groups.) The conviction trajectories for these groups is shown in Figure 5 below.

FIGURE 5: Trajectories of criminal convictions among the ESYTC cohort

Comparing the characteristics of the first two groups who began attracting convictions at the same age but who went on to have very different patterns of convictions thereafter, the only significant differences between these groups at age 12 appeared to be that the chronic group were more likely to be male and more likely to live in the 25% most deprived areas in Edinburgh. This suggests then that it is very difficult to predict, by age 12, which males from these areas (or indeed other areas) will become persistent offenders. While it was not possible to predict, at age 12, which young people would become desisters and which would become chronic offenders, important differences begin to emerge between the ages of 13 and 15 - the chronic group show a significant deterioration in key aspects of schooling and agency contact which is not mirrored in the desister group. The chronic group experienced a statistically significant increase in truancy and a large and statistically significant increase in school exclusion. Over this same period, the chronic group also reported significantly increased rates of adversarial police contact. Importantly these differences between the chronic and desister groups were not accounted for by differences in self-reported serious offending. Findings from the ESYTC therefore suggest that reduced commitment to and increased exclusion from school, as well as increased adversarial contact with the police, might have a part to play in driving up convictions in the mid to late teenage years. While it is possible that police contact and convictions may have increased the risk of school exclusion, the data tend to show that school exclusion preceded the increased police contact and convictions. The role of the school in bonding young people to society and to its prevailing moral code therefore seems to be have been demonstrated in the Scottish research literature.

Problematic drug misuse

In its most obvious form, drug use contributes to crime through the offences of the supply and possession of drugs. In 2012-13, there were 34,688 drug-related offences recorded by Scottish police forces^[47]. The majority of these (29,150) were for possession. Drug use also has a role to play in other forms of offending. In 2011/12, 21% of individuals seeking treatment or advice from drug services in Scotland claimed that they had committed crime to fund their drug use^[48]. We also know that 39% of respondents to the 2013 Prisoner Survey reported being under the influence of drugs at the time of their offence, and 35% stated their drug use was a problem for them on the outside^[49]. One-sixth of prisoners (16%) reported that they committed their offence to get money for drugs. Of 83 persons accused in homicide cases in Scotland in 2011-12, 5% (4) were known to be on drugs and 2% (2) were both drunk and on drugs^[50]. In the same year, 16% (10) of the 61 homicide victims in solved case s were reported to have been killed in a drug related homicide, i.e. a homicide motivated by the need to obtain drugs (or money for drugs), homicide of a supplier or consumer of drugs in order to steal proceeds of drug trade or homicide as a consequence of rivalry within the trade/between users or dealers. Finally, analysis of the Home Office Youth Lifestyles Survey shows that using drugs is a key predictor of serious or persistent offending in young people aged 12 to 17^[51].

However, the relationship between drug use and offending is complex and the notion that drug use causes offending may not be wholly accurate. A Home Office study^[52] that explored the risk factors associated with drug taking in young people revealed that these are very similar to those associated with offending. These include: involvement in serious anti-social behaviour; weak parental attitudes towards bad behaviour; being in trouble at school (including truanting and exclusion); having friends who get into trouble; being impulsive; and lacking empathy. It is possible then that offending and drug use simply have the same antecedents and are symptoms of the same set of conditions. In their 'general theory of crime', Gottfredson & Hirshi^[53] argue that drug misuse is not the cause of crime but is simply a symptom of low levels of self-control. They state - "In our view, the relationship between drug use and delinquency are both manifestations of an underlying tendency to pursue short-term, immediate pleasure"^[54].

The authors of the Home Office study suggested that, to promote resilience to drug use, policy intervention should focus (among other things) on early intervention with truants and on strengthening parenting skills. There is therefore a parallel here with interventions that have already been highlighted to as effective for reducing the risk of crime.

A thorough review of the evidence on the effectiveness of different methods of treating drug addiction was published by the Scottish Government in August 2010 to help inform Scotland's revised drug strategy (which marks a shift to a 'recovery' model of intervention). This review^[55] included an assessment of the findings from the international literature about what we know about 'recovery' in the addictions field generally and an evidence-based appraisal of 'what works' in this area as well as a more focused overview of the international literature on drug treatment effectiveness.

The review found that there is a considerable history of cohort studies assessing the short, medium and in some cases the long-term outcomes of a range of mechanisms of delivery of drug treatment which consistently show significant improvements across a range of indicators, including health, offending, risk-taking, substance use and social functioning. Differences in effectiveness between types of treatment (such as community detoxification, methadone maintenance and residential rehabilitation) have been less consistently reported in the evidence base but do suggest the following:

• There is a consistent evidence base supporting opiate replacement therapy in maintenance settings, but this requires not only prescribing but adequate psychosocial support and links to 'wraparound' care.
• Scottish outcome research has shown that while methadone maintenance leads to improved outcomes in a range of domains, it is associated with low rates of sustained abstinence.
• Continuity of care is a critical component of effective treatment systems, and there is a strong supportive evidence base around linkage to 12-step and other community 'aftercare' supports.
• There is an ongoing issue with psychosocial interventions - while there is a strong evidence base from trials, there is little evidence that these are routinely translated into everyday clinical practice.

The review also found that, while structured treatment has a key role to play, it is only part of the support that most people will need. Ongoing support in the community is essential for the recovery journey. The best predictor of the likelihood of sustained recovery from addiction generally is the extent of 'recovery capital' or the personal and psychological resources a person has, the social supports that are available to them and the basic foundations of life quality, i.e. a safe place to live, meaningful activities and a role in their community (however this is defined). Barriers to recovery include psychological problems (mental illnesses and the absence of strengths, such as self-esteem and self-efficacy), significant physical morbidities (including blood borne viruses), social isolation and ongoing chaotic substance use. The review therefore highlighted the role of communities as both a setting for recovery to occur and a foundation for supportive relationships and opportunities for vocational and personal growth, as well as a developmental platform for recovery.

Problematic alcohol misuse

While a direct causal relationship between alcohol and offending (including anti social behaviour) is often difficult to establish, the association is clear. Even a limited amount of alcohol impairs cognitive skills, meaning that people may misread social cues, make bad judgements about risk, or respond inappropriately in social situations. In particular, they are more likely to respond aggressively when they believe they are being provoked. According to the World Health Organisation, alcohol and violence are linked in a number of ways, with a strong association between alcohol consumption and an individual's risk of becoming a perpetrator or victim of violence^[56] . The following key facts further demonstrate the link between alcohol and offending:

• According to the 2013 Prison Survey, 45% of Scottish prisoners say they were drunk at the time of the offence, including two-thirds (68%) of young offenders^[58]
• An audit of Scottish Emergency Departments suggested that at least 70% of assaults presenting to these departments may be alcohol-related^[59]. This equates to a minimum of 77 alcohol-related assaults presenting to Scottish emergency departments every day.
• Analysis by Strathclyde Police shows that of the 5,000 prisoners processed by one Glasgow police station in 2006-7, over 60% were under the influence of alcohol and/or drugs. Of those detained for violence, two-thirds were under the influence of alcohol^[60].
• Over two-thirds (71%) of those accused of homicide (and whose drug status was known) in 2012-13 were either drunk or drunk and under the influence of drugs at the time of the alleged offence^[61].
• In 59% of violent crime in 2012/13, victims said the offender was under the influence of alcohol^[62]

There is also evidence to suggest that the role of alcohol in offending is increasing. Repeated surveys of new admissions to Glenochil and Polmont young offenders institutions have shown that the proportions of respondents who stated that alcohol had contributed to their offending rose from 58% in 1996 to 80% in 2007. In contrast, the proportion that blamed illegal drugs fell from 40% in 1996 to 30% in 2007.^[63]

However, as with drugs, the link between alcohol and crime is complex. Research commissioned by the Scottish Government on troublesome youth groups, gangs and knife carrying in Scotland reveals that young people involved in violence believe that drinking alcohol is a precipitator of violence. However, the agencies involved in tackling violence believed that while fighting was associated with alcohol consumption and / or drug taking, these were not seen as causal factors.^[64] Again, it may be that the association exists only among those already predisposed to violence.

NHS Health Scotland developed a set of logic models that outlined the mostly likely pathways to achieve low levels of excessive and irresponsible drinking, and the potential long term benefits to be accrued from a reduction in consumption^[65]. Based on known associations between consumption and crime, a reduction in violence, offences and anti-social behaviour is one of the key predicted long-term outcomes in the model. These models helped to inform the Scottish Government's alcohol strategy 'Changing Scotland's relationship with alcohol: A Framework for Action'.^[66], which explicitly recognises the potential criminal justice benefits from reducing consumption.

A detailed evaluation programme - Monitoring and Evaluating Scotland's Alcohol Strategy (MESAS)^[67] - is in place to consider the extent to which the actions set out in the alcohol strategy met intended outcomes. MESAS is led by NHS Health Scotland and includes both commissioned and in-house research and analysis. The aim of the evaluation is to provide more than just a final verdict on the effectiveness of the strategy but to track the implementation progress, reach and outcomes of the key actions In addition, the Alcohol and Offenders Criminal Justice Research Programme led by NHS Health Scotland is of particular relevance. There were 3 strands to this programme: a needs assessment of alcohol problems in prisoners and recommendations for service improvement; a scoping study looking at interventions in place in local areas for offenders with alcohol problems in community justice settings and; an evaluation of the feasibility and effectiveness of the delivery of alcohol brief interventions in the community justice setting.^[68]

Employment opportunities

A review of the evidence^[69] (largely from studies undertaken in the US) reveals that those who commit crime tend to be those who are in low paying jobs or not in work. It also shows that the communities in which crime, particularly violent crime, is heavily concentrated tend to be those with high unemployment rates. However, analysis of the relationship between crime and employment rates suggests that the relationship is not straightforward. A review of the evidence by Justice Analytical Services^[70] found stronger evidence of a link between crime and inequality than it did between crime and unemployment. Overall, a rise in inequality often corresponds to a rise in crime. One theory that explains the link is that higher inequality implies a larger differential between the expected payoffs from legal and illegal activities. Another (and one that helps to explain the link with violent crime) is that those living in areas of high inequality may have lower expectations about their chances of improving their social or economic status, and thus have a lower perceived opportunity cost of engaging in criminal activity.

Other commentators^[71] argue that the relationship between employment and offending is spurious and that the mediating variable is individual social control, which includes commitment and responsibility. One of the most important UK studies of criminal behaviours is David Farrington's longitudinal study of males who experienced their teenage years in 1960's London. The study tracked the cohort (of over 400 subjects) well into adulthood and found that although property crime was committed more frequently during periods of unemployment, this held true only for those who were already predisposed to crime (as measured by earlier self-reported offending and moral values). In other words, unemployment did not lead to crime among those who were not already predisposed to it^[72]. Sampson and Laub^[73] however believe that lack of employment can also explain the late onset of criminal behaviour among those whose early life experiences had previous disposed them to be law-abiding. They suggest that employment, particularly stable employment, is a key factor in promoting desistence from crime and also that lack of it can trigger late onset offending. Securing steady, satisfying employment may act as a positive turning point for young adults previously engaged in delinquency because employment provides access to social networks and a conventional lifestyle^[74], provide a structured daily routine, a sense of stability and life satisfaction, and social control that may be otherwise lacking in young adults' lives.

While it does seem clear that among those predisposed to crime, the risk of their offending increases during periods of unemployment, there is some debate about whether it is an underlying cause of offending (as opposed to a contributory factor). That said, Sampson and Laub's evidence about 'turning points' highlights the importance of stable employment as an incentive to conform to the prevailing social norms.

Two reviews of the effectiveness of interventions designed to improve the employment prospects of those at risk of offending were identified (though most of the interventions that have been reviewed have focused on convicted offenders). One of the difficulties of assessing the impact of these type of employment interventions is that the proportion of participants who drop out of the intervention can be very high. (In one evaluation, for example, only 18% of those assigned to the programme actually completed it)^[75]. Lower rates of offending among those who complete employment programmes may therefore be more to do with the characteristics of those to persevere than with the programme itself. Even so, findings from the reviews suggest, on balance, that there is no robust evidence of the impact of employment interventions on offending.

One of the reviews was a Campbell Collaboration review by Visher, Winterfield and Coogeshall^[76]. This identified only eight studies (all from the United States) that reviewed non-custodial employment programmes for convicted offenders and the majority of these were more than 10 years old. From these they concluded that "none of the interventions had a significant effect on the likelihood that participants would be rearrested". They go on to suggest that employment programmes may only be effective for motivated individuals and that standard employment programmes are unlikely to change motivation. Ex-offenders, they argue, typically have other needs that may preclude immediate employment, including limited education and learning difficulties, substance misuse, mental illness, and a lack of affordable or stable housing.

Similarly, Bushway and Reuter in their assessment of employment interventions with ex-offenders in the United States conclude that no programme (in-prison training, or transitional assistance) or pre-trial diversion has consistently shown itself capable of decreasing recidivism through employment focussed intervention^[77].

Bushway and Reuter also go on to consider the impact of employment interventions that focus on 'at-risk' youth - defined as high school dropouts, those from poor households or poor communities. The programmes took three main forms: (a) subsidised employment, (b) short-term training programmes and (c) intensive residential programmes providing vocational and life skills training. However, few of the evaluations measured change in criminal behaviour (crime control was a regarded as a secondary outcome of the programmes). While they found that subsidised work interventions for at-risk youths did increase employment prospects during the period of the intervention, they concluded that they did not increase long term employment prospects nor did they appear to be supportive of non-criminal behaviour. Short term training programmes were also found to have no long term impact on employment outcomes.

The most rigorously evaluated programme of residential training that Bushway and Reuter found was the federally funded Job Corps which enrolled 60,000 youth at an estimated cost of $1.3 billion. Over 80% of those enrolled on the programme were high school drop-outs. Two major evaluations were carried out, one in 1982 and another in 2000. The 2000 evaluation was the more robust evaluation and examined outcomes 20 months after the intervention. It found that, compared to the control group, the Job Corps participants were only 3% more likely to be working and that their weekly wages were 8% higher^[78] . Nevertheless, even with these smaller differences in employment outcomes, the evaluation found that Job Corps participants were 17% less likely to be arrested than the control group. They also highlight that the residential component was a key feature in that it provided those who were drawn from 'debilitating environments' with a structure that could facilitate the necessary commitment. It should also be noted that the primary focus of this intervention was on attaining academic and vocational credentials and so the reduction in offending could have been as much a result of educational gains as employment gains. This therefore seems consistent with Visher et al.'s conclusion that other needs have to be met before employment programmes can be effective - this includes addressing motivational and educational issues. Based on their review, Bushway and Reuter state that programmes aimed at moving people out of crime and into employment must provide social and educational supports (as well as reduce the attraction of crime and treat substance misuse) if they are to be effective.

Education

Education equips people with the skills and qualifications that improve access to legitimate opportunities to satisfy needs. Longitudinal research on offending has identified that offending is correlated with (among other things) low intelligence and low school attainment (beginning at primary school)^[79].

There are many ways in which schools can seek to improve educational attainment. This is not the place to review these general approaches. However Gottfredson et al's^[80] review of school-based interventions focused only on those designed to prevent the precursors of delinquent behaviour. The most common approach examined in the review focused on changing classroom instructional and management practices to enhance classroom climate or improve educational processes. Although the review did not describe these approaches in detail, the authors noted that this broad strategy had already been shown to have positive impacts on academic attainment. In terms of their impact on crime, substance misuse and school engagement, they conclude the evidence suggests that they are 'promising' interventions. However the results are more mixed where these strategies are not combined with some focus on self-control or social competency skills. This therefore suggests that, while there is a relationship between offending and educational attainment, this relationship could be mediated by other underlying factors such as low self-control.

One relevant approach (that combines some focus on social skills) is Glasgow City Council's use of nurture groups. Nurture groups are targeted on young children who do not have the necessary emotional maturity to cope with the exacting demands of school life. Typically, a nurture group is a room in the mainstream school that contains different areas reflecting home and school. Within this comfortable and secure environment the adults respond to the children's developmental needs and help equip them with the basic skills they need to function in the normal classroom.

The evidence on nurture groups is discussed in a Scottish Executive report that explores the links between mental health and behaviour in schools^[81]. The authors of this report explain that a body of research evidence is developing about the effectiveness of nurture groups, most of which has come from small-scale case studies. However, a larger project undertaken by Cooper, Arnold and Boyd^[82] showed that teachers were almost unanimously positive, both with respect to children's educational progress within the nurture group and in terms of the impact of the nurture group on the life of the school as a whole (by, for example, encouraging the development of more nurturing attitudes and practices throughout the school). The majority of parents felt that behaviour, educational progress and enjoyment of school had all increased. The authors conclude:

The evidence from this study suggests that such a holistic approach has the potential to produce positive outcomes across a wide range of variables, including social, emotional and behavioural and educational functioning of children; parents' attitudes towards their own children and the school; and the positive functioning of whole schools.

O'Connor and Colwell^[83] assessed children entering and leaving a nurture group. The study showed that while the short term effects are quite clear, with children's emotional and behavioural difficulties reduced to a level enabling them to participate in the mainstream classroom, over the longer term some still required a degree of nurturing in the normal classroom. Finally, a large-scale, controlled study of nurture groups across 32 schools in the City of Glasgow also reports quantitative gains in academic achievement when the progress of pupils in nurture groups is compared with that of matched children in mainstream classes^[84].

CONCLUSIONS

A recurring theme in this section has been the importance of parenting. It has been identified as having a key role in the development of self-control (low levels of which have long been identified as a factor in offending); reducing the risk of and building resilience to trauma (which again has been linked to offending); building strong social bonds and setting appropriate boundaries on behaviour. The fact that low self-control has been linked to a range of negative life outcomes underlines the importance of addressing this if we are to improve wider life chances. Furthermore there is some robust evidence that investment in parenting programmes is effective in promoting self-control and that we should continue to invest in these programmes.

Given the link between serious disruptive behaviour in early life and delinquency and offending in adulthood, and the evidence that parenting programmes can positively impact on this behaviour, support for parents in addressing behavioural problems in the early years of a child's life appears to be one of the most effective pathways to reducing crime. Furthermore, the evidence that serious disruptive behaviour in early life also has negative consequences for education, employment and relationships suggests the benefits would be felt across a range of national outcomes. However, there is some evidence from the Edinburgh Study of Youth Transitions and Crime^[85] that the effect of good parenting was weakened where families were living in more deprived and disorganized neighbourhoods. Tackling parenting in isolation will not therefore be the 'magic bullet' and other pathways to reducing crime will also be important.

As well as highlighting the role of parenting, the evidence base has also highlighted the important role that schools have to play in helping to address the underlying causes of offending. Schools are likely to have a key role in a range of pathways - the most obvious of which is in ensuring high levels of literacy and numeracy, but others are: setting appropriate boundaries on behaviour; promoting an inclusive wider society; and broadening horizons and recognising potential. Schooling will also play a diversionary/ incapacitating role - expressed bluntly, whilst in education, potential offenders have less free time to engage in criminal activity.

Gottfredson et al show that school-based interventions designed to prevent the precursors of delinquent behaviour have more promising results when combined with some focus on self-control and social competency training. Indeed a number of authors have highlighted that cognitive-behavioural approaches are successful in reducing crime. While there are many school-based interventions, one approach that has been highlighted above is the use of nurture groups. The growing body of research on this will be of interest to colleagues in Justice, Education and Health and there may be a focus for some cross-government collaboration to evaluate and promote some of the many interventions that being used in schools to promote positive behaviours. There is also some evidence, from Sampson and Laub's work and from the Edinburgh Study of Youth Transitions and Crime, that attachment to school is an important protective factor and that school exclusion risks propelling young people into more engrained offending.

Sampson and Laub's work on 'turning points' is also particularly useful in reminding us that, although early life experiences do 'set the scene' for prospects in adulthood, there are key factors in adulthood that can redirect life trajectories. Bonds to society in the from of stable employment and good family relationships (particularly marital attachments) seem to represent important incentives to remaining law-abiding. This is consistent with the evidence cited in Section Two that social ties, or the lack of them, affect the deterrent impact of criminal justice policies. While stable employment has been identified as an important turning point, unfortunately, there is no robust evidence of the impact of employment interventions on offending. The evidence seems to suggest that existing offenders and those at risk of offending tend to have other needs to be met before employment programmes can be effective - this includes addressing motivational and educational issues. Programmes designed to improve employment prospects must therefore provide other social and educational supports if they are to be effective. While there may be a limit to what social policy can do to help generate social bonds such as employment, a key implication for justice is that our response to offending (in the form of criminal justice sanctions) should do as little as possible to undermine prospects for stable employment and family relationships.

While there is a growing body of evidence that biological factors matter, and that some people are born with a higher propensity to offend, the evidence that the impact of these factors diminishes as people grow up suggests that social factors can over-ride these. Social policy is by no means rendered irrelevant and, for those born with a biological predisposition to offend, quality of parenting, early socialisation, schooling and social bonds are even more important.

Finally, while there is a wealth of evidence that links offending both to drug and alcohol misuse, for some the underlying causes of offending and this may be the same and stem from low levels of self-control. Addressing drug and alcohol misuse is already a key priority for the Scottish Government. A focus on improving self-control through effective parenting is likely to make a valuable contribution to these efforts.