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

PART TWO: DETERRING OFFENDING BY ENSURING THE COSTS OUTWEIGH THE BENEFITS

While the most obvious approach to deterring potential offenders is to rely on the criminal justice system to impose formal sanctions on behaviour, this review also explores whether it is possible to draw on informal methods of social control. Generating and sustaining strong cultural norms that encourage compliance relies on a number of approaches. One approach is to encourage the potential offenders to recognise that his or her behaviour not only generates costs to him or herself but also that there are wider costs that have to be borne by the victim and by wider society. This is an approach that is central to restorative justice. A second approach is to foster a moral or ethical obligation or commitment to the rule of law. Research has suggested that people are more likely to obey laws if they have an underlying trust in the judicial process. Perceived legitimacy exists when the policed regard the authorities as having earned an entitlement to command, creating in themselves an obligation to obey. A third approach which is explored here is to encourage communities to exercise informal guardianship of their own public spaces. While vigilante movements represent the most extreme form of informal guardianship, a more subtle approach is to promote a local culture in which communities have sufficient pride and sense of responsibility to report crime and collaborate with the police and formal agencies in tackling it. The last informal method of social control that is explored is to rely on public condemnation and social stigma of criminal behaviour as a means of deterring potential offenders. This recognises that the shame associated with attracting moral opprobrium from family, friends and the local community can have a powerful effect in moderating behaviour.

This section also reviews the evidence on the relationship between the severity, certainty and celerity (swiftness) of formal criminal justice sanctions and crime rates.

INFORMAL METHODS OF SOCIAL CONTROL

Recognising the impact of behaviour

In their theory of delinquency, Matza and Sykes^[86] suggest that people justify their behaviour and participation in a crime internally using 'techniques of neutralisation', such as denying that any real injury was caused by their actions. The evidence on the impact of interventions designed to promote empathy and compassion has already been reviewed in Section One - Addressing the underlying causes of offending. This suggested that interventions designed to promote emotional intelligence, particularly those that draw on cognitive-behavioural theory, have been effective in reducing anti-social and offending behaviour. There is, however, less evidence of the effectiveness of simply confronting offenders with the impact of their behaviour. Comparing the evidence on restorative justice with that on cognitive-behavioural therapy suggests that to impact the decision-making process requires more intense forms of intervention grounded in techniques of social psychology.

Ensuring that offenders understand the impact of their behaviour on the victims of their crime is the central objective of restorative justice approaches. Restorative justice is a broad term that covers a range of practices which share the aim of repairing the harms (including material and psychological) caused by criminal behaviour^[87]. The core values of restorative justice include: respect, accountability, consensual participation and decision-making, and the inclusion and empowerment of all relevant parties. This process of a restorative conference is thought to make it more difficult for offenders to neutralise their behaviour. Face to face contact with the victim, together with increased awareness of the harm their actions caused, is thought to challenge denials and justifications. Others have suggested that the process of restorative conferencing also provides a remedial opportunity for moral development in young offenders who may hitherto have had limited exposure to morally formative experiences^[88]. The offender's apology and the victim's forgiveness are generally regarded as essential parts of the process of emotional healing and the key to the successful outcomes of satisfaction and reducing recidivism^[89].

Although restorative justice aims to shame offenders by confronting them with the impact of his or her actions, the requirement that they make reparation provides the means of forgiveness. Shame in restorative justice is therefore 're-integrative' and thought to bring about better outcomes. All of the above suggests that the restorative justice process may have a differential impact on a young person's self-concept (that is, the construct of identity, self-esteem, self-efficacy and personal agency) than traditional criminal justice approaches.

New Zealand was the first country to place family group restorative justice on a statutory basis with the introduction of a new youth justice system in 1989. This focused on family group conferences made up of the young person who committed the offence, members of his or her family, the victim(s) or their representative, a support person for the victim(s), a representative of the police, and the mediator or manager of the process. There are three principal components to this process:

• ascertaining whether or not the young person admits the offence;
• sharing information among all the parties at the conference about, for example, the nature of the offence, the effects of the offence on the victims, the reasons for the offending, and any prior offending by the young person;
• deciding the outcome or recommendation^[90].

The New Zealand model was adapted in New South Wales and then later introduced to the UK by the Thames Valley Police.

Previous studies suggest that while restorative conferences can lead to a large drop in post-intervention offending rates in severe offences of youth violence, this was not the case for those convicted for less severe offences such as shoplifting or property offending^[91]. A review of 'restorative justice' was conducted in the UK in 2007^[92]. This review reported mixed results on rates of re-offending with significant decreases in reoffending against control groups in some studies, no significant difference in other studies, and some evidence of increased re-offending in two small ethnic minority samples in Australia and the US. Since then, a further evaluation report has been published on this issue in the UK^[93]. This reports a significant decrease in the frequency of reconviction over the following two years (when all three restorative justice schemes are considered together), with the likelihood of reconviction over the next two years tending towards the positive direction, although not statistically significant. A recent Cochrane Collaboration review of restorative justice conferencing was undertaken by the Queen's University, Belfast^[94]. Four randomised controlled trials were included in the review, which concluded that there was no difference in reoffending rates between those who took part in restorative justice conferences and those in normal court proceedings. There was also no robust evidence to suggest that restorative justice conferences have an impact on young offenders' sense of remorse and recognition of their own wrongdoing, although there was some indication that victims benefit from them more than they do from court proceedings. Overall, the quality of the available evidence was judged to be low, so the review concluded that further research is needed.

Moral/ethical commitment to the rule of law

Tyler argues that people are more likely to comply with social rules when they believe that the authorities responsible for those rules are legitimate^[95]. Tyler and others^[96] argue that social regulation and the encouragement of voluntary civic behaviour are difficult when authorities can rely only on their ability to reward and/or punish citizens. The alternative to such strategies is to focus on approaches based on appeals to internal values. Two key aspects are trust and legitimacy: the public must trust the criminal justice system to be fair and effective if they are to comply with it, and they must also accept that the police, courts and other criminal justice institutions have the legitimate right to exercise authority over the public. Research suggests that using fair decision-making procedures ('procedural justice') is central to the development and maintenance of supportive internal values. Those authorities that use fair decision-making procedures are viewed as more legitimate, and people are more willing to comply with their decisions. Research has suggested that there are four key factors that contribute to judgements about fairness: participation (the extent to which people are allowed to express their views about how a problem should be resolved); perceptions of the neutrality of the authority; perceptions of the trustworthiness of the authority (which can be a function of following professional rules of conduct); and being treated with dignity and respect.

This concept of procedural justice has been applied to the criminal justice system. On the basis of various surveys of the public, Tyler has demonstrated that public perceptions of the fairness of the justice system in the United States are more significant in shaping its legitimacy than perceptions of its effectiveness. Whilst certain people may disagree with some of these laws, they nevertheless obey them because they think complying with the authority that enacted them is the right thing to do. Tyler's main focus has been on the interactions between officials and the public, and the ways that the behaviour of officials builds or erodes institutional legitimacy. The police are the most visible agent of social control and the most high-profile institution in a justice system that is empowered to define right and wrong behaviour.

Hough et al^[97] demonstrate how procedural justice theory has been tested in the UK. The National Policing Improvement Agency (NPIA) fielded core questions on public trust and police legitimacy in a representative sample survey of England and Wales. The survey tested what relationships exist between measures of public trust in the police, measures of perceived legitimacy, and people's self-reported compliance with the law and cooperation with the police. The data demonstrate that trust in the police was an extremely powerful predictor of perceived legitimacy and that perceived legitimacy was, in turn, a powerful predictor of compliance with the law (even when controlling for personal morality). Whilst trust in police effectiveness is also an important predictor of people's sense of the risk of sanction, the perceived risk of sanction is not the most significant factor in compliance with the law. This suggests that formal methods of deterrence are not the quickest routes to securing compliance. The ways in which the police wield their authority in part generates their perceived legitimacy, and if they treat people unfairly, legitimacy suffers and people become cynical about human nature and legal systems of justice. This then leads them to view certain laws and social norms as not personally binding.

Tyler argues that normative compliance (which occurs when people feel a moral or ethical obligation or commitment to do so) is economically more viable and is more stable over time than instrumental compliance (which relies on deterring through sanction and punishment). The importance of protecting the rights of the policed (suspects and defendants) as well as promoting greater consumer satisfaction on the part of the 'law-abiding majority' is therefore crucial.

The theory of procedural justice is being further tested by exploring patterns in attitudes across the different legal system through the European Social Survey. The Euro-Justis project, a large European project with partners in seven EU member states, was conceived to develop a standardised set of survey indicators which could be used to assess criminal justice across Europe, including measuring confidence in the criminal justice system^[98]. As part of this project, the researchers developed a 45 question module on trust in justice, administered as part of the European Social Survey (ESS) in 2010. The total sample was around 39,000 adults in 20 countries.

The results of the survey demonstrate the variation across Europe when it comes to trust in justice and belief in the legitimacy of justice institutions. Overall, Scandinavian countries were found to be most trusting of their police and courts, while former Communist countries in Eastern Europe tended to be less trusting.

In questions regarding police legitimacy, it was found that there was a relatively strong national-level correlation between the sense of obligation people feel to obey the police and their perceived moral alignment with the police; countries where there is a relatively strong sense that the police and citizens have shared moral values also have citizens with a relatively strong sense of duty to obey the police. Similarly, there was a clear correlation between belief in the fairness of the judiciary and trust in judicial competence.

In a report on the UK data (sample size 2,422)^[99], Hough et al. focus on the results regarding questions on buying stolen goods, what these data tell us about attitudes towards the police, and how these data support a model of pathways to compliance with the police. These pathways are:

• Instrumental compliance: when people's decisions about compliance with the law are based on self-interest. The researchers found that there is a small, but statistically significant impact on the buying of stolen goods if people believe that the police are effective, and thereby that the risk of getting caught is high.
• Morality of the act: complying with the law because it matches your moral principles. The ESS data show that people who think it is wrong to buy stolen goods also tend to report not having bought stolen goods (which is likely to be because they haven't bought stolen goods, but could also be because of a reluctance to admit to doing things which they think are immoral).
• Moral alignment with the police: people are more likely to comply with the law if they feel aligned with the values of the police. People who believe that the police represent a moral exemplar also feel they have a corresponding role to act in accordance with the law.

The ESS data show that the latter two pathways to compliance with the police are stronger than instrumental compliance, supporting Tyler's argument that strengthening the public's alignment with the police and the law (normative compliance) is a more effective method of reducing crime than increasing people's perceptions of the effectiveness of the police and the likelihood of being sanctioned (instrumental compliance).

In terms of co-operation with the police and beliefs in procedural fairness, the ESS data show that people's experiences of contact with the police play a key role in shaping their views of procedural fairness. The survey asked people if they had been approached, stopped or contacted by the police in the past two years and how satisfied or dissatisfied they were with how they were treated. Both neutral contact (those who were 'neither satisfied nor dissatisfied' with their experience of police contact) and negative contact (those who were 'dissatisfied' or 'very dissatisfied') with the police correlated with lower levels of trust in police procedural fairness. Positive contact was positively correlated with trust in police procedural fairness, but did not significantly correlate with trust in police effectiveness. Moral alignment with the police is also strongly predicted by people's trust in the procedural fairness of the police, which suggests that people view the police as having a moral authority because they perceive the police to use their power and authority in ways which are appropriate and fair. Moreover, this association exists even when adjusting for wider forms of interpersonal and political trust; so even though people who are generally trusting of others are also likely to trust the police, there is something more specific about how people come to trust in the legitimacy and fairness of the police.

A recent Campbell review looked at evidence of the impact of police-led interventions designed to increase citizen perceptions of police legitimacy^[100]. The review synthesised findings from 30 experimental and quasi-experimental studies (covering 41 independent evaluations) of any type of public police intervention explicitly aimed at improving perceptions of police legitimacy or which explicitly utilised principles of procedural justice. A range of interventions were covered, including problem-oriented policing strategies, restorative justice conferencing and alternatives to traditional police complaints procedures, but the most common intervention type was some form of community policing. Overall the review concluded that interventions designed to improve perceptions of police legitimacy do have a positive impact, but the type of intervention implemented is not as important as the fact that the intervention involves procedurally just dialogue between the police and citizens. As such, any policing strategy could theoretically be used as a vehicle for improving perceptions of police legitimacy, as long as it involves some element of procedurally just police-citizen interaction e.g. the police actively encouraging citizen participation, demonstrating neutrality in decision-making and showing dignity and respect to citizens.

A Scottish Government review of the evidence on public attitudes to the justice system looked at what those attitudes are, what drives them, and what works in terms of improving public attitudes^[101]. The review set out to look at attitudes to both criminal and civil justice institutions, but found that there was insufficient evidence on attitudes to civil justice to draw any meaningful conclusions. One of the main findings of the review was that people's personal experience of the justice system is the key influence on their attitudes towards the system. The evidence supports Tyler's argument about the importance of procedural justice, as when people perceive the justice system to be fair this leads to greater confidence in and satisfaction with the system, and is also associated with people being more compliant with the law and cooperative in their interactions with the justice system.

Community cohesion and informal guardianship

Evidence suggests that there is a link between 'community cohesion' or 'collective efficacy' and rates of crime within neighbourhoods. The theory is that communities with high collective efficacy are more effective at exercising control or guardianship over their own public spaces and that this in turn deters potential offenders by increasing their perceived risk of detection. The seminal research on collective efficacy is that of Sampson and Raudenbush. For their purposes, they define collective efficacy as 'linkages of cohesion and mutual trust with shared expectations for intervening in support of neighbourhood social control'^[102]. In their empirical study in Chicago in 1995, they explored the relationship between visual signs of social and physical disorder, collective efficacy and predatory crime in neighbourhoods selected to provide a mix of race/ethnicity and socio-economic status.

One of their main motivations for the study was to test the 'Broken Windows' thesis; that visual signs of disorder (such as broken windows and graffiti) drive up serious crime by suggesting to potential offenders that residents are indifferent to what goes on in their neighbourhoods and are therefore unlikely to intervene or to call the police when they observe criminal events. (See section below on public condemnation of criminal behaviour.) However, Sampson and Raudenbush's conclusions are also relevant to an assessment of the impact of community cohesion on crime because they demonstrate that the higher the levels of collective efficacy within a community, the lower the rates of crime.

In their study, visual signs of social and physical disorder were systematically recorded through researcher observation in just under 200 neighbourhoods. Signs of social disorder included public drinking, groups hanging around, soliciting and fighting and signs of physical disorder included graffiti, empty beer bottles and drug paraphernalia. Collective efficacy was measured via a survey of 4000 households across the study neighbourhoods which included questions about willingness to intervene to prevent acts such as truancy, graffiti and fighting in front of one's house and perceived levels of trust and cohesion^[103]. Predatory crime (violence, burglary and robbery) was measured by combining officially recorded crime data with data on victimisation in the previous 6 months (which were also collected via the neighbourhood survey).

In addition, the researchers collected a range of other socio-demographic data (primarily from census records) that were thought to be linked to disorder, crime and collective efficacy and therefore needed to be controlled for in their analysis. These data included a measure of residential stability (the assumption being that some stability was necessary to make residents feel they had a stake in tackling disorder and crime); the density of residential (as opposed to commercial) population; perceptions of disorder among residents (which could impact on a sense of whether it was worthy trying to affect improvements and hence on collective efficacy); and the extent of economic disadvantage within each neighbourhood (which would impose economic constraints on action).

The analysis revealed firstly that concentrated disadvantage was the single most important predictor of both disorder and crime. However it also revealed that collective efficacy was a more powerful predictor of all three crimes (violence, burglary and robbery) than signs of social and physical disorder.

Before concluding that collective efficacy was (along with concentrated disadvantage) a powerful predictor of violence and burglary, Sampson and Raudenbush first had to test whether the direction of causality was running in the opposite direction- whether it was crime rates that were impacting on levels of collective efficacy. This could have been the case if, for example, crime was driving residents away from an area and reducing residential stability and if fear of crime was reducing interaction in public places. They tested this by controlling for homicide rates in the year preceding the fieldwork and found that the relationship between collective efficacy and current crime rates still held fast.

Sampson also tested the relationship between violence and collective efficacy in Stockholm, Sweden, which has a very different social structure and different levels of violence to Chicago.^[104] Nevertheless, the evidence suggests that the same relationships held true - rates of violence are predicted by collective efficacy and, as in Chicago, collective efficacy is promoted by housing stability and undermined by concentrated disadvantage.

Conversely, the criminologist Sandra Walklate has argued that there is a link between areas with high levels of community cohesion and high crime rates^[105] and that 'high crime areas can be experienced as highly ordered and safe places for the people who live there'^[106] (Walklate, 2002: 84). Areas which may appear to those outside them as being unsafe, chaotic and lacking in community cohesion, may actually feature dense and complex social networks with substantial bonding social capital. One feature of dense, tightly-bound communities is that they can be insular and suspicious of outsiders. In areas characterised by high levels of unemployment, crime and social exclusion, individuals who live in the area may have strong extended family and friendship networks, but very few links to those outside the community, for example to people from different demographic groups or to public sector organisations (a lack of 'bridging' and 'linking' social capital). This can engender distrust of, and hostility towards, outsiders and 'the authorities'. In such communities there may also be a strong culture of 'not grassing'^[107], with those who have been identified as a 'grass' being criticised, socially excluded, or worse. As a result people may be reluctant to engage with local authorities, the police and other public sector organisations, which leads to crime going unreported and hinders the success of initiatives aimed at tackling ASB and crime.

Walklate conducted a two and a half year qualitative research project looking at the fear of crime in two neighbourhoods in Greater Manchester^[108]. One of the communities was tightly-knit and cohesive, with close family and kinship ties and a strong local identity. In this community, she found that many residents expressed the view that you were safe from crime if you were a local, as people in the neighbourhood would not 'rob off their own'. The community was viewed as being 'self-policing', in the sense that people felt that people in the area would not victimise those in their own community, and that if they did that someone would 'sort it out'. On the other hand, it was also self-policing in that people are deterred from reporting crime to the police because of the very real threat of public shaming as a 'grass'.

In contrast, the other community studied in the research was marked by rapid social change and an influx of newcomers. Views on safety were markedly different in this neighbourhood, with older people regularly referring to high crime rates and a decline in neighbourliness and trust. Young people were more accepting of problems in the area.

So, Sampson and Raudenbush's research suggests, first and foremost, that social policy should seek to ameliorate concentrated disadvantage but that it should also seek to enhance the social conditions that foster the collective efficacy of residents and organisations as part of strong, resilient and supportive communities. The work of Walklate and others also suggests that in addition to community cohesion, communities also need to have trust in the police and the justice system, which requires building connections between those communities and the police and local authorities.

The traditional approach to improving community cohesion has been the physical rehabilitation of deprived areas but in recent decades we have seen a shift towards trying to empower community residents to take preventative action themselves for example to tackle crime. According to Hope^[109], part of the rationale for this approach is that transferring resources to local communities gives them a stake in conformity and relieves the frustration of blocked aspirations. Community empowerment is also thought to encourage residents to stay in an area and take more interest in and responsibility for their neighbourhood. In a similar way, Sampson^[110] discusses the links between collective efficacy and community policing. He argues that by fostering greater civic involvement by residents in the social life of their neighbourhood, the police act as a catalyst to developing a sense of local ownership of public space and an increase in informal social control. He describes, for example, evidence from the Chicago Alternative Policing Strategy that demonstrated that 'beat meetings' between police and the community provided a means for civic involvement that has "been problematic in many poor communities". (The evidence in the effectiveness of community policing in reducing crime is discussed in more detail in the 'deterrence' section of this paper.)

While commentators have argued that we should be fostering collective efficacy, there is, as yet, no reliable evidence that encouraging communities to take action themselves to deter and prevent crime has actually been effective in reducing crime. Welsh and Hoshi^[111] reviewed four programmes that were designed to mobilise communities to participate actively in planning and implementing crime prevention activities. Each of these programmes was carried out in the United States in high crime areas. Unfortunately, the methodological rating of the evaluations was low and only one measured impact beyond the immediate outcome. Results were also shown to be mixed. The authors therefore concluded that the effectiveness of mobilising the community to tackle crime is unknown. We would also argue that testing for a relationship between community mobilisation efforts and crime rates is not enough and that Sampson and Raudenbush's work suggests we should be testing for a more sophisticated interplay between community-driven crime prevention interventions, collective efficacy and crime rates- with collective efficacy being the pivotal variable. We should also be taking an interest in the many other pathways of fostering collective efficacy.

Public condemnation of criminal behaviour

The role of parents and the wider community in setting appropriate boundaries on behaviour has already been discussed in Part One of this paper (which discussed addressing the underlying causes of offending). Ideally, this approach should socialise citizens into accepting the prevailing social order from an early age. However, for those who have not been effectively 'socialised', communities (as well as wider society- through the formal criminal justice system) still have a role in deterring them from offending by sending a clear message that this behaviour will not be tolerated.

Much of the relevant literature on the role of the community in deterring crime focuses on the way in which a community conveys visual signs to its residents and visitors about its attitude to towards crime. The theory is that a physical environment that has been neglected and allowed to run-down can have a role in breeding crime. According to the 'Broken Windows' thesis^[112], unchecked signs of disorder signal to potential offenders that residents are indifferent to what goes on in their neighbourhood and are therefore unlikely to intervene or to call the police when they observe criminal events. A further theory in the 'Broken Windows thesis is that visual signs of disorder send the signal that an area is unsafe. Law abiding citizens therefore avoid the area and leave it even more vulnerable to more disorderly behaviour and serious crime. Some commentators^[113] argue that not only do law-abiding citizens avoid the area but also that businesses and services leave the area and that this, in turn, undermines the social and economic conditions that sustain law-abiding communities. This thesis was very influential in American policing policy in the 1990s and led, for example, to New York's 'zero tolerance' approach.

There is evidence from Sampson and Raudenbush's testing of this thesis (discussed above) that, after controlling for other characteristics of neighbourhoods, there does seem to be a relationship between observed social and physical disorder and robbery (but not with violence or burglary). Sampson and Raudenbush conclude that areas with greater cues of disorder appear to be more attractive targets for robbery offenders. However collective efficacy appears to be a more powerful predictor of rates, not only of robbery, but also violence and burglary than signs of physical or social disorder. This is not to say, however, that signs of disorder are irrelevant to understanding these types of crimes. Disorder may be important for understanding residential stability patterns, investment by businesses, and overall neighbourhood viability which will in turn have an impact on efforts at building collective response and hence an indirect effect on crime. In other words, eradicating disorder may indirectly reduce crime by stabilizing neighbourhoods.

Whatever the causal mechanism, there are a number of research studies that have shown that removing visual signs of disorder have had an impact on crime. Sloan-Howitt and Kelling^[114] show how a clean up programme, that was undertaken to remove graffiti from all train cars and stations on the New York subway system, significantly reduced the incidence of new graffiti. Carr and Spring^[115] also found similar improvements in Victoria, Australia following a programme promoting the rapid repair of vandalised train equipment. Interestingly, they also found that crimes reported against the person declined by 42%.

FORMAL METHODS OF SOCIAL CONTROL

The severity and certainty of punishment

A comprehensive review of deterrence research was commissioned by the Home Office in the late 1990s. The review^[116] examined studies that tested the statistical relationships between changes in the certainty or severity of punishment and crime rates. Most notable of these studies are Farrington, Langan and Wikstrom^[117], and Langan and Farrington^[118] which examine crime and punishment trends in England and the U.S. in the 80s and 90s. This research, confirming earlier deterrence studies, finds substantial negative correlations between the likelihood of conviction (a certainty measure) and crime rates. The evidence on the severity measure was, however, weaker. Any negative correlations between sentence severity and crime rates generally were not sufficient to achieve statistical significance. The Farrington, Langan and Wikstrom study compared the English and American (as well as Swedish) trend data. Given that the U.S. penalty levels were substantially higher than English levels during the periods studied, the absence of a finding in that study of strong correlations particularly undermines the notion that severity of punishment has a marginal deterrent impact.

By way of caution, the authors of the Home Office review remind us that changes in criminal justice policies can have no deterrent effect unless they alter potential offenders' beliefs about sanction risks. Thus even where crime rates are statistically associated with changes in the severity of punishment, this would not establish a deterrent effect unless there were also evidence that potential offenders were aware of those changes. The review therefore goes on to examine the considerable body of recent literature on 'perceptual' deterrence. This literature examines the link between potential offenders' beliefs concerning likelihood or severity of punishment, and their reported decisions or expectations of offending. However, these studies largely focus on certainty variables, such as perceived likelihood of prosecution, and do not shed much light on questions of severity effects. The report concluded that the studies reviewed did not provide a basis for inferring that increasing the severity of sentences generally is capable of enhancing deterrent effects.

A more recent review by Michael Tonry^[119] argued that there is evidence that changes in enforcement and sanctions can effect some kinds of behaviour (e.g. tax evasion and driving offences) but that there is little credible evidence that they affect all types of crimes. Tonry cites research conducted from the mid 1970s to the late 1980s that examined the deterrent effect of new laws that prescribed mandatory minimum sentences for certain crimes. This research suggests that either there was no measurable deterrent effect or that there were short-term effects that quickly subsided. He also discusses the 'tough on crime' initiatives that were implemented in America during the 1990s, including the 'zero tolerance' policy in New York City and the 'Three strikes and you're out' law in California. While it was claimed they were associated with a dramatic reduction in crime, there were no robust studies that demonstrated this. Tonry argues that downward trends in crime began before the initiatives were announced and that neither the timing nor the rate of decline were any different in New York than in other states. While econometric reviews of the research have tended to reach more positive conclusions about the deterrent impact sentences, these have been criticised for their failure to take account of the social context of offending - the notion, for example, that the deterrent impact of criminal sanctions will depend, among other things, on the potential offender's own assessment of the likelihood of receiving the sanction and his/her ability to achieve the goal legitimately, and informal social controls.

The Home Office review^[120] of these perceptual studies, and of a few available studies of actual offending behaviour, also provide additional confirmation for the hypothesis that social ties, or the lack of them, affect the deterrent impact of the criminal-justice policies -- with persons having strong social ties (i.e., strong links to families, local communities, etc.) being the more readily deterred by prospects of being apprehended.

Increasing the risk of detection

One of the most obvious ways of increasing the certainty of punishment is to increase the likelihood of detection. The evidence on how policing practice can impact on detection is therefore discussed next.

Sherman and Eck^[121] describe how the deterrent effect of having a police agency has been demonstrated by studies that examine changes in crime rates before, during and after police strikes. The evidence from these studies consistently shows that crime rates increase markedly during strikes. Whether adding more officers to an existing police force has any impact on crime is, however, less clear. In a review of the evidence exploring the relationship between police numbers and crime rates, Bradford^[122] posits that until the 1990s, there was a lack of clarity in terms of the impact that increasing police numbers had on crime reduction. However, consideration of recent, more robust evidence found a degree of consistency in their findings, namely that while reduced property crime is associated with increased numbers of police officers (it is estimated that a 10% increase in police numbers would result in an approximate 3% reduction in property crime - the elasticity rate of property crime being -0.3), the link between police numbers and violent crime is weaker. Bradford concludes that there is relatively strong evidence to suggest that there is a relationship between increased police numbers and reduced crime, particularly in terms of property-related crime. However, given that his review was not systematic, Bradford urges a degree of caution in generalising from the results. It is also suggested that more evidence is required before a causal link between higher police numbers and lower crime can be established.

Unless studies can demonstrate how extra police officers are being deployed to tackle crime, the absence of a causal mechanism makes it difficult to interpret any statistical associations between police numbers and rates of crime. Studies that examine the impact of deploying more police officers to specific activities have far more potential to inform policy.

Studies that have examined the impact of increasing local patrols in crime hotspots have shown that this is associated with reductions in crime. Sherman and Eck^[123] review 11 studies and conclude that the more precisely patrol presence is concentrated at the 'hot spots' and 'hot times' of criminal activity, the less crime there will be in those places and times. The study which applied the most scientific rigour (an experimental design) was Sherman and Weisburd's analysis of hotspot patrolling in Minneapolis in the late 1980's. In this study the police department reorganised its entire patrol to test a pattern of directed patrols at randomly selected crime hot spots during peak times. These extra patrols were resourced by reducing patrols in low crime areas. Both the hotspot and non-hotspot areas were subjected to over 7,000 randomly selected hours of observation by researchers over the course of a year. Sherman and Weisburd's analysis revealed that, while crime-related calls for service increased in both groups, the increase in the areas that received no extra patrols was up to three times as great.

Braga et al.'s review of the effects of hotspot policing on crime^[124] also concluded that hot spot interventions are an effective crime prevention strategy. The review included 19 US studies (of 25 interventions), 10 of which were randomised control trials and 9 quasi-experimental studies. The interventions included problem-oriented policing, increased patrol strategies and drug enforcement operations. Twenty of the 25 interventions reported substantial crime control benefits in the target areas, and a meta-analysis of reported outcome measures found that the crime reduction effects in intervention areas compared to control areas was small but statistically significant. The interventions had a positive impact on violent crime, public order offences, drug offences and, to a lesser extent, property crime. The researchers also found that diffusion of crime control benefits was more likely to be a side effect of hot spot policing interventions than crime displacement. Only 3 of the studies reported substantial displacement of crime into areas surrounding the treatment area, whereas 8 studies suggested possible diffusion effects. In terms of the type of intervention, the meta-analysis indicated that problem-oriented policing interventions were approximately twice as effective as those which simply increased police patrols in the target area, although this should be interpreted with caution as there were a relatively small number of studies of each type of intervention reviewed.

Similar evidence from the UK is provided in a Home Office evaluation of a police initiative in Humberside that was established in 2000^[125]. Its aim was to tackle personal robbery and alcohol-related disorder through community engagement and high visibility patrols. Two types of foot patrols were provided: targeted patrols in hotspot areas during 'hot times'; and general patrols across the entire city. This was delivered by an additional 14 police officers providing, in total, an extra 24,000 hours of high visibility patrols. The evaluation found that after a year there was a 16% reduction in recorded personal robbery in the hotspot area, compared with a 5% increase across Humberside (the police force area) and a 15% increase across the UK generally. The fact that there was also a drop (albeit a lesser drop) in personal robbery in an adjacent area that did not attract additional patrols, was explained by benefit diffusions.

A 2012 Campbell Collaboration review looked at the effectiveness of "pulling levers" focused deterrence strategies in reducing crime^[126]. Pulling levers approaches usually involve selecting a specific crime problem, e.g. gang homicide or street drug dealing, and conducting research to identify key offenders or groups of offenders. An interagency group is formed, including the police, probation and parole officers etc, in order to undertake an enforcement operation targeted at the key offenders, using every available "lever" to sanction offenders. Offenders are communicated with directly, for example at mass "call-ins", where they are told why they are under scrutiny, what enforcement action will be taken if they continue to offend, as well as confronting them with moral voices from within their communities, such as crime victims, to attempt to get offenders to face the impact of their behaviour on others. As well as enforcement, multi-agency co-ordination is used to help direct offenders to appropriate services, such as employment services or treatment for substance misuse.

Ten U.S. studies using non-randomised quasi-experimental designs were included in the review. No randomised experimental studies were identified by the researchers. Nine of the ten studies reported significant reductions in violent crime, including a 63% reduction in youth homicides in Boston and a 44% reduction in gun assault incidents in Lowell. Three of the ten studies also looked at possible crime displacement and diffusion effects, and two of these reported statistically significant reductions in drug offences or violent crimes (depending on the nature of the intervention) in areas adjacent to those where the interventions took place.

Despite the lack of randomised experimental evaluations of pulling levers interventions, the review concluded that the available evidence is very positive, but that caution should be taken with regard to the effect sizes reported, as the evidence suggests that less rigorous research designs are associated with stronger reported effects. The researchers highlight that this sort of targeted approach appears to alter offenders' perceptions of the risk of being caught and sanctioned. The review also highlights the benefits of multi-agency working between the police and other criminal justice agencies, not only to increase the risk of detection through increased police presence, but also in redirecting offenders away from crime through the provision of social services and employment opportunities etc. They also suggest that pulling levers approaches increase collective efficacy in the communities they are implemented in, by engaging with community members to address offender behaviour at call-ins, for example. However, because of the nature of the evaluations reviewed, the researchers were unable to fully unpack the way these focussed deterrence approaches work and robustly evaluate the extent to which different aspects of the interventions contributed to the overall effectiveness of the programmes. Because of the multi-dimensional nature of these interventions, it would be beneficial for future work to tease apart the different strands in order to assess how the interventions work, and whether there are some elements which are more effective than others.

There is also evidence about the impact of community policing, a key feature of which is the involvement of community and other partners in identifying, prioritizing and solving neighbourhood problems. A review of evidence on the impact of community policing that was commissioned by Justice ASD and undertaken by Mackenzie and Henry^[127] shows that community policing has been associated with a reduction in both crime and disorder, although there is stronger evidence for its effectiveness in reducing disorder than crime. Evaluations of particular community policing programmes have tended to find reductions in crime (usually measured by victimisation surveys) in some, but not all, of the areas studied. The positive results in relation to the reduction of disorder have been suggested to be related to two strands of the community policing approach in particular: foot patrol and problem solving. However, as the authors note, there are several features of the concept and practice of community policing which render an evaluation of the effectiveness of this model of policing difficult; one of which is that there is no consistent definition of what community policing entails.

Overall, substantial evidence indicates that the way in which police officers are deployed has the greatest impact on preventing serious crime, for example through directed patrols, proactive arrests and problem solving at high crime hot spots.

Celerity of punishment

While the link between certainty and severity of punishment has been reasonably well explored in the research literature, the link between celerity (swiftness) of punishment and offending has not. Only two studies could be identified to help test this pathway of the model and neither provide convincing evidence that increasing the celerity of punishment will help to reduce crime.

The first is a study by Bouffard and Bouffard^[128] that examined data from several thousand 'driving under the influence' cases processed in a specialized court in the United States over the course of 6½ years. They examined the relationship between reductions in driving under the influence rates and changes in the certainty, severity and swiftness of case processing over time. The results suggest that swiftness, though often overlooked, may be an important component of effective deterrence.

A further study was based on a survey of undergraduates at the University of Arizona. The researchers used a drunk driving scenario to explore perceptions of what would deter them from this form of offending. The research concluded that variation in sanction certainty and severity would impact on offending behaviour but that, although there was some evidence of a positive effect of celerity, it was not statistically significant^[129].

CONCLUSIONS

There is consistent^[130]evidence that increasing the certainty of punishment is effective in reducing crime but only weak evidence of a relationship between severity of punishment and crime rates. There is very little evidence on the celerity of punishment. Neither of the two studies that were identified for this review provided robust evidence of impact. In terms of increasing the certainty of punishment, limited but relatively strong evidence suggests that there is an effect of increased police officers on reduced crime, particularly in terms of property related crime. However, further evidence is required to establish a direct causal link between the two. Increasing police numbers can be effective if greater resources are deployed and targeted appropriately. For example, increasing patrols in crime hotspots has been shown to be associated with a reduction in crime. While community policing has also been linked to a reduction in crime, the many elements involved in community policing make it difficult to identify exactly what it is about this approach that seems to work.

The evidence on the impact of formal criminal justice measures is far more developed than the evidence on informal approaches and much of this latter evidence is still tentative and developing. Nevertheless, in an era of reduced public spending, it is important that we try to understand how informal approaches that require less intensive investment might be able to complement our formal efforts to deter crime. Keeping abreast of the new evidence on this is therefore crucial.

The research on the impact of restorative justice is probably the best source of evidence on whether we can deter potential offenders by confronting them with the social costs of crime (and make it more difficult for them to 'neutralise' their behaviour). The evidence on the effectiveness of restorative justice in reducing crime/recidivism is mixed and there is a need for more high quality studies to be conducted. Nonetheless, there is some evidence to suggest that this approach is promising.

While the evidence on the impact of the criminal justice system suggests that increasing the certainty of punishment can have an impact on crime, Hough et al's work on procedural justice suggest that the perceived legitimacy of the criminal justice institutions is even more strongly correlated with willingness to comply with the law (even when controlling for personal morality). This suggests that formal methods of deterrence may not be quickest or the most economically viable method of securing compliance. The importance of building public trust in the agencies of the criminal justice system has to be emphasised, with the European Social Survey module and Campbell review adding further weight to the importance of due process and the principles of treating all those who come into contact with the criminal justice system with respect and dignity.

While there is some evidence of a link between signs of disorder and crime, collective efficacy seems to be a more important underlying factor. The links between collective efficacy and crime rates suggest that some communities are effective in exercising their own informal guardianship. Whether or not we can foster that sense of collective efficacy through social policy though is not clear. However, the traditional approach of investing in the physical infra-structure of communities has in recent decades, shifted towards also recognising the importance of community empowerment and fostering civic participation. The Glasgow Community Health and Wellbeing Research and Learning Programme (GoWell), is a 10 year programme investigating the impact of investment in housing and neighbourhood regeneration in Glasgow on the health and wellbeing of individuals, families and communities. It is anticipated that this programme may be able to provide analysis of changes in the incidence and geography of crime in these communities and shed light on the impact that regeneration has had on crime and antisocial behaviour.

Another approach is to try to foster collective efficacy and empower communities by involving them in the design and implementation of various crime prevention initiatives. This is consistent with the salutogenic approach that has been adopted in many health improvement initiatives. However, there is, as yet, no reliable evidence that this approach has been effective in reducing crime.