Crime, fear of crime and mental health: Synthesis of theory and systematic reviews of interventions and qualitative evidence

Aim and methodology of the review

The aim of the review of theory is to synthesise the available theoretical frameworks regarding the pathways between crime, fear of crime, health and well-being and the built environment, to develop a logic model for interventions and a causal map of relevant contextual factors. The findings of the review of theory were used to inform the design and interpretation of the systematic reviews that form the main part of the overall evidence synthesis.

The methodology employed for the review of theory was a pragmatic non-systematic review. The searching and selection of material were iterative, with phases of literature searching alternating with phases of synthesis and theory construction and testing. Search sources included Google Scholar, MEDLINE, Criminal Justice Abstracts and suggestions from subject experts within the review team. In addition, there was a strong emphasis on ‘pearl growing’ methods such as citation chasing. The selection of studies was purposive and context-sensitive and informed by the emerging theoretical picture at each stage; a priori inclusion criteria were not applied. As far as possible, selection was guided by the goal of theoretical saturation, although, because of the broad scope of the review, saturation could not be achieved in every area of the review. In the selection of theoretical studies, those with a scope similar to that of the present review and those that most adequately took into account the complexity of the relevant factors and their inter-relations were prioritised. In the selection of empirical data, studies with a robust methodology, particularly systematic reviews, were prioritised, as far as possible and appropriate. However, this prioritisation was purely informal in nature. Because of the widely varying quality of the evidence base in the different fields and questions investigated, consistent and explicit criteria could not be used.

The synthesis of data was based initially on an identification of similar theoretical concepts across the literature and an integration of these concepts into an overarching framework. The linkages between concepts were drawn initially from the most relevant theoretical literature and then filled out with reference to the empirical literature. As discussed in the following sections, our method was highly interpretive, with a strong emphasis on theory construction as an essential part of synthesis.

The synthesis resulted in two separate but linked models: a complex map of relevant contextual factors and a simpler and more linear logic model of interventions. This dual method of causal mapping has not been widely used in previous studies. It was adopted here because, owing to the scope and complexity of the larger map, intervention points and mechanisms could not be clearly identified within it. In addition, we had already made an a priori decision to focus on only one of the potential areas of intervention within the larger map, namely the built environment, so the logic model of interventions helped to clarify the consequences of this decision.

Appendix 1 includes a selection of theoretical models from previous research, with brief comments indicating how these have been utilised in the construction of our causal model. The selection of models focuses on those relating to fear of crime and/or crime and health; although we utilised causal models from the theoretical literature on other topics, such as the built environment and health, our use of these was generally more selective.

Methodological context

In this section we present some background to the choice of methods described in the previous section. We examine three methodological theories that have formed our thinking: realist synthesis, critical interpretive synthesis (CIS) and Baxter et al. ’s recent work on conceptual frameworks. 18 These theories are relevant, first, because all three have addressed the challenges of synthesising diverse types of evidence and, second, because they have all, in different ways, addressed the relation between evidence synthesis and theory construction. This section presents a brief overview of the three theories and then situates the methods used for this review with respect to these theories and the broader methodological landscape.

The realist synthesis approach has been recommended as particularly appropriate for the synthesis of disparate and complex data. 16,32,33 The characteristic feature of realist synthesis is an emphasis on the theory-laden and context-dependent nature of praxis within intervention programmes, systems and institutions. Hence, the synthesis of evidence about interventions or systems is inseparable from an engagement with the theories implicit in those systems. This engagement may frequently take the form of an analysis of causal pathways (although in most realist syntheses these analyses have been relatively linear, in the sense described in Models and theories in evidence synthesis). For proponents of this method, the synthesis of diverse evidence types (effectiveness data, qualitative data and theory), and the use of iterative searching and purposive selection, are required to understand ‘what works, for whom, how, and in what circumstances’. 33

Critical interpretive synthesis34,35 is similar to realist synthesis in some respects. One difference is that CIS emerged from the development of methods for qualitative synthesis rather than the synthesis of effectiveness data, for which there has generally been a stronger emphasis on theory construction as part of the synthesis process. 36 However, the proponents of CIS argue that it is not limited to the synthesis of qualitative data but can also be used for the synthesis of multiple types of evidence. CIS draws on Noblit and Hare’s37 method of meta-ethnography to develop the fundamental distinction between an ‘aggregative’ or ‘integrative’ approach, such as a traditional meta-analytic review of quantitative data, and an ‘interpretive’ approach, exemplified by CIS. Aggregative approaches seek primarily to draw together the available evidence and summarise it, and rely on the existence of well-specified concepts that can be used to carry out the summary without any deeper critical engagement. Interpretive approaches, by contrast, involve engaging critically with the conceptual frameworks found in the literature and developing overarching ‘synthetic constructs’. Here, CIS draws particularly on Noblit and Hare’s37 ‘lines-of-argument synthesis’, a type of synthesis in which the construction of new theoretical content is indispensable, by contrast with ‘reciprocal translational analysis’, which is limited to translating between studies to develop common concepts.

The third approach examined is that adopted by Baxter et al. 18 They present a methodology for developing a causal map for complex interventions (which the authors explicitly describe as integrative rather than interpretive in Dixon-Woods’ sense). 34 Initially the authors used a previously agreed framework38 to categorise the potential causal influences in the field under discussion. They conducted a non-systematic review of diverse types of evidence, consulted with an expert reference group and then coded the selected papers in depth using an approach derived from qualitative analysis. From these data, a causal map was constructed and revised using an iterative process. Although useful, Baxter et al. ’s methodology is less theoretically elaborated than those described above, and their description of the synthesis process is brief and does not draw on the methodological literature on qualitative synthesis (as reviewed by Barnett-Page and Thomas36).

Our methodology does not exactly line up with any of the three described above. It is closest to that of Baxter et al. 18 in the focus on causal mapping as the representation of theoretical content. However, our coding procedure was less formalised than that of Baxter et al. ;18 in addition, we would categorise our overall approach in Dixon-Woods et al. ’s terms34 as interpretive rather than aggregative (although see the discussion of this point below). On a practical level, the main item of guidance we draw from the three approaches described here is the use of iterative searching and purposive selection of studies, informed by emerging theoretical constructs. More broadly, we draw on their insights about the relation between synthesis and theory construction, and have sought to link these to relevant recommendations from the literature on causal mapping. (To our knowledge, few studies – with the partial exception of that by Baxter et al. 18 – have attempted to bridge these two distinct bodies of theory.)

However, our review diverges from these methodologies in two ways. The first point concerns the place of our review of theory in the overall evidence synthesis project. Like the methodologies described above, our review of theory includes diverse types of evidence including empirical data as well as theoretical constructions. However, unlike them, it is not designed to stand alone but to stand alongside the systematic reviews, which form a clearly separate phase of the project. Realist synthesis and CIS are comprehensive approaches explicitly designed to be an alternative to traditional systematic reviews, not a supplement to them (this point is less clear with regard to Baxter et al. 18). Hence, the greater synthetic power of these approaches must be set against the fact that they are substantially less transparent and reliable than systematic reviews, in those areas where systematic reviews are possible and appropriate. Our methodology represents an attempt to utilise the strengths of both approaches by combining a non-systematic critical review of theory with conventional systematic reviews in a way that maximises the potential to transfer insights and concepts from one to the other, while maintaining their methodological separation intact.

The second point of difference concerns the implicit assumption in all of these methodologies that the synthesis is different in kind from the primary materials included in the review, and that it stands, as it were, above rather than alongside the latter. Even in realist synthesis, with its emphasis on the theoretical content of interventions, it is clear that the theory developed in the synthesis is intended to be more inclusive and powerful than that implicit in the primary studies. By contrast, many of the ‘primary’ studies included in our review of theory are themselves exercises in wide-ranging theoretical mapping and hence of the same character as our review itself. The relation of the review to the included theories might be described as one of dialogue, rather than inclusion. Hence, the distinction between interpretive and aggregative approaches may not be applicable, as much of the primary material that is being synthesised consists itself of synthetic constructs, and so requires interpretation before it can be aggregated.

This was particularly true for our review of theory because we located no body of theory covering the whole scope of the review, or any shared consensus on the framework to be adopted. As set out in Chapter 3, there are highly sophisticated bodies of theory within particular areas, including crime prevention; the social and individual determinants of fear of crime; the links between the built environment and health; and the links between fear of crime and health or health behaviours. However, few researchers have attempted to map out the pathways between all of these factors simultaneously. Moreover, even within specific fields, consensus on the theoretical frameworks is often lacking. This is particularly true of theories of fear of crime (discussed in Chapter 3, Fear of crime: measures and contexts), on which researchers in the field frequently disagree on fundamental questions of methodology and definition.

Moreover, these diverse bodies of theory come from a wide range of disciplinary perspectives, including criminology and policing, sociology, psychology, public health and urban planning. Concepts in one field may not line up with those in another. For example, the concept of the built environment is only imperfectly translated to the policing field by the concepts of physical disorder, a narrower concept with a focus on visible problems, or place-based strategies, a concept that includes the social as well as the physical environment and is more closely linked to interventions. 39,40 More deeply, the theoretical bases underpinning work in different fields may be incompatible; for example, much recent fear of crime research has drawn on an expressivist paradigm that emphasises the social meanings of individual action,41,42 whereas the literature on environmental crime prevention largely remains within a rational choice perspective. 43

As a result, the synthesis could not proceed on the basis of a commonly defined vocabulary of concepts, as no such vocabulary was available, but had to proceed by drawing together heterogeneous theoretical constructs. In this context, the attempt to aggregate and translate between concepts to build a coherent causal map necessarily involved interpretation.

For reasons of space, this report does not present a systematic overview or glossary of all of the concepts used in the theories that form the sources for our synthesis, nor a full discussion of the deeper paradigms that underlie them. Hence, a full account of the issues raised above, and a record of all of the decisions made in constructing the synthesis, cannot be given. We have described specific issues in the relevant sections of Chapter 3 when they are consequential for the design or interpretation of the causal models.

Crime and disorder

In principle, all types of crime are included in the theory review; however, three broad categories are likely to be especially relevant. For simplicity, only these are included as subconcepts in the model. Violent or potentially violent crimes against the person, such as rape, assault and mugging, are likely to cause substantial physical and/or psychosocial health harms for the victim and may often form the focus of fear. Drug- and alcohol-related crimes include a range of offences, such as violence, criminal damage, driving and public order offences, and crimes such as theft and burglary to fund drug habits; they often have a spatial patterning reflecting the locations of drug markets, alcohol outlets and the night-time economy. ‘Environmental’ crimes are those that impact directly on the quality of the physical environment, such as criminal damage, vandalism and graffiti; this category may also include non-criminal forms of antisocial behaviour such as littering. Of course, these are very different phenomena. In particular, the inclusion in the same concept of environmental crimes and serious violent crimes does not imply an endorsement of a punitive agenda towards the former. In general, the scope of this project precludes a fully critical perspective on the concept of crime and disorder.

[Two further points should be added here on the definition of crime. First, some non-violent crimes against property, such as burglary and car theft, are also salient in that they may have substantial psychological and emotional impacts and be widely feared44 (on the distinction between crimes against the person and crimes against property relative to the fear of crime, see Chapter 3, Emotional response: further considerations). Second, given the environmental dimension of the review project as a whole, ‘domestic’ crimes such as child abuse and intimate partner violence will be less salient than ‘stranger’ crimes. This is in no way to minimise the extent or impact of domestic crimes, only to observe that they are of less relevance within the overall conceptual framework adopted for the review.]

Fear of crime

Fear of crime is a complex concept that raises a number of challenging definitional and methodological issues and is discussed further in Fear of crime; measures and contexts. The two most important subconcepts here are the individual’s perceived risk of being a victim of crime and his or her emotional responses to crime, such as worry or anxiety. The model also includes individual attitudes (e.g. beliefs about the nature and extent of crime) and perceived vulnerability to crime, as these are known to be closely bound up with fear. However, the perceived and actual rates of crime in the local area, and the individual’s objective risk of crime, are not included within the fear of crime concept, although they are linked to it. As discussed in Fear of crime; measures and contexts, it is difficult to draw a clear boundary around the concept of fear of crime, given its complex linkages with the social environment and individual psychology, and our definition represents a particular theoretical position that is not shared by all researchers in the field.

Health and well-being

Our perspective on health and well-being is comprehensive, including all forms of physical and mental health outcomes; health behaviours such as physical activity, and social well-being broadly conceived, including interpersonal interaction and social capital. As such, our concept of health and well-being overlaps substantially with the concept of the social environment (see Social environment).

Built environment

The built environment includes factors relating to the physical environment insofar as they are shaped by human activity. In particular, it includes the design of public spaces such as streets and parks, land use policy more generally (e.g. zoning regulations), public and private transportation and the architecture and design of residential housing. In the model, these are summarised with the subconcepts ‘housing’ and ‘public space and transport’. It also includes people’s interactions with the environment and the physical and geographical distribution of social factors. This category includes, for example, the sociodemographic make-up of communities, or social or legal restrictions on people’s movement within the built environment. In the model this is represented by the subconcept ‘neighbourhood and community factors’, which overlaps with the social environment.

Social environment

The subconcepts in the social environment cluster are based on a previous review of theory. 45 ‘Neighbourhood and community factors’ are defined in the Built environment. ‘Social inequalities’ refers very broadly to the effects of sociodemographic factors such as socioeconomic status (SES), ethnicity and gender, including structural inequalities and individual discrimination. ‘Interpersonal relationships and networks’ includes more local-level dimensions of social interaction and can be taken to include measures such as social capital and social cohesion or integration. As such, it overlaps with the health and well-being concept, as we take the latter to include the social dimensions of well-being as well as individual physical and mental health.

National policies

The role of national policies is not specified in detail in the model as a full discussion of their effects, particularly on health and well-being outcomes, lies outside the scope of this project. For our purposes, the main influences of interest are on the built environment and on crime. However, it is important to bear in mind that all of the other factors and relationships are shaped, more or less directly, by government policy, as well as by other macro-level determinants such as the economy.

Background

In this section we explore fear of crime in further depth. Fear of crime receives particular attention here for two reasons. First, there are complex relationships between the subconcepts that may be relevant for understanding the evidence and specifying the scope of the project. Second, there has been considerable debate about the concept’s meaning and coherence, and about the appropriateness and interpretation of the measures used. In discussing the findings of the fear of crime literature in the following sections, we have not always been able to give detailed attention to these issues, so this section serves as an overall guide to interpretation. This section is itself limited in scope: we do not engage with the full body of evidence on the determinants of fear of crime, only those that are of relevance to the model. In addition, we cannot here engage with the broader context of the debates around fear of crime research, particularly its political role, although the highly politicised nature of many apparently technical debates should be borne in mind. 46–50 Some of these issues are addressed further in the following sections.

Because of its extent and the ongoing contestation of fundamental issues, it is difficult to give a clear and uncontroversial overview of the field of fear of crime research. One basic distinction might be between, first, an older tradition, going back to the 1960s, which has been based on a broadly positivist, data-driven model of research, and which has focused on using observational research to explore the determinants of fear; and, second, a newer critical tradition, drawing on psychology and symbolic interactionist sociology, and strongly influenced by feminist thought, to understand fear as expressive of a broad range of attitudes and anxieties, and as rooted in the day-to-day reality of individual lives. Broadly, the two research paradigms have tended to correspond to a methodological divide, with the older positivist tradition emphasising quantitative survey measures and the newer critical tradition emphasising qualitative and ethnographic research. Only relatively recently, particularly with the work of Stephen Farrall, Jonathan Jackson and colleagues,49,51 has a viewpoint that draws together these two perspectives, and which utilises both qualitative and quantitative methods, emerged.

Within the first, more data-driven tradition, four main theories, in the sense of perspectives emphasising particular causal factors, can be distinguished in the literature. 52–56 Different authors divide up the theoretical field differently (e.g. some would include sociodemographic factors as a theoretical perspective in its own right), but these four are the main theories identified in the literature. The first is vulnerability theory, which emphasises the role of vulnerability to crime (defined further in Perceived vulnerability) in producing fear of crime and focuses particularly on explaining differences in fear between sociodemographic groups. The second is social disorder theory or social disorganisation theory, a more ecological approach that emphasises the role of local physical and social environments in engendering fear. The third is victimisation theory, which sees fear of crime as primarily driven by actual crime victimisation, and holds that it can be explained by the same factors as crime itself. The fourth is social integration theory, which emphasises the role of strong social networks and attachments at a local level as protective factors that may reduce fear.

Researchers with a more synthetic perspective have seen these theories as reducing in turn to two paradigms, a rationalist and a symbolic paradigm. 57 A rationalist view of fear of crime would see it as based primarily on approximately correct estimates of the risks and potential consequences of victimisation, whereas a symbolic view would see fear as bound up with the social meanings and representations of crime and disorder. However, this division does not line up perfectly with the theories described above. Victimisation theory is clearly rationalist; vulnerability theory is rationalist to some extent, but may need to appeal to symbolic accounts in explaining how estimates of the consequences of crime spill over into assessments of risk. Social disorder and social integration theories are usually expressed in symbolic terms, but they may have a strongly rationalist component in that disorder and low social integration may be causally linked to crime and function as roughly accurate indicators of actual risk (see Social environment and crime). As discussed further below (see Chapter 3, Perceived risk and emotional responses), teasing apart the rational and symbolic, or cognitive and affective, components of people’s behaviour may frequently be difficult, particularly in the context of secondary research. More generally, some researchers in the critical tradition have questioned the construct of ‘rationality’ in this context, arguing that judgements of rationality or (implicitly) irrationality are difficult to justify with reference to fear58 and that the dichotomy is in any case a blunt instrument in understanding the place and meaning of fear of crime in individuals’ lives and social relationships. 59,60

In terms of our model, each theory can be seen to emphasise a different set of factors in terms of the strength of the hypothesised links to fear of crime outcomes. Accordingly, further information on the empirical grounding of each theory is divided between the relevant sections of the report. Vulnerability is discussed further in Perceived vulnerability; social disorganisation and social integration theories focus on the links between the social and physical environments and the fear of crime and are examined primarily in the section on these links (see Built environment, social environment and fear of crime); and victimisation, likewise, is examined in terms of the link between crime and fear of crime (see Crime and fear of crime). However, before examining these links, it is necessary to clarify what exactly is being measured in fear of crime research.

The unclarity in this central question has often been remarked on in the literature. Ferraro and LaGrange61 reviewed the measures used in fear of crime research, finding that many heterogeneous and not directly comparable measures had been used and that many measures were inadequately validated. Although considerable methodological work has been carried out since, the points that they make are still substantially valid today. 49,55 Studies that describe themselves as measuring ‘fear of crime’ may measure perceived risk of crime, perceived crime rate, feelings of safety, general anxiety or episodes of worry about crime (and many particular measures are possible for each type of outcome). This makes it challenging to interpret the findings of these studies, as apparently subtle differences or ambiguities in the measures used may have a substantial and unpredictable impact on the findings.

In addition, even when the measures themselves are valid, the responses may be subject to bias. For example, social desirability bias relating to gender roles has been argued to account for almost all of the observed gender difference in affective fear outcomes. 62,63 The immediate context in which questions are asked, such as the time of day, may have an impact on responses. 64

Because of these problems, some researchers have concluded that the concept of fear of crime, as used in these studies, is largely an artefact of the methods used to measure it, and that it does not correspond to a meaningful social reality. 65 Other researchers have argued further that the confusion is not purely adventitious but reflects the fact that fear of crime is an inherently vague and inclusive phenomenon that acts as an attractor for a wide range of ill-defined worries66 or for a general ‘urban unease’ engendered by the density and loose social controls of urban environments. 67

The critical tradition in fear of crime research has developed this latter point further, exploring how fear of crime may embody and express broader attitudes towards social change and diversity, politics and policy, and ‘security’ in a deeper sense than statistically low risk. 49,59,68–73 This work has emphasised the social construction of crime, risk and fear and argued that these high-level concepts may obscure the diversity and complexity of individual experiences of fear. It has also highlighted the way in which discourse about fear of crime in research and policy has functioned to perpetuate social hierarchies, by treating crime and fear in an uncritical fashion which occludes the social inequalities that underlie fear, especially the prevalence of male violence against women,70,71 and the question of whose interests inform the accepted definitions of fear. 72

Much of this theoretical work is rooted in detailed qualitative research, the findings of which cannot be engaged with in detail here; the more detailed findings are discussed in Chapters 6 and 7, to the extent that they overlap with our reviews of empirical data. In the context of this chapter, four theoretical points are particularly relevant. First, fear of crime is not a free-floating social phenomenon (as both abstract causal modelling and the more positivist tradition of research sometimes imply), but makes sense only when situated in particular physical locations, and in individuals’ lives and personal concerns. 68,70 Perceptions of space and the physical environment at a local level may interact with the broader determinants of fear in complex and unpredictable ways. 70 Second, this research suggests that some of the links hypothesised by the model between fear and other factors, such as the environment or individual attitudes, may be better thought of as expressive relations between social meanings than as cause–effect relations between really distinct factors (see also Chapter 2, Models and theories in evidence synthesis). In particular, the simplistic dichotomy between ‘subjective’ fear and ‘objective’ crime rates may obscure the social dynamics underlying both phenomena.

Third, as suggested earlier, the impacts of fear of crime are highly unequally distributed, and these inequalities tend to closely shadow the existing power relationships within society. The experience of fear of crime as a pervasive factor in one’s day-to-day existence is one that disproportionately affects women, ethnic minorities and people living in material disadvantage. For many people, fear of crime may refer as much to the latent violence that is implicit in discriminatory social structures as to the manifest violence that is measured by crime statistics; the inescapability of such fear, and its symbolic resonance with the marginalisation and devaluation of oppressed groups, may amplify its effect on mental health and well-being. Some scholars have utilised the concept of ‘spirit injury’ to encapsulate this link between individual victimisation and structural inequality. 74,75

Fourth, this literature provides a holistic sense of how fear of crime may act as a window into high-level social structures and dynamics, by acting as an illustration of how individuals’ deep psychological need for security is played out at the social level. Ulrich Beck’s76 thesis of the ‘risk society’, which suggests that risk is the central trope of contemporary societies, that the nature of risk transcends quantitative estimations of likelihood and that people tend to seek individual solutions for social or trans-social risks, has been a productive theoretical reference point for some of this work. 68 A somewhat different approach is represented by the work of Taylor and Jamieson,66,77 who see the fear of crime as a symbolic nexus that expresses concerns about national as well as personal status. We cannot here engage in any depth with these sociological theories, but one important potential implication is that the vague and inclusive nature of fear of crime may be as much a strength as a weakness, as it may enable researchers to grasp a complex domain of social and psychological reality with a single measure. 49,78

In any case, it is clear that any attempt to synthesise the findings on fear of crime, and draw them into a more inclusive theory, will need to carefully distinguish the subconcepts that make it up and the different measures that may be used to investigate fear. Our model includes four subconcepts: perceived risk, or the individual’s estimate of how likely he or she is to become a victim of crime; emotional responses, including the whole range of affective reactions to crime or the threat of crime; perceived vulnerability and individual attitudes. These subconcepts and their inter-relations are examined briefly in the remainder of this section. The research linking fear to the other factors included in the model is covered in Causal map: relationships.

Perceived risk and emotional responses

In our model, the central axis of fear of crime is made up of the two subconcepts of perceived risk and emotional response, or one’s estimation of the likelihood of victimisation and one’s feelings of anxiety or worry about crime, which can broadly be described as the cognitive and affective components of fear respectively. 79 [The action-oriented or ‘conative’ component, which some researchers include as a third dimension,55,79 is regarded as a separate concept (‘avoidance behaviours’) in our model.] The relation between them is complex. They may not always be closely linked: it is possible to know that one’s risk of crime is high without being emotionally concerned, or conversely to be highly worried while estimating the risk as low. At the same time, there may not always be a clear subjective separation between the two for individuals. Studies that have directly compared perceived risk and the emotional dimensions of fear have been inconsistent in their findings, but have generally found that they are never perfectly, and rarely very strongly, correlated;80–85 these findings are reviewed by Chadee et al. 85 The strength of the association between perceived risk and emotional responses has been found to vary substantially depending on demographic variables such as gender and on the specific crime types investigated. 81,83,86

Perceived risk and emotion have not always been clearly separated in the research. Given the weak correlation between them, this means that interpreting the findings of such research may be problematic. For example, questions such as ‘Do you feel safe in your area at night?’, which have been used in many fear of crime studies, could be interpreted as referring either to one’s affective ‘feelings’ of safety or danger or to one’s estimate whether one actually is safe. 61 In addition, even if risk and emotion could be cleanly separated (as, for example, in Ferraro’s87 risk interpretation model), there are numerous potential indirect links, as shown in the model: emotional responses may be driven by a number of factors that also influence perceptions of risk (e.g. perceptions of the environment) but in different ways. A further complication (which is not explicitly included in the model) is that, as well as individuals’ own risk, the perceived risk of others, particularly partners or children, may have an impact on fear and behaviour. This ‘altruistic’ or ‘vicarious’ worry appears to be widespread. 88,89

Given the inconsistency of the findings on perceived risk and affective fear, it is difficult to isolate factors that may help to explain the relationship. One complex of factors that may be relevant concerns the social and moral meanings of crime. Jackson et al. 42 observe that, unlike other negative outcomes that may be feared, crime represents a deliberate attack on social norms and the social order. The ‘how dare they’ factor involved in affective reactions to crime, then, may complicate the link to perceived risk. This may explain why other harms that have a negative impact and which have been found to form the focus of fear more often than crime, such as illness, accidental injury or unemployment,90,91 do not appear to have the same affective valence, because their social meanings are different. (However, Jackson et al. 42 do not explore the broader social meanings of ‘the “how dare they” factor’, for example in relation to social inequalities; see the discussion of ‘spirit injury’ in Background.) A related point is that anticipatory emotional responses, such as those involved in fear of crime, tend to relate to a repertoire of mental imagery more than to a detailed analysis of risk. 47,55 This may drive the tendency noted earlier for fear of crime to act as a point of articulation for broader social concerns. Some researchers have argued on this basis for much more complex multidimensional quantitative instruments to capture fear, although this proposal has not been widely taken up. 55

Emotional response: further considerations

Apart from the basic distinction between perceived risk and emotional response, a number of further refinements have been suggested regarding the measurement of the latter. The first is the temporal dimension of fear, particularly the distinction between ongoing ‘dispositional’ fear and ‘transitory’ or episodic fear. The second is the distinction between normal, reasonable fear and fear that is excessive or pathological. The third, which also bears on perceived risk, has to do with the specific types of crime that are feared.

Psychological research on fear and anxiety distinguishes ‘state’ anxiety, which is episodic in nature and responsive to particular situations, from ‘trait’ anxiety, which is a relatively stable and ongoing property of individuals. 92 State and trait anxiety are generally conceived in this research as intraindividual tendencies and, as such, have been found to be not strongly correlated with fear of crime. 93 More broadly, however, the distinction can be usefully adapted in the fear of crime context to distinguish ongoing ‘dispositional’ anxiety about crime and ‘transitory’ fear of crime, which is experienced as a discrete event. 79 These distinct constructs can then be accessed using general, non-time-specific measures of anxiety (e.g. ‘How much are you afraid of . . .’) and measures that ask about the frequency of worry in a given time period respectively. Several empirical studies show that measures of frequency give substantially different results from generic measures. 49,55,94–96 Using measures of frequency tends to lead to substantially lower estimates of fear, that is, a large number of respondents say that they are anxious in general, but report no or very few specific instances of worry. In addition, the two types of measure appear to be influenced by different factors, suggesting that two distinct constructs are in play. In particular, frequency of worry has been found to be fairly well predicted by perceived risk. 60

It has been further argued that transitory fear tends to occur in response to particular situations, particularly the immediate threat of victimisation, whereas dispositional fear is driven by broader factors beyond the particular situation. Some researchers have seen this as a way to distinguish ‘formless’ or ‘expressive’ anxiety from ‘concrete’ or ‘experiential’ fear,49,65,95,96 and thus to isolate that component of expressed fear of crime which is responsive to actual crime, rather than to broader social attitudes and concerns. The argument in these papers is highly sophisticated and draws on a wide range of evidence and cannot be adequately discussed here; briefly, however, one might question whether time-specific fears necessarily correspond to concrete threats (rather than, for example, to expressive concerns about particular places or people encountered) and non-time-specific fears correspond to broad expressive factors (rather than, for example, to concrete but ongoing indicators of high crime levels). In addition, as noted earlier (see Background), researchers in the critical tradition might argue that latent structural violence – such as that involved in maintaining gender or ethnic inequalities – is as important as manifest violence in explaining fear of crime and should not be dismissed as merely ‘expressive’.

The second clarification is to distinguish normal or ‘functional’ fear, which acts as a cue to behave in an appropriately cautious manner (e.g. by locking doors), from excessive or ‘pathological’ fear, which generates anxiety sufficient to lower one’s quality of life and which is not assuaged by routine precautions. 97,98 (It should be noted that this is not the same as the distinction between ‘rational’ and ‘irrational’ fear, as the focus is on the effects of fear rather than its relation to objectively measured crime.) The distinction is important, as normal or functional fear may well not be problematic or especially negative, contrary to the presumption in much research and policy discourse that all fear of crime is a problem per se. Many people who report worry about crime also report that this worry has no effect on their quality of life. 98 Across the population as a whole, 64% of respondents to the British Crime Survey report that fear has a low impact on their quality of life, 31% report that it has a moderate impact and 5% report that it has a high impact, although the proportion reporting a high impact is considerably higher for some groups in the population. 91

The third point regards what specific types of crime are feared. Many studies have used generic measures such as ‘Is there any area near where you live, that is, within a mile, where you would be afraid to walk alone at night?’99 Other researchers have criticised the lack of specificity of such questions and have preferred to use measures that can explore the differences between different crimes. For example, the British Crime Survey100 uses a multiple measure of worry that includes worry about the following crimes: burglary from the home; mugging, car theft; theft from the car; rape; physical attack by strangers; being insulted or pestered and being attacked because of skin colour, ethnic origin or religion.

Again, like the time-specific measures discussed above, crime-specific measures have received some attention as a means to distinguish realistic fears driven by crime risk from more formless expressive anxieties. However, as discussed in Crime and fear of crime, there appears to be relatively little reason to think that such measures do reflect more accurate judgements of risk.

There is also a debate in the literature about how many different crimes need to be measured to obtain an accurate picture of fear of crime. Some studies using multiple measures have found that fear of different crimes can largely be reduced to, first, fear of crimes involving physical harm and, second, fear of crimes involving loss of property. 101 Others argue that certain crimes, particularly rape and sexual assault for women, may be particularly salient in terms of fear and create a ‘shadow’ effect across fear of crime in general. 102 It is clear that crime-specific measures cannot be straightforwardly regarded as accessing different parts or aspects of fear of crime. Some findings show that substantially greater numbers of people report worry about specific crimes than report feeling unsafe in general. 103

In addition, decisions about which types of crime to investigate with respect to fear often depend on previous methodological decisions (often linked to broader sociological or criminological perspectives). The question of what counts as ‘crime’ is highly important for the interpretation of fear55 but has received limited attention, particularly in more data-driven quantitative research. In particular, most of the fear of crime literature has tended to focus on crimes committed by strangers in public places. The fear of violent crime in private spaces, committed by known offenders or intimate partners, has not been a salient theme in most fear of crime research, although research indicates that it is widespread and serious. 70,104

One general point that emerges from this literature is that the effort to distinguish realistic from expressive fear of crime on the basis of more precise quantitative measures, and thus resolve the ‘risk–fear paradox’, has had limited success overall, although research on this point is ongoing. Moreover, the critical tradition provides cogent reasons to think that the distinction itself may be based on a misconception (see Chapter 7, Fear and rationality). In terms of the model, it might be hypothesised that, if a sharp distinction were shown to be tenable, this would lead to stronger links between crime and realistic fear on the one hand, and expressive fear and health and well-being outcomes on the other, but a less coherent model overall, in the sense that realistic and expressive fear would themselves represent two substantially distinct phenomena. The resolution of the risk–fear paradox would thus tend to block any attempt to see fear as linking crime and well-being outcomes at a community level. However, the considerations in this section suggest that the paradox has not been satisfactorily resolved. If so, there may be more scope to see risk and crime as linked to well-being outcomes in a holistic way, even if the evidence for the individual links is frequently ambiguous.

We return in Fear of crime and health and well-being to the question of how different measures of fear relate to health and well-being outcomes. To anticipate, the distinction between cognitive and affective measures may, tentatively, be reflected in the research findings, which appear to suggest that emotional responses have a greater impact on health and well-being than perceived risk; this is intuitively plausible and is reflected in the model by the more direct connections of the former to health and well-being. Both dispositional and episodic measures of fear have been found to be associated with health outcomes. However, the other distinctions explored here do not seem to have been explored with regard to their health and well-being impacts. Intuitively, it might be expected that pathological fear is more damaging than normal fear, and fear of serious violent crimes more damaging than fear of less serious crimes and crimes against property. Further research would be valuable here.

Perceived vulnerability

Vulnerability has been seen as encompassing three concepts: risk, perceived negative consequences of crime and perceived control. 60,105 For our purposes, the concept of risk is not included in vulnerability, as it has its own subconcept in the model; vulnerability can then be seen as the combination of an individual’s perceptions of the severity of the consequences of crime and his or her ability to exert control, that is, to defend him- or herself against attack. These factors may relate to an individual’s ability to defend him- or herself physically, to the social resources on which one can draw to counteract crime, or to situational factors such as the presence of other people who may be able to assist. 105,106

Vulnerability has been of interest particularly in explaining inequalities in the social distribution of fear. 103 That is, it may help to explain why certain groups, particularly women, older people and socioeconomically disadvantaged groups, have a greater fear of crime but less objective risk of victimisation; their fear is a response to their greater vulnerability and the greater impact that crime has when it does occur. However, assessing the value of vulnerability in explaining these differences is challenging. Many studies have not sought to measure vulnerability directly but have used demographic variables directly as proxies for vulnerability. 56,103 However, when self-rated vulnerability has been measured, it has been found to be a better predictor of fear than age, gender or disability. 107 The effects of perceived consequences and control have been found to be substantially less important than those of perceived risk, which is not a dimension of vulnerability by our definition. 60 In addition, some of the findings on demographic differences in fear have been questioned. As noted earlier, some researchers have questioned whether differences between men and women may not be an artefact of the measures used;62,63 others have found that the purported difference between older and younger people tends to disappear, or even reverse, when the questions focus on fear of specific crimes. 87,108,109 As a result, although it remains true that vulnerability is associated with fear of crime and has a role in explaining the genesis of fear for individuals, its value in explaining the social distribution of fear remains open to debate. For our purposes, the most important insight of vulnerability theory is that factors other than perceived risk may have a substantial impact on fear, and hence reducing risk may have a limited impact on fear.

Individual attitudes

The model also includes an ‘individual attitudes’ subconcept. To some extent, this serves to capture the broader expressive factors outlined earlier, relating, for example, to the social meanings of crime as they impact on individual judgements and emotional responses. More specifically, a range of attitudes may be relevant in explaining the fear of crime, including perceptions of police effectiveness or attitudes to policing110 and broader political and social attitudes, such as those regarding law and order or social change. 111 Again, as with vulnerability, attitudes are included in the model primarily to indicate the halo of factors that may impact on individuals’ fear, and as a reminder that the central drivers in the model do not fully account for fear of crime outcomes.

Crime and health

Crime may impact on health in a range of ways,112 which can broadly be grouped into two categories, namely direct and indirect impacts. 113 Direct impacts include physical injuries caused by violent crimes against the person and the psychological trauma that may accompany crimes involving violence or the threat of violence, or crimes such as burglary that involve intrusions into the private sphere. In the model, this is represented by the link from violent crime to physical and mental health. Indirect impacts include a wide range of negative effects that crime can have at a community level, for example by exacerbating social problems that impact on health. This distinction corresponds roughly to that between an individual perspective on crime and health and a social perspective.

The individual perspective, which focuses on the direct impacts of victimisation on individuals, has been the primary focus of the literature on crime and health. 113–115 These physical and mental health impacts on victims are often substantial and long-lasting. 116,117 ‘Domestic’ crimes, including child abuse and intimate partner violence, may have particularly serious health impacts. 118–120 However, at a community level, the health impacts are likely to be less substantial, because serious violent crime is relatively rare. In 2010–11 there were approximately 2.2 million incidents of violent crime in England and Wales,121 representing approximately 42 incidents per 1000 people per year. Dolan et al. 122 estimate the total health loss from the direct physical and psychological impacts of violent crime as being equivalent to 0.0024 quality-adjusted life-years (QALYs) per person per year. However, this may be an underestimate as it includes costs relating to victims only rather than also including those relating to witnesses or victims’ families or friends, and the figures on which it is based may underestimate certain types of crime, particularly domestic crimes. In addition, crime and the health burdens of crime are highly unequally distributed, so the health impact is likely to be substantially higher than the average for some subgroups of the population.

The community- or social-level perspective on crime and health presents a more complex picture. Violent crime has been found to be associated with a wide range of negative health status outcomes at a neighbourhood level, including all-cause mortality,123 coronary heart disease124 and preterm birth and low birthweight,125 as well as health behaviour outcomes such as lower levels of physical activity. 126,127 Exposure to community violence is also known to be associated with negative physical and psychological health, particularly for children and young people. 128,129 However, although these associations are well established, the causal pathways involved remain open to debate in many cases.

Taking a social perspective on crime and health also demands a theoretical shift, analogous to that from the individual perspective of clinical medicine to the population perspective of public health and epidemiology. Perry130 argues that approaches to the prevention of violent crime still have much to learn from public health more generally, in that approaches known to be valuable within public health have not been widely applied to questions of crime. In particular, violence prevention has tended to focus on ‘high-risk’ subgroups in the population, rather than shifting the mean of the population as a whole, and on secondary rather than primary prevention. Taking a public health-informed approach to crime would imply de-emphasising questions of why specific individuals engage in criminal activities in favour of asking why crime rates and types vary across populations and areas. Perry also argues, drawing on Farmer,131 that it would imply a greater focus on the ‘structural’ violence latent in unequal and unjust social orders, not only on the manifest violence measured by crime statistics. This point relates to the idea of ‘spirit injury’ discussed earlier (see Background), and suggests that a population-based approach to crime and health will need to take into account the indirect as well as the direct impacts of crime, and to engage critically with the concept of crime itself.

As already discussed, such a population-level approach has been widely adopted in the observational epidemiological literature on crime and health. However, it has generally not informed the development of intervention strategies. Winett’s132 review on violence in the USA as a public health problem found that, although authors tended to identify social and structural causes for violence, the interventions that they proposed targeted individual behaviour change and improved public health practice and de-emphasised social factors. Winett’s132 findings suggest a need for a more contextually informed understanding of crime and health and of the potential for interventions to ameliorate the health impacts of crime.

A further body of research has examined the links between alcohol availability or use and violence. These links may be complex: alcohol consumption may increase risky behaviour, inhibit the ability to avoid violence, increase the risk of being a victim of violence and increase the risk of violent tendencies developing in those exposed to alcohol in utero. 133 A summary of the epidemiological and criminological literature134 notes that, although problem drinking is associated with intergenerational transmission of intimate partner violence and of violence perpetration and victimisation for both men and women, and is significantly related to violent offending, the causal link between alcohol consumption and violent behaviour remains questionable. (Throughout this report, when an association or finding is referred to as significant, we mean ‘statistically significant’ unless otherwise stated.) However, the author notes that the economic literature does suggest a causal link through studies examining price changes and alcohol outlet density. This potential link is of particular interest for our purposes. Two systematic reviews of studies examining the effects of changes in alcohol outlet density have found a positive association between alcohol outlet density and increased alcohol consumption and related harms, including injuries and violence. 135,136 The causal direction of the link between high outlet density areas and alcohol consumption rates is unclear,136 although outcome evaluations of interventions do exist in addition to observational studies.

Finally, as well as the pathways from crime to health described above, there may also be pathways in the other direction, insofar as people with health problems, particularly serious mental health problems, may be at greater risk of crime;137–139 this is represented in the model by the pathway from mental health to crime.

Crime and fear of crime

The link between crime and fear of crime is conceptually obvious but empirically complex. Until recently a long-standing truism of fear-of-crime research was that objective risk of crime was poorly correlated with perceived risk and affective fear outcomes. Victimisation theories of fear of crime posit that fear is largely driven by the lived experience of victimisation. However, this theory does not appear to be strongly supported by the data. Although research does tend to show some relationship between victimisation experience and fear, it is not as strong as might be expected. 140–142 However, this may depend on the measures used. Some researchers have found that victimisation is associated with frequency of worry, as opposed to dispositional measures. 49 Repeated or multiple victimisation may also be more strongly associated with fear than one-off or occasional victimisation,117 although it is less clear that it has more severe mental health impacts.

At a broader level, it is unclear to what extent individuals’ perceptions of their own risk represent accurate estimates of the probability of victimisation (as measured by area-level crime rates or individual-level predictors of risk), or are responsive to changes in the latter. Some studies have found that most individuals are ‘pessimists’ in that their estimated risk of crime is substantially higher than their actual measured risk. 143 However, other studies with a more specific focus have found the opposite result; for example, women’s estimations of the risk of sexual assault have been found to be relatively ‘optimistic’. 144 Such results have led some researchers to speak of a ‘risk–fear paradox’. 145,146

Empirical studies of the correlation between risk and fear tend to show that there is a relation between the two, but that it is not very strong. Most studies do find that there is a statistically significant relationship, but also that it explains only a small amount of the variation in fear. Again, there is considerable controversy about which measures of fear best access the relationship. For example, some researchers hypothesise that measures which access worry about specific crimes (as, for example, the British Crime Survey measure) may be more closely related to objectively measured crime rates than those that access anxiety about crime in general. However, there does not appear to be any trend towards a stronger relationship with objective risk in studies that use the former type of measure of fear147,148 than in studies using more global measures of anxiety. 149–151

The conclusion would seem to be that the strong formulation of the ‘risk–fear paradox’, which states that there is no relationship at all, should be rejected, but that in a weaker form – namely, that the primary explanation of fear, however defined, must be sought elsewhere than in objective risk – the paradox is borne out by the observational quantitative data. Without reviewing these data systematically and dealing in depth with the methodological issues, this conclusion cannot be fully secure; nonetheless, it seems probable (see Chapter 7, Fear and rationality).

The question that then arises is which other factors may explain variation in fear, given the relatively limited importance of objective victimisation risk. One type of explanation that has received considerable attention is the various factors that may affect individuals’ estimations of risk. In the model, this is shown, first, by the cognitive heuristics and biases that influence individuals’ judgement of crime rates and of their own risk of being victimised in relation to actual (objectively measured) crime rates and risk; and, second, by the various other factors, including the perceived social and physical environment, social and media representations of crime, and interpersonal relationships, that influence fear outcomes. As we use it here, the distinction between objective and subjective risk refers simply to two types of measure. We are neutral with regard to the epistemological debate over whether divergences between them should be understood as subjective distortions and misunderstandings of an objective reality, or whether they constitute two equally valid, but potentially incompatible, ways of constructing reality. The role of environmental factors is discussed in Built environment, social environment and fear of crime; here, we focus on the role of social and media representations and risk heuristics.

Media representations and reporting of crime have often been put forward as a source of people’s understanding of crime and risk. However, the empirical data on media consumption and fear of crime are equivocal: although some studies support a link,152 others do not. 153,154 Some theorists have argued that the media do not act alone in influencing individuals’ perceptions, but in association with a repertoire of social knowledge, perceptions and imagery that derives from others’ experiences and perceptions of crime and disorder; in addition, as briefly noted earlier, these perceptions relate to people’s broader views about politics and social change. 145,155 Hence, these perceptions may also be influenced by individuals’ structural position within a community and their relationship to it,70,73,155 and by the wider structures of socioeconomic, political and cultural relations within society. 78,103,156 The role of the media and social representations, therefore, inter-relates with that of national policy and social inequalities, as well as with local-level features of individuals’ social networks and environments.

The notion of cognitive heuristics is drawn from psychological research on individual decision-making under conditions of risk and uncertainty. This research derives from the classic work of Tversky and Kahneman157 who define three commonly used heuristics: representativeness, availability, and anchoring and adjustment. Representativeness is when people judge the probability of a hypothesis (e.g. related to our focus here, ‘I may be a victim of crime’) by considering how much the situation resembles available data (previous experience, media and social representations). Availability is when individuals estimate the frequency of an event (criminal acts), or a proportion within a population (‘who commits crimes’), based on how easily an example can be brought to mind. Anchoring and adjustment is the tendency for people to base their assessments or judgements about risk on one piece of information or ‘anchor’. 158 In crime research, anchoring and adjustment can be seen at work in people’s answers to crime surveys, with perceptions of how the rates of particular crimes have altered differing between common crimes, in which the anchor tends to be personal experience, and serious violent crime, in which the anchor is more likely to be national media coverage. 159 These heuristics are likely to create substantial divergence between actual and perceived risk of crime (and may possibly account for the potential for intensive policing interventions to increase fear by raising awareness of crime160).

A further dimension in the perception of risk, which is not explicitly represented in the model, is that the social and psychological levels are not independent but may be mutually reinforcing. That is, these cognitive heuristics may act not only on an individual level but also on a social level, particularly in the choices made by mass media about when and how to report crime news. 47 Theoretical frameworks such as the social amplification of risk theory161 emphasise the role of differential communication patterns in constructing the social meanings of risk. Media representations draw on people’s everyday concerns and create feedback loops connecting them to the social and political discourses around crime, as well as policy and practice, hence ‘amplifying’ these concerns at a social level.

Built environment and crime

The theorising of links explaining the relations between the built environment and crime, and to inform the development of interventions, goes back to the 1960s. These theoretical approaches have focused on identifying features of the built and social environment that provide opportunities for or encourage criminal acts, which may be then adjusted to reduce or discourage criminality. 162 These theories, which include ‘defensible space’ and ‘broken windows’ theories,43 as well as the ‘space syntax’ approach developed by Hillier et al. ,163,164 are perhaps best summarised in terms of Crime Prevention through Environmental Design or CPTED. 162,165,166 It should be noted that, although the ‘broken windows’ theory is of clear relevance to this review, its application in policing practice has tended to emphasise more traditional approaches, with an emphasis on increasing prosecutions for minor crimes (also known as ‘zero tolerance’ policing) and on reducing ‘social disorder’ such as public drinking, with the built environment a relatively minor theme. 167,168 As such, ‘broken windows’ policing is tangential to our concerns here.

‘Design’ in the CPTED paradigm covers a wide range of permanent and semi-permanent physical and social environment features at different levels, from individual houses to whole neighbourhoods. Also important are physical features resulting from human activity, which fall into one of two clusters: signs of lack of caring or incivilities (litter, graffiti, vandalism, abandoned buildings, vacant lots) or signs of investment and involvement. 169 CPTED includes a range of strategies that focus on territoriality, surveillance, access control, activity support, image management and target hardening. 170 CPTED approaches emphasise the creation of environments that foster a strong sense of ‘ownership’ or ‘territoriality’, as this improves the chances for natural surveillance (‘eyes on the street’). Such surveillance reduces the likelihood of potential offenders committing crimes as they perceive a greater chance of being ‘caught in the act’, and is further strengthened by designing in formal surveillance measures (CCTV, security patrols, Neighbourhood Watch). Controlling access to areas aims to reduce access to potential targets for crime, thereby creating an increased sense of risk in potential offenders. A further aspect of this is target hardening, which aims to increase the efforts that offenders make in carrying out crimes by restricting access through the use of physical barriers, such as fences, gates, locks, alarms and lighting, and security patrols. In contrast, activity support includes using design features to encourage specific patterns of usage of public space. This may be to locate ‘unsafe’ activities (e.g. monetary transactions) in ‘safe’ areas (e.g. those with good surveillance) or to encourage natural surveillance opportunities through mixed usage of public spaces and the numbers of ‘eyes on the street’. Avoiding the neglect of public spaces and routine maintenance sustains a positive image of places/spaces and thus avoids them becoming ‘magnets’ for criminal activity. 170 CPTED approaches have been taken up in larger-scale policy initiatives such as Secured by Design,171 which recommends the integration of crime prevention into the design and planning process.

This approach to crime prevention views much criminality as being the result of situational opportunities, that is, that the built or social environment provides (or fails to restrict) opportunities for individuals to commit crimes. The theoretical underpinnings of this view rely heavily on classical rational choice theory, which posits that criminal behaviour (like all behaviour) results from rational, self-interested action that maximises individual utility. Thus, criminal activity occurs when the individual perceives that the potential benefits outweigh the potential costs (punishment, social stigma). This approach also draws on two further theories. Routine activity theory suggests that crime occurs when there is a temporal and spatial intersection of a motivated offender, an attractive target and a lack of capable guardianship. The theory of social disorganisation suggests that, within disorganised communities, or communities lacking in collective efficacy, the breakdown of informal social controls facilitates the emergence of criminal cultures.

One potential weakness of CPTED approaches is the potential for crime to be displaced to an adjoining area, another time, other targets, by other means or a different type of crime – the emphasis on securing a particular space or physical area from crime leads to the spatial displacement of crime to another area nearby. 170 Thus, evaluation of the effectiveness of CPTED in reducing crime or fear of crime needs to include analysis of any displacement effects of the initiatives. The approach has also been criticised for focusing on public outdoor spaces to the neglect of private indoor or domestic spaces, which may be the sites of serious abuse. 172 Similar criticisms can be levelled at the focus on fear-of-crime theories and practices that assume private or domestic spaces to be places of safety. Another criticism is the potential of CPTED approaches to produce a ‘bunker mentality’, leading to reduced social cohesion or withdrawal of certain parts of the community behind the lines of defensible space;170 more broadly, CPTED has been critiqued as an aspect of the privatisation and ‘securitisation’ of formerly public space, and the sanitisation of (what is perceived as) social disorder. 173

The empirical evidence base for CPTED and related theories is mixed. Observational studies (reviewed by Schneider and Kitchen43) have found some links between crime and variables such as density, traffic volume and street connectivity. However, many are cross-sectional with inadequate controlling for confounders, and systematic research syntheses are lacking, so the reliability of this literature is limited (although the ‘space syntax’ theorists mentioned earlier have started to develop more sophisticated methodologies for investigating these questions). The literature on the effectiveness of interventions at the level of the built environment in reducing crime is considerably more robust, and several rigorous systematic reviews have been conducted. Welsh and Farrington174 find that CCTV is effective in reducing crime although, when disaggregated by setting, a significant effect was observed only for car parks and not for any other setting (public transport, public housing, city centres). Improved street lighting and neighbourhood watch schemes are effective in reducing crime. 175,176 Multicomponent interventions using CPTED approaches are effective in reducing robberies in retail settings. 177 There is little robust data on the effectiveness of CPTED-type approaches in residential or community settings, although some findings are promising. 39 Finally, there is some evidence that Secured By Design approaches are associated with reduced property crime. 178

The relation of these theories and empirical findings to the further outcomes that we are interested in – health, well-being and fear of crime – is unclear. Few studies have sought to directly evaluate the health or well-being benefits that may accrue to the population from successful crime reduction interventions. Fear of crime outcomes are discussed in Built environment, social environment and fear of crime; the theories discussed above have tended to see the reduction of crime and fear of crime as going hand in hand, although it is unclear to what extent this is empirically supported (we return to CPTED theory in Chapter 7, Crime Prevention through Environmental Design).

Social environment and crime

A comprehensive overview of theories of crime causation and the social environment, and relevant empirical data, is beyond the scope of this study. In particular, the relations between social inequalities and crime cannot be explored here. However, one theoretical perspective that is relevant centres on the idea of ‘collective efficacy’. This perspective emphasises the influence of the social environment on rates of crime and criminal motivation. Collective efficacy, defined as the level of social cohesion (based on informal social controls) within a neighbourhood or community, along with the willingness of its members to intervene on behalf of the (perceived) common good, has been suggested as a means to explain differences in crime rates, including homicide, between neighbourhoods with similar aggregate demographic characteristics. 179 Collective efficacy is embedded in the structural and cultural characteristics of a neighbourhood, which either support or undermine the density of social ties of kinship, friendship and familiarity and the levels of participation in collective action. 179 The original theorists focused on informal social controls of the activities of children and young people, although more recent work has examined the role of collective efficacy in intimate partner violence and homicide. 180 The idea of collective efficacy also underpins community development approaches to crime reduction, within which residents are supported to establish their own organisations to address issues within the community, including crime and fear of crime. 181–183 Rather than specific actions designed to influence individual criminal behaviour, this approach ‘is a framework for action which establishes the necessary preconditions through which individual criminal motivation or behaviour can be changed through routine practice’ (p. 422). 181 Thus, the focus is on relationship building within the community, the development of institutional, structural and economic assets, collective engagement and actions and sustainable institutionalised change. 181

It should be noted that the theory that collective efficacy protects against crime has been subject to critique, on the basis that certain forms of social connectedness and collective action may in fact serve to condone and perpetuate certain kinds of crime, for example by providing social capital to offenders within criminal subcultures, or by promoting norms of inequality that provide social legitimation for crimes such as intimate partner violence. 184,185 As with many of the other links examined in this review, then, the relation between the social environment and crime is not straightforward, and quantitative variables describing social structures may be either positively or negatively associated with crime, depending on the precise measures used, the types of crime considered and the broader social context.

Broader social factors may also have an impact on crime. For example, socioeconomic inequality has been found to be associated with violent crime at a national level. 186,187 However, space precludes any detailed discussion of such broader links.

Built environment, social environment and fear of crime

The model presents the built and social environments as each acting independently on both perceived crime rates and perceived individual risk, to reflect the fact that these environmental variables may act ‘globally’ on crime and perceptions of crime, or locally on perceived individual risk (e.g. particular people or locations encountered in one’s daily routine that are perceived as dangerous). These local variables and pathways may be very specific in nature and the model does not distinguish the specific factors that may be involved. This section covers the built and social environment together because the main theoretical paradigms have tended to treat both as aspects of the same underlying issue.

The pathways linking the environment to fear of crime have often been seen as closely related to those linking the environment to crime (examined in Built environment and crime and Social environment and crime). The pathways in the case of fear may be somewhat more challenging to investigate. Environmental factors may affect fear not only cognitively, by affecting people’s perceptions of crime rates or individual risk, but also by directly affecting their emotional responses to their environment and exacerbating feelings of threat or anxiety, which may be expressed as fear of crime. In addition, as noted above, some researchers argue that fear of crime is a nebulous concept that may often reflect broader dissatisfaction with the physical and/or social environment. Hence, in some cases, it may be more appropriate to think of the perceived environment as a dimension of fear in its own right, rather than as distinct from and causally linked to fear.

Among the theories of fear of crime discussed earlier (see Fear of crime: measures and contexts), social disorder theory most clearly focuses on the role of environmental factors. The central concept here is that of ‘incivilities’ or ‘disorder’, which encompasses both problems in the built environment, such as vandalism and abandoned properties, and problems in the social environment, such as people dealing or using drugs. The most usual form of this theory is symbolic rather than rationalist, in Elchardus et al. ’s57 terms. That is, ‘disorder’ in the form of problems in the built or social environment may play the role of a symbolic indicator of social problems, economic deprivation or political neglect, and hence increase fear independently of its effect on actual crime.

Other theories posit both broader symbolic influences from the environment on fear of crime, and factors that may act rationalistically on fear by increasing actual crime risk. The former include social integration theory, which emphasises the role of local-level social networks (a dimension of the social environment in our model) in increasing fear and may be linked to the collective efficacy theory of crime (see Social environment and crime), which suggests that weak social integration and collective norms tend to exacerbate crime. However, researchers have hypothesised both positive and negative associations between social integration and fear. Some suggest that social support or social capital may act as a buffer and help to lessen fear of crime; others suggest that higher social interaction may increase fear of crime by increasing communication regarding crime and heightening the perception of risk. 188 The latter include the theories examined earlier linking certain features of the built environment, and weak social integration or collective norms, to crime (see Built environment and crime and Social environment and crime respectively), which suggest that environmental problems may form part of a complex of causal factors that increase crime.

Social disorder theories in their narrow form appear to be borne out by the observational data, which show a consistent association between perceived social and physical disorder and perceived risk or fear of crime. 149,189–194 The association of perceived disorder with affective measures of fear has been found to be weaker than its association with perceived risk,84,190 although other findings indicate that the link between the physical environment and fear of crime persists even when perceived risk is controlled for. 149 Perceived disorder has also been found to correlate more strongly with fear of property crime than with fear of personal crime; this has been hypothesised to relate to the more ‘patterned and predictable’ nature of the former. 190 These findings suggest that disorder impacts on fear primarily as an indicator of crime risk.

However, other research indicates that individuals’ perceptions of social and physical disorder are a better predictor of their fear than objectively assessed measures of these problems;195,196 in the model, the environment is separated from the perceived environment to reflect this distinction. This finding may relate to the point made earlier about expressive and causal links. That is, it suggests that the observed association between disorder and fear may result less from individuals using disorder as an indicator of risk than from the fact that the perception of environmental conditions is already laden with social attitudes and judgements,197 such that the observed relation between perceived disorder and fear is more an expressive linkage between two nexuses of social meaning than a cause–effect relationship. 198 (This type of link may also be expressed as a correlation between the perceived built environment and the perceived social environment, directly or through broader constructs such as ‘neighbourhood attachment’. 199)

Theories that posit a broader role for the environment are less clearly supported by the empirical data. Regarding the built environment, some research has found larger residential buildings to be associated with fear of crime while being more weakly associated with actual crime rates; this has been linked to ‘defensible space’ theories discussed in Built environment and crime. 200 However, other studies have not found an association between building size and fear. 201 With regard to the broader urban environment, green space has been found to be associated with better perceived safety, although this is less true in dense urban areas. 202 Non-residential land use has been found to be associated with lower feelings of safety, although the association does not survive controlling for actual crime rates. 203 It is also unclear to what extent the potential pathway from the built environment to fear of crime extends to the health and well-being consequences of the latter. One study finds that the association between perceived crime and mental health outcomes was increased by better residential environments. 204

Regarding social integration or social cohesion, as measured, for example, by people’s perceptions of their relationships with their neighbours or their participation in community organisations, the literature does not suggest a relationship with fear in either direction. Some studies find a significant association between stronger social interaction or social cohesion, at least on some measures, and a lower fear of crime. 194,205,206 However, many studies find no consistent association. 52,207–212 It should be noted that these studies are heterogeneous in the measures used for both fear of crime and social integration, and a more detailed analysis might reveal a clearer pattern. Many of these studies also present more complex analyses looking at different environmental factors. For example, one study finds that social interaction may buffer the negative effects of the built environment on fear. 210

Moreover, it is clear that there are correlations between the social environment and the built environment, although this linkage cannot be explored in detail here. Of particular interest are potential correlations between built environment factors, such as patterns of land use, and collective efficacy outcomes. 213,214

Local-level factors such as area SES have also been found to be correlated with fear, although the pathways involved are not entirely clear. 215 Moreover, national policies may have an effect on fear of crime. One study finds that more generous social welfare policies are associated with higher levels of perceived safety. 78

The potential for interventions at the level of the built environment is discussed further in Intervention pathways (logic model). Interventions at the level of the social environment may include crime reduction interventions such as Neighbourhood Watch, as well as broader programmes to promote social integration and cohesion. Police-initiated programmes to engage and educate communities about crime and risk may also be seen as social environmental interventions.

Built environment and health and well-being

There is a substantial body of literature linking factors in the built environment to health and well-being outcomes, which can be only very briefly summarised here. The main theory has been environmental stress theory, which sees well-being as determined by the balance between stressors, such as noise, traffic, poor housing, overcrowding or the threat of crime, and countervailing protective factors, such as social integration. 216,217 The causal interactions here may be complex, as some psychosocial protective factors may themselves be influenced by the environment. In addition, a number of longer pathways may operate: the pathways from the built environment to crime and fear of crime may join up with those that lead to health and well-being; and, similarly, the built environment may affect health through its impact on the social environment. These pathways are often hard to distinguish in practice, and also difficult to isolate from the impacts of socioeconomic disadvantage on individuals and communities.

Many of the associations between the environmental stressors listed above and poorer mental health outcomes are well supported by robust empirical research. 216,218 Several of these factors, particularly traffic and poor housing conditions, may also impact directly on physical health. There is some evidence that built environment interventions, particularly housing improvements, can have positive health impacts. 219

The environmental stress concept is useful for our purposes because it forms a framework for thinking about crime and fear of crime as pathways between environmental determinants and health outcomes, and substantiates the general concept that poor environments can impact on health by increasing day-to-day stress and anxiety. Thus, crime has been investigated as part of a larger complex of environmental stressors. 220 More specifically, three longer pathways may link the built environment, health and crime or fear of crime. First, what we have called ‘environmental crime’ has impacts on the built environment, although it is unlikely that these make a substantial contribution to the health impacts of the latter. Second, features in the built environment may also be linked to the social environment and social well-being in ways that are relevant to crime and fear of crime. Third, the built environment may make a difference to health behaviours, particularly outdoor physical activity. The perceived and actual qualities of the built environment, such as accessibility and aesthetic qualities, are known to be associated with physical activity;221,222 such perceptions may interact with the determinants of crime and fear of crime in the built environment.

Fear of crime and health and well-being

There are three types of potential pathway from fear of crime to health and well-being. First, the worry and anxiety induced by fear of crime (in the affective sense) may impact on mental health more broadly; in addition, the psychological distress involved in fear is a mental health outcome in its own right, hence the overlap in the model. In the other direction, poorer health may also exacerbate fear of crime and its health effects. Second, fear of crime may lead to avoidance behaviours such as limiting one’s movements outside the home, which may impact negatively on health behaviours such as physical activity and on social interaction. Third, fear of crime may impact on social well-being at a community level by decreasing social cohesion and increasing neglect, with consequent effects on residents’ well-being; it may further influence some the decision of some residents to move home, hence potentially changing neighbourhood composition in ways that may have community-level well-being impacts.

The literature on fear of crime and health is relatively small but reasonably consistent, once the measures are disaggregated. Several studies have found that affective measures of fear of crime, worry about crime or feelings of unsafety are associated with poorer mental and physical health. 208,223–228 The study by Jackson and Stafford229 is not included here because it focuses on the opposite causal pathway (see next paragraph) and because it uses the same fear of crime data as Stafford et al. 227 Several outcomes have been found to be associated with fear, including self-reported general health,223,224,226 mental health,208,223,226–228 physical functioning,227 quality of life227 and a composite index of self-reported general health and physical functioning. 225 There is also some evidence of an association between higher perceived crime, or lower perceived safety, and poorer health and well-being outcomes, although the findings here are more equivocal and complicated by the strong association of both types of outcome with SES;204,230 other studies have found no clear association. 231,232

The main pathway accounting for the effects found in these studies appears to be the first listed above, namely the psychological distress created by fear of crime and the further effects of this distress. In addition, poor health, particularly mental health, may increase perceived vulnerability and hence fear of crime. The British Crime Survey91 found that 20% of those with bad or very bad health said that fear of crime had a high impact on their quality of life, compared with 5% across the population as a whole. Qualitative studies have also found that fear of crime tends to have particularly negative impacts on those with existing mental health problems. 233 This indicates that, as well as the pathways from fear of crime to health outcomes, there are also pathways going in the opposite direction. 229 The direction of causality is difficult to establish from the quantitative data alone, but there is reason to think that the reverse pathway – from poor health to fear – may be substantial. This should be borne in mind when assessing the theoretical possibility of improving well-being by reducing fear because, if much of the association between the two outcomes is explained by this reverse pathway, the impact of fear reduction on well-being may be limited.

As noted earlier (see Emotional response: further considerations), many researchers have called for greater precision in the measurement of fear of crime, on several dimensions. With respect to the distinction between perceived risk and emotional responses, the findings cited earlier may suggest that the latter have a greater impact on health and well-being outcomes although, because of the non-systematic nature of this review, this finding should be regarded as indicative only. With respect to the distinction between dispositional fear and episodic fear, both have been found to be associated with health outcomes; of the studies cited earlier, five use non-time-specific measures208,224,226–228 and two measure frequency of worry. 223,225 With regard to the distinction between functional and pathological fear, limited data are available on health outcomes, although this would be a promising avenue of investigation. Studies have elicited respondents’ perceptions of the impact of fear on quality of life91,98 and used this to make the distinction between functional and pathological fear, but we did not locate any studies that make the distinction independently and then measure the association with quality of life or other well-being outcomes. Potential differences in the health impacts of fear by type of crime have also, to our knowledge, not been investigated.

The mental health impacts of fear of crime are probably relatively limited across the population as a whole. The utility loss has been estimated at 0.00065 of a QALY per person per year or around one-fifth of the mental health impacts of crime itself,122,234 although to our knowledge no modelling work has been carried out on the basis of more in-depth empirical studies such as that by Stafford et al. 227 to provide more accurate estimates.

The second pathway linking fear of crime to health outcomes is through the behaviours adopted to lessen the perceived risk of victimisation. This pathway may operate at a place-specific level: the percentage of respondents avoiding certain areas or neighbourhoods because of their concern about crime has been measured at 48% in a sample of the general US population99 (see Table 2.40) and 84% in a sample of women in the UK. 88 However, it may also operate on a more global level, particularly for women, as constraints on behaviour resulting from fear become internalised and normalised as an attitude of constant vigilance235–237 that intersects with the broader dynamics of gender norms and the social control of women’s behaviour. 70,88

These avoidance behaviours may be linked to health and well-being outcomes in two ways. First, avoidance behaviours may limit interpersonal interaction, leading to poorer mental health. 227 Again, there may be a feedback loop here whereby limiting social interaction also increases fear in its turn. 140 Second, concerns about neighbourhood safety may lead people to change their behaviour to limit outdoor physical activity, leading to poorer physical health. This pathway has received considerable attention in the theoretical literature. 238 There is some evidence for an association between perceived safety and physical activity, although the findings in the literature are mixed. 239,240 The association appears to depend substantially on demographic characteristics, with safety having a greater effect on physical activity for older people209 and women. 209,241 Measures of affective fear are also associated with lower physical activity. 223 The extent to which these pathways influence the further health consequences of lower physical activity is unclear. One study found that measures of avoidance behaviour (not going outdoors) because of fear are associated with several mental and physical health status outcomes in a sample of older people;242 interestingly, the associations are generally stronger for men than for women in this study. In addition, some findings suggest an association between lower perceived safety and obesity, although the effect is only borderline significant. 243 In addition, as discussed below, physical activity appears to mediate at least some of the association between fear of crime and health outcomes.

The third pathway hypothesises that fear of crime leads to decreased trust and cohesion within communities and to individual withdrawal, with a series of feedback loops at community and policy levels leading to progressive decline in the social and physical environments, with consequent well-being effects at a social level. 55,188 In the model, this can be seen as the loop that goes from fear of crime to interpersonal relationships, and from there (as part of the social environment) directly and/or via the built environment to the perceived environment and back to fear of crime. However, although this pathway has received considerable theoretical attention, it is difficult to substantiate empirically because, as noted earlier (see Built environment, social environment and fear of crime), research findings on fear of crime and the social environment are not consistent and, even when a link is well established, the direction of causality is not clear. In particular, as actual environmental problems appear to predict fear considerably less well than perceived problems, the existence of strong feedback loops between fear and actual environmental change remains to be proven.

The relative importance of the pathways is difficult to quantify. Ross223 investigates psychological distress and physical activity as mediators of the effects of fear on general health. Stafford et al. 227 investigate physical activity and social interaction as mediators for mental and physical health outcomes. Both find that a substantial amount of the association is explained by these pathways, but not all. Possible measurement error aside, this may suggest a role for broader effects such as those hypothesised by the third pathway above, or it may reflect pathways from other variables not included in these studies.

Initial inclusion criteria

The following inclusion criteria were applied sequentially (i.e. studies had to meet all criteria to be included):

1. Does the study report data on crime, safety and/or the fear of crime or evaluate an intervention intended to reduce crime and/or the fear of crime? (Note: crime or fear of crime need not be the sole or primary focus of the study. Violence and antisocial behaviour are considered to be crime outcomes, but illegal drug use per se is not. Conduct problems, behaviour problems or aggression are not crime outcomes and studies on these without crime or fear of crime data were excluded.)

2. Does the study report data on at least one of the following: fear of crime (including perceived safety or risk, worry or anxiety about crime, or fear of specific crimes or of crime in general, but not including perceived crime rates or perceptions of crime, e.g. the extent to which crime is ‘a problem’); mental health; physical health; health behaviours (e.g. physical activity), or social well-being (e.g. social interaction)? (Note: violence is considered to be a health outcome for this criterion, as is illegal drug use; however, crime or antisocial behaviour, if not further specified, are not. Educational attainment is not a health outcome. This criterion should be interpreted strictly for quantitative studies, but broadly for qualitative studies.)

3. Does the study report data on variables relating to the built environment, on an intervention that involves some change to the built environment or on perceptions of the environment? (For qualitative studies this should be interpreted at abstract stage to include factors relating to the local-level social environment, e.g. social cohesion, collective efficacy.)

4. Is the study a primary research study or a systematic review? (A systematic review is one that clearly reports its search strategies and inclusion criteria. Systematic reviews were retained for reference searching but were not included in their own right.)

5. Was the study conducted in a country that is a current member of the Organisation for Economic Co-operation and Development (OECD)? These are Australia, Austria, Belgium, Canada, Chile, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israel, Italy, Japan, the Republic of Korea, Luxembourg, Mexico, the Netherlands, New Zealand, Norway, Poland, Portugal, Slovakia, Slovenia, Spain, Sweden, Switzerland, Turkey, the UK and the USA.

6. Is the study either a qualitative study (e.g. one that involves in-depth or semistructured interviews, focus groups, ethnography) or a study that evaluates an intervention and presents quantitative outcome data that were measured both before and after the intervention and/or for both an intervention and a comparison group? To be included here the study had to sample both an intervention and a comparison group; studies with data from one group and one time point that compared these outcomes with data from other sources (e.g. the British Crime Survey) were excluded.

7. (For qualitative studies only) Was the study conducted in the UK?

We attempted to retrieve the full texts of all studies that met the inclusion criteria at abstract stage. When the full text was located it was rescreened using the same criteria. Of the full-text studies, 79% were screened independently by two reviewers (n = 498), with differences resolved by discussion. Agreement on inclusion prior to reconciliation was 96.0%, with Cohen’s kappa = 0.92. The remainder were screened by one reviewer alone.

After this initial screening process several further exclusion criteria were adopted to focus the reviews further. These criteria differed between the review of effectiveness and the review of qualitative evidence.

Categorisation of the effectiveness studies

As a wide range of interventions are included in the effectiveness review, the findings are presented by intervention category. The studies have been divided into seven categories of interventions. Of these, categories 1–4 include interventions whose main aim was to reduce or prevent crime and/or the fear of crime, whereas categories 5–7 include interventions that did not have such an aim but which measured relevant outcomes in the evaluation:

1. interventions with a main focus on improving the security of homes

2. the installation and/or improvement of street lighting

3. the installation of CCTV systems

4. multicomponent interventions for crime prevention with a main focus on public areas, several of which are based on the principles of CPTED

5. housing improvement and/or relocation schemes

6. large-scale multicomponent area-based regeneration initiatives

7. smaller-scale environmental improvements in public areas.

In the following sections the evidence for each category of intervention is discussed in detail. For crime-focused interventions (categories 1–4), all relevant outcomes are included in the synthesis, including fear of crime, health status and health behaviour outcomes, and social outcomes (e.g. social interaction). However, relatively few studies measured health and social outcomes: only two studies (both in category 1) measured health outcomes and only six (in categories 1 and 4) measured social outcomes; no studies in categories 2 or 3 measured any health or social outcomes (i.e. fear of crime outcomes were the only ones within the scope of the review for these intervention types). For non-crime-focused interventions (categories 5–7), only fear of crime outcomes have been synthesised as robust systematic reviews already exist on the health outcomes of these types of interventions. 219,308 (For more detail on the outcomes included under fear of crime, see Appendix 8.)

It should be recalled here that interventions not involving a substantial change to the physical built environment were excluded (criterion 3). In particular, this excluded policing and judicial interventions such as increased police patrols or changed models of policing practice such as ‘hot spots’ policing; changes to judicial or sentencing practice, such as drug courts; and interventions working with offenders, for example in the probation system. Interventions aiming to target the broader social determinants of crime without an environmental dimension, such as alcohol taxation or educational interventions, were also excluded. (Some of the included interventions may have included components of these types, but interventions that focused only on these strategies were excluded.)

A summary of the direction of effect and significance of the findings for each study and outcome category can be found in Appendix 6. For the synthesis in this section, as described in Chapter 4 (see Data synthesis), overall effect sizes have been summarised using median change scores. The statistical significance of the findings in each case is based only on the analyses presented by the study authors (we have not carried out further tests for statistical significance). Within each section, findings from controlled studies are presented first, followed by those from uncontrolled studies. The studies are identified by the author names in the left-hand column of Table 1 rather than by references to specific reports, as in many cases the information reported is drawn from several distinct linked reports.

1. Home security interventions

This category includes five studies, two controlled and three uncontrolled before-and-after studies. All studies in this category were conducted in the UK. Two studies received an A quality rating (Allatt, Brownsell) and the remaining three a C rating (Halpern, Matthews a, Matthews b). Health outcomes are available for two of the studies (Brownsell, Halpern) and social outcomes for one (Halpern); the other three assessed fear of crime outcomes only.

2. Street lighting

The single largest category of interventions in the review consists of street lighting improvements (n = 16 studies). Four studies use controlled before-and-after designs (Atkins, Painter d, Painter e, Painter f); the remainder are uncontrolled. All but one of these studies were conducted in the UK, most between the late 1980s and late 1990s. Only fear of crime outcomes are measured; no study assessed health or social outcomes. Four studies are rated A for quality (Atkins, Painter d, Painter e, Painter f), two B (Bainbridge, Painter c) and the remainder C (Barr, Burden, Davidson, Herbert, Knight, Painter a, Painter b, Payne, Vamplew, Vrij).

Two studies (Knight, Vrij) investigate the short-term effects on fear of improvements to lighting in a small well-defined area, with follow-up times of a few weeks. The other 14 studies are more ecological in nature, seeking to investigate the longer-term effects of improving lighting over areas ranging from a few streets to whole neighbourhoods. Of the latter, 10 are uncontrolled before-and-after studies whereas four use control groups, generally matched for demographics and area type. Because of the seasonal nature of the hypothesised intervention effect, follow-up times are generally either 1–3 months or around 1 year. Most of these studies involved improvements to lighting rather than its installation in areas that were previously unlit, and refer to the BS 5489 standard as the minimum standard of illuminance achieved by the intervention; in some cases there were further changes such as the replacement of low-pressure yellow bulbs with high-pressure white ones. A wide range of fear-related outcomes were measured, with several studies including 10 or 20 distinct outcomes. A feature of this category of studies is the use of measures of avoidance behaviour (e.g. ‘Are there areas you avoid at night because of crime?’) in conjunction with measures of affective fear (on these measures, see Appendix 8).

3. Closed-circuit television

Six studies, three controlled and three uncontrolled before-and-after studies, evaluate the effectiveness of closed-circuit television (CCTV). Only fear outcomes are measured by these studies, with none including health or social outcomes. All but one (Musheno) of the studies were conducted in the UK, and all six received a low (C) quality rating. Five of the studies evaluate the installation of local CCTV systems, with four looking at one system each (Brown, Ditton, Squires a, Squires b) and one (Gill) evaluating nine separate systems as part of a national evaluation for the Home Office; the sixth (Musheno) evaluates a slightly different CCTV programme.

4. Multicomponent crime prevention interventions

This category (nine studies) includes a range of complex, multicomponent interventions with a focus on reducing crime through changes to the built environment. Intervention components often include those examined in the previous sections (security improvements, lighting improvements and installation of CCTV) as well as more general environmental improvements (such as landscaping or graffiti removal). In addition, many include non-built-environment components such as changes to policing practice (e.g. community-oriented policing) and/or social programmes (e.g. drug treatment, employment initiatives). Several of these interventions (Donnelly, Felson, Kaplan a, Kaplan b) were explicitly based on CPTED theory (on CPTED more generally, see Chapter 3, Built environment and crime, and Chapter 7, Crime Prevention through Environmental Design), and even those that do not employ the technical vocabulary of CPTED appear to use similar ideas. Thus, they emphasise the role of features such as natural surveillance, access control and territoriality, as part of a broad-based approach to addressing crime and the fear of crime through the built environment. In terms of study design, one was a RCT, five used controlled before-and-after designs and three were uncontrolled. However, many of the controlled studies used non-equivalent control groups. In terms of study quality, all were rated C except one that was rated B (Mazerolle) and one that was rated A (Arthur Young). All but one (Webb) of the evaluations in this category were carried out in the USA. The main outcomes of interest to this review are fear-related although several studies also measure social outcomes; none measure health outcomes.

5. Housing improvement

Seven studies evaluated the effect of housing improvement programmes, or the construction of new housing, on fear of crime outcomes (Barnes, Blackman, Critchley, Foster, Glasgow Centre for Population Health, Nair, Petticrew). (It should be noted that pure relocation interventions, such as the Moving to Opportunity intervention in the USA,309 were excluded from this review as they do not directly involve changing the environment.) All of these studies were conducted in the UK. Three studies included renovation of existing housing (Blackman, Foster, Nair), one focused on the provision of new housing (Petticrew) and three (Barnes, Critchley, Glasgow Centre for Population Health) included elements of both. In terms of design, five used a formally controlled before-and-after design and two an uncontrolled design; however, only one of the controlled studies (Critchley) clearly reported comparative data for the outcomes of interest to this review, with the others reporting only within-group findings or incomplete data. In terms of quality, two studies were rated A (Critchley, Petticrew) and the remaining five C (Barnes, Blackman, Foster, Glasgow Centre for Population Health, Nair).

6. Regeneration and area-based initiatives

Evaluations of two national regeneration initiatives in the UK (New Deal for Communities and the Single Regeneration Budget) were included in the review (Beatty, Rhodes). Both of these interventions were very wide-ranging programmes including many different local initiatives: some focused on crime prevention but the majority targeted other issues such as employment or housing or community development. Both evaluations, one a controlled study (Beatty) and one an uncontrolled before-and-after study (Rhodes), were conducted on timescales of several years and measured a wide variety of outcomes. One study was rated A (Beatty) and one C (Rhodes).

7. Improvements to public areas

Two studies, one controlled and one uncontrolled, looked at environmental improvements to public areas that were not primarily intended to prevent crime. Both studies were rated low quality (C).

Summary of findings

In general, the evidence base covered in the review of effectiveness has considerable limitations. As noted earlier (see Characteristics of the studies and quality assessment), study quality is generally poor, with very few studies using a robust design with adequately matched control groups. The analyses conducted by study authors were often rudimentary, with many not conducting any tests of statistical significance: of the 47 included studies, almost half (n = 22) did not report testing for significance on the relevant findings at all (Arthur Young, Bainbridge, Barr, Brown, Burden, Critchley, Ditton, Felson, Herbert, Kaplan a, Matthews a, Matthews b, Musheno, Nair, Painter a, Painter b, Painter c, Payne, Squires a, Squires b, Vamplew, Webb); only 10 analysed differences over time between an intervention group and a comparison group, and the analysis method used was questionable even in some of these (Atkins, Beatty, Brownsell, Cohen, Foster, Fowler, Gill, Halpern, Mazerolle, Painter f). Another concern about the effectiveness evidence is the high degree of heterogeneity in the outcome measures aggregated as ‘fear of crime’ in this report; this issue is discussed further in Appendix 8. There is also considerable heterogeneity within several categories in terms of intervention, population and setting. Finally, to the extent that the findings can be synthesised, they frequently present conflicting results, with no clear direction of effect.

The findings thus do not support any strong conclusions on the effectiveness of interventions. To the extent that tentative conclusions can be drawn, the evidence suggests that most of the interventions considered are not effective in reducing fear of crime. When positive findings appear to emerge from single-group studies, they are rarely confirmed by more robust designs with matched comparison groups. The interventions that are most promising appear to be home security interventions (category 1), at least in certain contexts, and improvements to public areas (category 7). CCTV (category 3) appears to be the least promising of the interventions, with consistent evidence of ineffectiveness for fear outcomes. Findings in the other four categories are more mixed, with some positive findings, but they do not provide strong evidence of effectiveness for fear outcomes. Evidence from one study298,299 indicates that there may be a possibility of adverse effects for some interventions.

One factor that may explain some of the variation in intervention effectiveness is the extent to which interventions are primarily conceived, and presented to the target audience, as efforts to reduce crime. Interventions that are explicitly directed towards crime and have strong police involvement – particularly CCTV (category 3) and most of the multicomponent crime prevention interventions (category 4) – appear to be largely ineffective for fear outcomes. By contrast, most of the more promising interventions appear not to have been presented or conceived as primarily crime reduction initiatives. This finding tends to support the concern, already noted in the review of theories [see Chapter 3, Intervention pathways (logic model)], that crime reduction interventions may have an adverse ‘iatrogenic’ (or ‘alytogenic’) effect on fear, by raising awareness of crime among the targeted population. Although our review did not find any evidence of significantly increased fear as a result of such interventions, this effect may explain why little evidence of decreased fear was observed.

In terms of broader health and social outcomes, limited evidence is available [note that data on these were extracted only from crime-focused interventions, that is, categories 1–4 (home security, street lighting, CCTV and multicomponent crime prevention interventions)]. Several of the studies in category 4 measure social cohesion outcomes such as perceived trust or friendliness; the findings, although inconclusive, suggest that these interventions are generally ineffective for such outcomes. The study by Brownsell et al. provides some promising evidence of improvements in health status outcomes, although most are not significant (Brownsell). However, given that the population and intervention investigated in this study (telecare for older people in sheltered housing) are rather unusual with respect to the review as a whole, this finding is arguably of limited relevance. Only one study (Halpern) attempted to measure health status outcomes of a crime reduction intervention in the general population. Although this study is small and methodologically not very robust, its findings are positive, with significant improvements in mental health status and promising findings for social cohesion outcomes. Other than this, however, there is a lack of evidence on whether environmental crime reduction interventions can improve population health outcomes.

Study funding

The question of study funding is particularly relevant with respect to the studies of street lighting (category 2). Many of these studies are funded by the lighting industry. Of the 16 studies, nine explicitly report being funded wholly or partially by lighting companies (Barr, Davidson, Herbert, Painter a, Painter b, Painter c, Painter d, Painter e, Painter f), with the author of a further study being affiliated to a lighting company (Knight). In three cases funding is unclear (Bainbridge, Burden, Vrij), and only three studies appear to be wholly independently funded (Atkins, Payne, Vamplew). It is noticeable that the three that are independently funded show decidedly less improvement than the average across the studies, with two studies finding no change in fear outcomes (Atkins, Vamplew) and one a slight trend towards increased fear (Payne). Although it obviously cannot be concluded that the industry-funded studies are problematic per se, and included here are several of the more robust studies in this category, the question of funding should be borne in mind in interpreting the findings.

Population subgroups and inequalities; fear of crime outcome measures

Findings on the targeting of interventions and differences between subgroups are set out in Appendix 7. Findings on differences in effectiveness between age groups are mixed. There appears to be little difference in effectiveness between men and women. There is very little data on ethnicity and none on SES.

Findings on the outcome measures used for fear of crime are set out in Appendix 8. Many different measures are used for fear of crime. The heterogeneity of outcome measures is a potentially serious limitation of the effectiveness results. Of particular concern are those studies that measure only cognitive outcomes (i.e. perceived safety or risk), as these may have limited impact on affective outcomes or well-being more broadly (see Fear of crime and health and well-being and Chapter 7, Fear and rationality). Several of the studies with substantial positive effects (Cohen, Felson, Palmer, Webb) actually only measure such cognitive outcomes, and it is open to question whether positive effects on these outcomes have any impact on well-being.

Limitations of the review

The review of effectiveness was carried out using full systematic review methodology, with highly sensitive searches, using no date or language restrictions and using robust procedures for screening, quality assessment and data extraction. However, there are some limitations to the review.

The scope of the interventions included in the review was restricted, and interventions without a substantial environmental component were excluded. This means that we do not know to what extent interventions such as policing may be able to reduce fear of crime.

Only a restricted range of outcomes were extracted from the included studies. In particular, we did not look at crime outcomes, for reasons of practicability (as extracting crime outcome data not only would have been a substantial task in its own right but also would have complicated the methodological data extraction and quality assessment). As a result, we cannot say how the findings on fear of crime relate to the effectiveness of the interventions in reducing crime (which was in almost all cases their primary aim). However, as noted earlier (see Crime and fear of crime), observational research tends to find that crime rates and fear of crime are not closely associated. As set out in the theory review, then, the pathway that runs from interventions to crime reduction to reductions in fear of crime may not necessarily be the most promising one. Hence, our inability to substantiate this particular pathway in detail may not be as serious a limitation as it seems. This issue is discussed further in Chapter 7 (see Broader implications for interventions).

Although quality assessment was carried out and is fully reported in this chapter, and was used to inform the interpretation of the study findings, lower-quality studies were not excluded and inclusion criteria relating to study design were not stringent. As a result, many of the included studies are not reliable, and the findings tend to show that positive results in lower-quality studies are not replicated in higher-quality ones. Hence, the findings should be regarded as indicative only. This is true throughout the review, especially for those categories of intervention in which few or no high-quality studies were available (particularly categories 3, 4 and 7).

Security measures

Participants’ views of physical security measures, such as locks, fencing or other physical access control measures and secure entry systems, are mixed. Several participants report feeling safer as a result of having locks or entry systems (Farrall, Whitley) and some participants report increased fear as a result of the perceived inadequacy of physical security measures (Farrall).

However, negative perceptions of security measures are also frequently expressed. One participant argues that visible security measures would increase risk: ‘Of course I haven’t put a burglar alarm in. I can’t afford it. Anyway if I did something like that they’d only think I’d got something worth nicking’ (participant, female, 30s, Squires).

Further, there is a widespread perception that visible security measures increase fear. One study (Nelson) that focuses on security shutters for shop windows found that several participants saw them as increasing fear – and as creating an unpleasant atmosphere more generally in areas where they are widely used, described by participants as ‘creepy’ and ‘dead’ – and two other studies (Taylor, Waters a) found similar perceptions of shutters and security gates. The metaphor of a fortress or prison is frequently used by participants to express the unwelcoming and depressing effect of excessive visible physical security in the home (Farrall, Pain a, Waters a, Whitley).

This perception of security measures as increasing fear is also linked to a broader theme of anger at the need to install them. Participants link the need for physical security measures to a broader anger at the breakdown of social order that crime represents (see further Crime as a social symptom on the social meaning of crime). This particular theme elicits reactions of anger that are not as noticeable with respect to other environmental crime prevention strategies such as lighting or CCTV (Farrall, Koskela).

I’ve got locks on my windows, I’ve got the place like a blooming fortress you know? And I don’t really see that one should have to live like that, I mean I’m vexed with society, for having to live like that.

Participant, Farrall

Street lighting

Participants who are asked for their views on how to reduce crime or the fear of crime frequently mention improved lighting (Alexander, Cozens, Crime Concern a, Crime Concern b, Valentine a, Waters a). Participants in seven studies mention that dark or poorly lit locations are particularly associated with fear of crime (Crime Concern a, Crime Concern b, Koskela, Pain b, Pain d, Valentine a, Waters a), and in a further five studies that their fear is greater at night (Farrall, Innes, Little, Nelson, Taylor). Darkness is mentioned as a cause of fear particularly by women and older people, but is expressed across all study populations:

We have a lane right next to us which I avoid when it’s getting dark. Anywhere on my own, either by myself or even with my husband, if there’s not much lighting there I feel very nervous.

Participant, female, 20s, Koskela

You feel a little bit more aware at night and the kind of situations you put yourself in.

Participant, Waters a

There appear to be two inter-related ways in which lighting is seen to impact on fear (which are not strictly separate and may be linked by the idea of cognitive mastery over the environment; see the discussion in Fear and knowledge of the physical environment). The first is that it increases visibility and minimises places where attackers can hide (Koskela, Valentine a, Waters a) (see also Environmental visibility):

I mean the lighting is totally inadequate for ladies walking through on their own. [. . .] you couldn’t actually see anything which was worrying because you expect someone to leap out of a dark corner. [. . .] if anyone was set on attacking you there are plenty of places they could hide.

Participant, female, Valentine a

The second is that good lighting makes locations feel more pleasant and welcoming. This may be linked to other physical determinants of fear such as cleanliness and building layout. In particular, the threatening aspect of poorly lit locations is strongly associated with their being isolated, with few other people around (Crime Concern b, Koskela, Valentine a, Waters a) (see also Presence of others):

This part of the station is cleaner, brighter and there’s no litter. It makes you feel more comfortable, it looks as though someone is responsible and it is looked after.

Participant, male, Crime Concern b

It’s a perception thing isn’t it? You might not actually be any safer but you feel safer because you think there must be someone in there if there’s lights on.

Participant, Waters a

Corresponding to these two mechanisms, the link between improved lighting and reduced fear may fail to work for two reasons. First, even where lighting is good, visibility may remain poor because of the layout of the built environment (Crime Concern a). Second, in many cases, broader contextual factors may outweigh the effect of lighting in determining the emotional resonance of particular locations:

I mean when I was a child we lived in the country and it was all dark lanes with no lights, but we never felt afraid.

Participant, Valentine a

[W]hen I was in Pakistan, I went out to the shops at night, and walked round to visit people, and yet I wouldn’t do that here, which is silly really because statistically it’s far more dangerous in Pakistan than in England.

Participant, Watson

One study focusing on the perceived relationship between lighting and crime found considerable scepticism about a link (Pain d), and in one further study participants who recommended improved lighting also expressed serious doubt about its likely effectiveness in reducing fear (Koskela). These more sceptical attitudes may be related to the more complex views of lighting mentioned above, and suggest that other factors associated with poor lighting, such as the absence of other people, are more important as determinants of fear:

In some areas we do need more lighting.

Q: Would that make you feel safer?

Not really because I think it’s going to happen anyway. Alleyways where there’s no housing, definitely I wouldn’t want to walk down, even with lights.

Participant, female, 20s, Koskela

Several specific aspects of lighting are also identified as important by participants, including colour and brightness (Crime Concern a, Pain b, Pain d, Waters a), consistent levels of illumination (Crime Concern a, Waters a) and maintenance (Pain d, Valentine a).

Apart from the direct link between darkness and fear as it applies to both the spatial distribution of light (i.e. poorly lit locations) and its temporal distribution (night vs. day), there are also indirect links to do with the way that spaces are used socially. That is, locations that are unlit are also seen as threatening because of the presence of young people or drug users, and part of the fear experienced after dark is to do with the different social norms that apply late at night, for example the presence in city centres of groups of drunk people (Innes; Valentine a) (see also Alcohol and drug users as threats).

Closed-circuit television

Relatively few data are reported in the included studies on perceptions of CCTV. The views that are reported are mixed. Several participants express the belief that CCTV is a deterrent to crime (Crime Concern b, Squires, Waters a) or suggest CCTV as a crime prevention measure (Alexander, Cozens). A few participants express strong support for CCTV (Crime Concern b, Cozens, Pain b). However, only a few participants explicitly say that CCTV reduces their fear, and one explicitly denies this: ‘CCTV makes you feel that if anything happened, they might be caught. But I don’t feel it makes me feel safer’ (participant, Crime Concern b).

There is considerable scepticism about the effectiveness of CCTV in reducing crime. Several participants express doubt whether cameras are monitored (Crime Concern b, Squires), or observe that CCTV cannot directly, of itself, impact on crime (Crime Concern b, Nelson, Pain b, Squires): ‘It’s just a video camera – it just records you getting beat up!’ (participant, Pain b).

The most detailed data on CCTV come from an evaluation of the installation of CCTV (Squires). Several participants in this study thought that crime and fear had been reduced as a result, but many thought that crime had just been displaced elsewhere, or even that young people involved in antisocial behaviour were ‘playing up’ for the cameras. Moreover, for several participants in this study, CCTV is an inadequate substitute for investment in infrastructure and policing, and exacerbates the sense that disadvantaged areas are not a priority for policy-makers:

What’s the point of putting up **** cameras. The estate is just crap. They should spend their **** money improving this dump . . . there should be more things for the kids, decent places for them to go, youth clubs and the like.

Participant, female, 30s, Squires

Why are they doing this? It is the housing that needs the money spent on it.

Participant, female, 20s, Squires

Environmental visibility

An important dimension of the built environment that relates to fear is environmental visibility, both in the sense that one can be seen by other people and in the sense that one can see them. Both aspects of visibility relate to lighting (see earlier section on street lighting) and to the layout of the built environment. The first aspect of visibility relates to what in CPTED theory is called ‘natural surveillance’ – the sense that the environment allows others to see what is going on. In this sense the importance of visibility in reducing fear is made explicit by several participants (Cozens, Crime Concern b, Koskela) and seems to be implicit in much of the data concerning the presence of others (see Presence of others). Non-residential environments are sometimes seen as more threatening because of the lack of visibility in this sense (Koskela).

The second aspect of visibility – the sense that the surrounding environment is visible and contains no hidden areas – is also very important. As with many of the qualitative themes, this has both a practical and a more symbolic side. On a practical level, visibility reduces fear by ensuring that potential attackers can be seen. Environments that present obstructions to visibility are consistently seen as threatening. Particularly problematic in this regard are vegetation, obstructions such as pillars and hiding places such as side alleys (Crime Concern a, Crime Concern b, Koskela, Valentine a, Waters a).

On a more symbolic level, visibility contributes to a sense of ‘openness’ and resonates with a broader sense of the readability of the environment, both physical and social (Burgess, Farrall, Waters) (see the discussion in Fear and knowledge of the physical environment and Fear and knowledge of the social environment):

I think the key to it all is sight, being able to see all around you, not just in front [. . .]

So you need to be able to get away?

Yes, you need the space.

Participant, female, 45–65 years, Burgess

As this quote indicates, visibility in this sense is also associated with freedom of movement. Areas with restricted visibility often also create a sense of being ‘trapped’ (Crime Concern a, Waters a).

Cleanliness and neglect

Dirt, decay, graffiti and evidence of physical neglect – what is known in the criminology literature as ‘physical incivilities’ – are mentioned as drivers of fear by participants in 11 studies (Cozens, Crime Concern a, Crime Concern b, Farrall, Goodey, Innes, Little, Parry, Valentine a, Waters a, Waters b).

To a substantial extent, this association appears to be driven by a number of more specific considerations that link to other determinants of fear. First, environments that appear cared for are associated with the presence of others and the potential availability of help, which help to protect against fear (Crime Concern b, Waters a) (see Presence of others).

Second, neglected physical environments are an indicator of problems in the social environment and of the breakdown of social cohesion (on which see Social cohesion and interpersonal networks and Crime as a social symptom respectively). Evidence that residents and others are not committed to the maintenance of the physical environment is taken to entail a lack of commitment to broader social norms. Similarly, in institutional settings [such as in Waters’ study of universities (Waters a)], care for the physical environment indicates a broader commitment to the welfare of the people who use that environment:

The house the other side of me [. . .] the people who own the houses they were from, they were living away and filling them up with unsatisfactory people really because they don’t care a fig. They don’t look after the place, it’s in a terrible state.

Participant, Waters b

Third, a personal environment that is pleasant and cared for contributes to a general sense of well-being, which in turn creates a sense of safety (Farrall, Parry, Whitley):

Well you need all your repairs done, you need all your repairs done, you need everything doing to your house so it looks nice, you need your walls and fences putting up so it looks decent when people are walking around there

You need to feel safe.

Participants, Parry

Fourth, there is an association between the appearance of neglect, specific styles of planning and architecture, the perception of material and social disadvantage and perceived risk of crime. These form a nexus of meaning that amplifies the link between neglect and fear of crime, by associating it with a set of broader perceptions of the character of specific locations or neighbourhoods (Farrall, Valentine a, Waters a):

[B]ut you get into the sort of grotty areas and you see all the vandalism and the litter and you just realise what kind of area you are in. If I was on my own I think I would be very nervous, you know in those sort of areas.

Participant, female, Valentine a

[T]he pre-fabricated and the tower block [. . .] I don’t like that it reminds me of a bad neighbourhood, a council, derelict place, but after seeing the rest of the place it really felt more community and a much safer place, a much nicer place.

Participant, Waters a

Finally, for children specifically, the presence of physical hazards such as dog faeces and broken glass, and particularly the presence of debris from drug use, contribute to a more general feeling of unsafety, in which the danger of accidental physical injury and the risks of a high-crime environment are not clearly distinguished (Goodey, Trayers, Turner).

Presence of others

As already mentioned, an important determinant of fear is the absence of other people. Places where there are other people around are experienced as less threatening (Burgess, Crime Concern a, Crime Concern b, Koskela, Pain b, Waters a, Valentine a). ‘When I come off the train late, most important is seeing people around, for me that’s most important in making me feel safe’ (participant, Crime Concern b, female).

This perception is expressed by participants across all population groups. However, women in particular mention the threatening aspect of isolated places more often, and link it specifically to scenarios of attack (Burgess, Crime Concern a, Koskela, Valentine a):

Well, I don’t often see people when I’ve been in the forest. [. . .] If I did meet a big man there, I mean it would be so easy to do things to me because there’s no one around. I think that’s very frightening.

Participant, female, 16 years, Burgess

However, the association between isolated places and fear is not universal and may be specific to residents of urban areas. In one study, residents of rural areas expressed the opposite perception, that crowded environments increase fear (Crime Concern a).

Several participants make the point that not all people are equally reassuring. Studies of particular settings – public transport (Crime Concern b) and university campuses (Waters a) – emphasise the reassuring effect of a visible staff presence in such settings. More generally, participants recognise that their fear of crime is by its nature a fear of other people, especially those who are ‘odd’ or not ‘normal’. In many cases the theme of the presence of other people is implicitly coded with expectations about how the latter are likely to behave:

[B]ut I wouldn’t cross it after dark . . . the trees, odd people hanging about and long distances where there’s no one else about.

Participant, male, Crime Concern b

I mean what really makes me feel safe is just seeing other people, other normal people around. I mean there’s safety in numbers whether they’re strangers or not.

Participant, female, Valentine a

Perceptions of specific groups as threatening, particularly young people in groups and alcohol or drug users, are discussed further in Young people as threat and Alcohol and drug users as threat respectively. However, in some cases, even people who may be perceived as a threat in other contexts are preferable to the absence of others (Crime Concern a).

Area knowledge

A consistent finding in the qualitative studies is that areas with which one is familiar are perceived as less threatening than areas that are not known well (Crime Concern a, Koskela, Valentine a):

How true it is that one often feels safer in your local area . . . I just feel safer because it’s my local area and I know what happens there and I feel more confident.

Participant, female, Crime Concern a

I mean I feel safer walking around this sort of area than I do in the town centre, but you get people who don’t come from Whitley, they’d rather walk round the town centre at night than Whitley. I’ve always lived here. I know the area pretty well.

Participant, female, Valentine a

The link between familiarity and reduced fear is often direct in that knowledge of an environment makes it feel less threatening regardless of its characteristics. In several cases, participants recognise that those unfamiliar with their area might see some aspects of it as cues for fear, but because of their familiarity with it they do not see these aspects as threatening (Crime Concern b, Watson): ‘I think it’s all right round here, I mean you see gangs of kids but they’re only young and it doesn’t bother me because it’s familiar, I mean I’ve always lived round here’ (participant, Watson).

Along similar lines, two studies find that clear and accurate signage and information, for example in public transport settings, can help to reduce fear (Crime Concern b, Waters a).

Part of the fear-reducing effect of familiarity has to do with perceptions of risk. Bannister’s study provides a particularly clear illustration of this. Participants were asked to mark on a map areas seen as unsafe at night. Most participants marked as unsafe only areas that they did not know well, such that a large proportion of residential areas were seen as unsafe by at least one participant, but none by all of the participants (the only areas seen as unsafe at night by all participants were parkland and a shopping centre). Thus, crime is generally perceived to occur ‘somewhere else’:

That place is horrible, it’s full of junkies.

[immediately] No, it’s not, that’s where I live.

Oh, I’m sorry, it’s just that I’ve not been there recently.

Participants, Bannister

However, there are also more complex links between familiarity and reduced fear, which relate to a number of other themes. Of particular importance are interpersonal relationships and social cohesion at area level, which appear to be highly important determinants of fear (see Social cohesion and interpersonal networks); much of the reassuring effect of familiarity appears to relate to one’s social networks in known areas. To some extent it may also be an artefact of the different processes by which risk is estimated in areas known through direct experience compared with areas that are known mainly by reputation (see further Social stigma and area reputation) or media reports (see Mass media). The remainder of this section focuses on the more direct association between familiarity and reduced fear.

Much of this association is practical in nature in the sense that knowing an area helps to reduce one’s vulnerability to or risk of attack. Knowledge can reduce vulnerability in that physically evading an attack, if necessary, will be easier in an environment (either physical or social) that one knows well (Valentine a). It can also help one avoid higher-risk areas (Moran, Seabrook).

Familiarity may also help to reduce the perceived risk of being victimised. Ethnic minority and lesbian and gay participants mention how their perceptions of different spaces affect their assessments of risk (see Ethnicity and Sexuality respectively). Young people in several studies also mention that leaving their ‘own’ areas may be intrinsically risky, as outsiders are likely to be the targets of attack regardless of their behaviour or appearance:

Would you go into different schemes [areas]?

Never.

I wouldn’t go in [area].

Why?

It’s no in Newhouse and they just fight whoever they don’t know.

Participants, Turner

What you can’t do is expect to be safe if you go up the other end and into someone else’s area.

Participant, male, 14 years, Walklate

More speculatively, the theme of familiarity may also link to the ‘legibility’ of the social environment in general, that is, the broad predictability of others’ social behaviour. This connection is discussed further in Fear and knowledge of the social environment.

Social cohesion and interpersonal networks

The sense of belonging socially to a specific neighbourhood, and knowing a range of people locally, appear to be important protectors against fear. Participants frequently observe that they are not fearful because they know people around them, particularly when they are long-established residents and have rich social and family networks (Farrall, Innes, Valentine a). As already suggested, the tendency for people to feel less fearful in their own areas (see Area knowledge) may be largely explained by the effect of interpersonal relationships:

I think it’s the whole atmosphere living in the Glodwick. Everyone knows everyone, so you’re not a stranger in your own town. And you just feel so safe, just in your own street and your own area.

Participant, Innes

Here I feel safe. It’s funny it’s mostly been the same families, the mothers have had the houses and then the daughters have carried on, and their sons have carried on. [. . .] I know a lot of people in this area. I grew up here, my friends are here.

Participant, female, Valentine a

This link between social belonging and reduced fear is complex. In some cases, such as those cited above, social belonging appears to directly create a sense of personal security. In other cases, people’s sense of being ‘known’ in an area may refer more specifically to having a reputation such that potential criminals know it would be unwise to attempt crimes against them, or to the knowledge of a semi-explicit ‘code’ of values that rules out committing crimes against people who are known locally (Farrall, Squires, Valentine a, Walklate). These more practical concerns link to the theme of ‘self-policing’, which, as discussed in ‘Self-policing’ and ‘grassing’, suggests a more ambivalent side to social cohesion:

We don’t get any trouble, I’ve a lot of friends and family on this estate and we can look after ourselves . . . if you know what I mean

Participant, male, 50s, Squires

Given the associations between interpersonal knowledge and perceived safety, fear tends to focus specifically on outsiders coming into the area. A sense that neighbours recognise each other helps to ensure that outsiders will be recognised and observed, which is seen to reduce risk (Farrall, Valentine a, Valentine b).

Some participants suggest that maintaining social relationships can help to reduce fear, even when they are relatively superficial; they need not involve close friendships or extensive day-to-day interactions:

Community spirit is there in that sense. We all have each others’ phone numbers on this street, and while we never go into each others’ houses we would all use them if we thought something was up. So you don’t feel isolated, not at all.

Participant, Pain a

However, even these more distant but still trusting relationships may be absent in some contexts. SES appears to be a relevant factor here. One study that directly compares a more middle-class area and a more working-class area finds considerably greater perceived social cohesion in the latter, helping to reduce fear (Valentine a). Residents of the middle-class area reported shallower and less trusting relationships with those living nearby, and less close interaction with them, which may increase fear.

Finally, to a much greater extent than with the theme of interpersonal familiarity, that of social cohesion links explicitly to theoretical discourses about the causes of crime. Crime is seen by participants in several studies [this is particularly a focus of Farrall et al. ’s study (Farrall), but is also echoed elsewhere] as driven by a breakdown in trust and social cohesion. This links to the theme of crime as social symptom, discussed later.

Young people as threat

Several specific types of people are mentioned as cues for fear. Young people are frequently mentioned in this regard, particularly when they are ‘hanging about’ in groups (Crime Concern a, Crime Concern b, Little, Nayak, Pain a, Squires, Valentine a, Walklate):

When I go to get my pension I am very aware of teenagers hanging about and such like, you feel they’re watching you.

Participant, Pain a

There is a problem with teenage children hanging around. They are hanging around at the moment, well they’ve been there for a while now at the bottom of the school drive.

Participant, female, Valentine a

Although participants in several studies discuss specific experiences of crime and antisocial behaviour by young people (Farrall, Nayak, Pain a, Walklate), in many cases the fear of young people does not appear to be motivated by any direct experience. Several participants explicitly recognise that this sweeping fear is ill-founded (Crime Concern b, Day, Pain b):

My old age group, I find a lot of people, when they see a gang of youths, they get a scowly grumpy face, [. . .] And I’ve found quite often if they’re walking past and I give them a smile, I get a smile back.

Participant, female, 70s, Day

Young people themselves are also aware of, and explicitly criticise, the tendency to stereotype them (Nayak, Seabrook, Walklate). In particular, young people in several studies complain of being hassled by the police on the basis of such stereotyped views (Alexander, Nayak, Pain b).

The lack of specific motivation in many cases for the fear of young people, and the prominence of the theme of ‘hanging around’, suggest that this fear is mainly driven by the particular way in which young people use public spaces, rather than by a direct estimate of risk.

Alcohol and drug users as threat

Another group who are often the focus of fear are people drinking or using drugs. The main threat from alcohol and drug users is the possibility of verbal or physical aggression, or being mugged for money (Crime Concern a, Crime Concern b, Day, Farrall, Innes, Waters b). Some participants express this more broadly in terms of the unpredictability of people’s behaviour when they are under the influence of alcohol or drugs (Farrall, Valentine a). The presence of large groups of drunk people at pub closing time is particularly threatening (Crime Concern a, Moran, Valentine a). As noted earlier (see Cleanliness and neglect), the presence of drug-related detritus such as used needles is also a particularly threatening aspect of the physical environment, especially for children.

‘Self-policing’ and ‘grassing’

In certain contexts – generally highly cohesive but materially and socially disadvantaged communities – there may be a considerable amount of ‘self-policing’. This term refers to people administering informal punishments for crime themselves, without involving the police. Particularly severe punishments are reserved for co-operating with the police or ‘grassing’. This phenomenon is particularly a focus of the study by Walklate et al. (Walklate), but the findings of that study are echoed in several others.

Many participants report feeling reassured by knowing that their community is willing to police infractions of its moral code. This is seen to both reduce the risk of crime occurring – at least for ‘insiders’ to the community – and increase the chances of an effective response when crime does occur. This is linked to the theme of social cohesion (see Social cohesion and interpersonal networks) in that effective self-policing requires both strong interpersonal relationships and a widely shared set of norms (Farrall, Innes, Squires, Walklate):

There’s a low chance of people our age group getting robbed. Not round here, it’s quite a close community [. . .] I don’t know how to say it but like, if there is a problem, it’s going to be sorted . . . We look after our own.

Participant, Innes

There’s also a positive side sometimes. [. . .] It has its own rules as well. They sort things out their selves.

Participant, Walklate

Such strategies may also operate within more specific communities: ethnic minority participants in two studies describe cases in which crimes were resolved without police involvement as a result of strong pre-existing relationships within the ethnic communities affected (Innes, Valentine a).

An important dimension of self-policing is that there is a ‘code’ that governs criminal activity and proscribes victimising ‘locals’ (Farall, Walklate). To this extent it is clear that the potential protective effect of self-policing is limited only to ‘locals’ or social insiders; the impact on outsiders is likely to be negative: ‘Oldtown is a great area if you are a member of the community, went to the local school and grew up with the local villains, but terrible if you’re an outsider’ (participant, female, Walklate).

Moreover, in self-policing communities, getting a reputation as a ‘grass’ is likely to lead to considerable harassment and in some cases serious violence (Squires, Walklate).

Ethnicity

Ethnic minority participants in several studies report the perception that they are at greater risk because of their ethnicity (Burgess, Crime Concern a). A few participants report a serious fear of racially motivated attacks, which has had a substantial impact on their lives (Hopkins, Squires):

We get people throwing stones at the house and swearing at us. . . . When I go out people are saying ‘**** black bastard’ it is terrible. [. . .] This place is like a prison. I can’t go out, I [. . .] can’t sleep [. . .] because of what they might do.

Participant, male, 50s, Asian, Squires

The ethnic coding of different neighbourhoods, and its relation to fear, is particularly a focus of two studies (Hopkins, Innes). They find that areas are strongly coded in terms of the majority ethnicity of their residents, and that this has an impact on fear, for white as well as for minority ethnic participants: ‘I wouldn’t feel happy walking down this area here, just outside my area, it’s like a white area. I wouldn’t be happy walking down there because I’d feel more vulnerable’ (participant, Innes).

This ethnic coding appears to reinforce the effects of familiarity and social relationships (see Area knowledge and Social cohesion and interpersonal networks respectively) to produce strong differentials in perceived risk, although it can make a difference even in unfamiliar areas (Hopkins).

Several participants express a fear of minority ethnic people, particularly black men, although in many cases they also express a critical attitude towards their own fears (Hollway, Valentine). As with a number of the other themes examined in this section, a sense of unfamiliarity and unpredictability appears to underlie this fear: ‘Blacks are threatening because you can’t see the expressions on their faces so you can’t tell what they’re thinking’ (participant, female, Valentine).

Several participants in the study by Farrall et al. link these ethnic fears to broader theoretical discourses, particularly a critique of multiculturalism (Farrall), although this link is not clearly reported in other studies.

Gender

Women’s and men’s experiences of fear of crime appear to be substantially different. In general, much fear of crime might be more accurately characterised as women’s fear of being attacked by men (Crime Concern a, Pain a, Valentine a). ‘Our main fear is attack from a man, especially where it is isolated and lonely . . . the whole issue about walking at night is about being attacked by a man’ (participant, female, Crime Concern a). (It should be noted here that almost all of the included studies focus on fear of stranger attack; the small amount of data on intimate partner violence is set out in Domestic and intimate partner violence.)

Several different factors appear to explain the difference in fear between men and women. To a substantial extent, the difference reflects a difference in the crimes that are feared. In particular, sexual assault or rape is the main focus of women’s fear (Goodey, Koskela, Valentine a): ‘I think without a shadow of a doubt, you know, I’d rather be killed than raped, you know, stabbed than raped’ (participant, female, Koskela).

Some of the differences in expressed fear may also be due to the differences in gender norms, which make it less acceptable for men to express fear (Goodey). Such differences are reinforced by media reporting (see also Mass media), which is seen to focus on crimes against women, particularly rapes (Goodey, Valentine a). There is also a widespread perception that women are more vulnerable to attack because they are less physically strong (Pain a, Valentine a).

Some participants suggest that the pervasiveness of low-level sexual harassment and minor sexual crimes, such as indecent exposure, tend to maintain an awareness of the possibility of more serious crimes. Valentine’s study (Valentine a) focuses particularly on this as an explanation of women’s fear, but her findings are echoed in other studies (Burgess, Taylor):

Whistling, that frightens me. [. . .] I walk from Reading to Newtown every morning, the quickest way is along the canal, but there’s an awful lot of blokes there and they shout things to you. [. . .] I hate walking down there, I only go that way if I’m in a real hurry.

Participant, female, Valentine a

Women’s fear of crime appears to be considerably more pervasive and inescapable than men’s. This may partly be explained by the role of common, day-to-day reminders of the possibility of crime, as mentioned above:

You’re never safe at any time. If somebody wants to go out and attack a woman, they’ll do it.

Participant, female, Koskela

It can happen anywhere, anytime, so you can’t really do anything about it. I suppose the onus is always on the woman to be on your guard.

Participant, female, Valentine a

One gender-specific environmental intervention on which views were sought is women-only carriages on trains (Crime Concern b). Participants in this study were generally sceptical about the idea.

An important aspect of gendered fear is what some writers have called ‘vicarious’ fear. Several male participants express vicarious fear for their wives or partners (Farrall, Pain a, Valentine a). In some cases, this vicarious fear may be a source of anxiety for the women involved (Valentine a) and may lead to substantial restrictions on women’s activity (Farrall, Goodey, Pain a, Valentine a). Another form of vicarious fear that often takes a gendered form is parents’ fear for their children, which is explored further in Parents and children. As discussed there, many parents express greater fears for girls than for boys and place greater restrictions on girls’ behaviour.

The effects of fear of crime on women also appear to be more far-reaching than those on men, particularly relating to restrictions on activities. These differences are explored further in Restricted movement and activities.

To some extent, gender differences in the fear of crime are understood by many participants as being part of a broader context of power imbalances between men and women, although detailed data on this point are found only in one study:

I mean if men were getting raped and things were happening to them it wouldn’t be so bad, but because it’s just women, it’s just so annoying. I mean a lot of things are equal, but a lot of things won’t be exactly equal if women can’t go out.

Participant, Valentine a

Perhaps because of this awareness of the broader issues involved in fear, many female participants express a high degree of scepticism about the possibility of effectively addressing it.

[Q: What would make you feel safer?]

[. . .] I don’t know. I don’t think there is anything that would be kind of any safer. Nah, nah. Apart from maybe being in a wee fortress kind of thing.

Participant, female, 30s, Koskela

The one thing I would like would be to [. . .] be able to walk the streets how you want, what time you want and know you’ll get home in one piece without being violently abused. It’s the one thing [. . .] I’d love to see, but I can’t see it ever happening.

Participant, female, Valentine a

Some participants express a sense that fear of crime is not related to objective risk or vulnerability, as any non-zero risk is seen to be unacceptable. The pervasiveness of women’s fear of crime, and the widespread scepticism about solutions to it, relate to this sense that it is risk per se, rather than greater or lesser probability, that determines fear. Women’s fear is similar to parents’ fear for their children (see Parents and children) in this respect:

I don’t want to put myself into a position where I feel threatened. I don’t know how likely it is but I’m not prepared even to take the risk.

Participant, female, Koskela

It can happen anywhere, anytime, so you can’t really do anything about it.

Participant, female, Valentine a

Several participants suggest an increase in their own fear, or in social attitudes relating to women’s fear, over time, although it is unclear what the causes of this change might have been (Koskela, Valentine a).

Sexuality

One study focuses particularly on lesbian and gay participants (Moran). Several participants in this study report a fear of anti-gay attacks, frequently based on experiences of harassment ranging from relatively minor attacks to serious attacks. Participants describe a process of assessing the risk of violence in public spaces:

[Y]our subconscious is working a million miles per hour just calculating, ‘Is this a nice safe place to go or is it dodgy? Should I go out again before I get my head panned in?’

Participant, Moran

Domestic and intimate partner violence

As already mentioned, very few data are presented on domestic violence or intimate partner violence as opposed to attacks by strangers (see further Limitations of the review). Only one study presents views on fear of intimate partner violence, specifically women’s fear of spousal abuse, and relatively few data are reported (Koskela). Women in this study who had not experienced intimate partner violence tended to see their own risk as low. Generally, the data in this review do not support any conclusions on fear of domestic violence or intimate partner violence.

Parents and children

Another population who account for a substantial amount of fear of crime are parents, particularly parents of young children, who often express vicarious fear for their children. Abduction is particularly feared (Dixey, Farrall, Pain b). ‘I think all parents’ biggest fear is that somebody might abuse them, more than the fear of them getting killed or dying of an illness’ (participant, Dixey).

As with some of the data on women’s fear of sexual assault (see Gender), the fear of child abduction appears to be highly pervasive and not responsive to objective risk, because of the extreme seriousness of the feared outcome (Dixey, Koskela, Valentine b):

You hear so many stories, you know, and read so many things in the papers. And I know the chances are one in probably millions but you still always think, you know, it could be yours and you don’t want it to happen, you don’t want her to disappear.

Participant, female, Valentine b

Some parents report considerable impact on their psychological well-being as a result of this fear: ‘It makes you so darn mad, it chews you up inside. When you talk to people in the pub, you find that they are also all wound up about these things too’ (participant, Dixey).

However, in some cases parents report feeling social pressure to be seen to act in the interests of children’s safety, which does not necessarily correspond to their own fears (Valentine a).

Some participants report a greater fear for girls than for boys (Valentine a), although others feel that there is little difference (Dixey).

As well as parents’ vicarious fears for children, one participant expresses increased concern for her own safety since having a child:

I think it is different now, just in case anything happened to me, for her [daughter’s] sake. [. . .] For years I’d walk home through the university and I was never really bothered, but I don’t know, maybe round here it’s different.

Participant, Watson

As a result of these fears, many parents report placing considerable restrictions on children’s activities. These are discussed in Parental restrictions on children.

Talk about crime

Another feature of the social environment that may impact on fear is talk about crime. Several participants describe how their information about crime, and more generally their perceptions of safety, are shaped by discussions with others (Alexander, Goodey, Valentine a), although several participants recognise that such information is often unreliable (Valentine b):

Everything centres around the hairdressers, you hear everything there. You say I’m going out tonight and you say where you’re going and they say [. . .] ‘Such and such happened there’, or ‘My Jane’s friend went there and got attacked’, and so then you’re wary of going there.

Participant, female, Valentine a

Crime talk in more formal contexts, such as those provided by Neighbourhood Watch or workplace safety schemes, may also increase fear (Farrall, Valentine a) (see also Mass media).

Talk about crime particularly drives the perception that certain neighbourhoods are unsafe (see Social stigma and area reputation) and may account for feelings of unsafety with respect to areas that people have little or no direct experience of (see Area knowledge).

However, talk about crime not only is a contributor to fear but also can form part of a more general sharing of experiences and resources that may help people deal with the threat of crime (Valentine a). More broadly, talk about crime forms part of the texture of social interaction, which contributes to people’s local knowledge and sense of social belonging (see Area knowledge and Social cohesion and interpersonal networks); as such, it may indirectly reduce fear:

So what is it like to live here?

It’s canny like. Everyone knows everyone, an’ if summat’s gannin’ doon, we’re the first t’hear aboot it.

Everyone’s textin’ us what’s gannin’ doon – we’re always one step ahead!

Participants, Alexander

Experiences of victimisation

Previous experience of crime appears to be an important determinant of fear for many participants. Several participants, particularly those who have been the victims of violent crimes, report high levels of fear as a result, which in some cases may persist for years after the attack. These fears are often associated particularly with the location where the attack occurred, but also shape people’s experiences more broadly. The most damaging experiences reported in the data appear to be violent hate crimes, including anti-gay attacks (Crime Concern a), rape (Valentine a) and racially motivated crimes (Squires, cited above, see Ethnicity). However, in one study (Farrall) some victims of burglary also report serious impacts. Vicarious fears, such as parents’ fears for children, can also be increased by personal victimisation (Pain c), and witnessing crime can contribute to fear (Crime Concern b).

As noted earlier (see Gender), women’s experience of relatively minor sexual crimes or harassment appears to be an important contributor to fear. Many female participants describe experiences of harassment, indecent exposure or the threat of sexual attack, which raise their awareness of the possibility of serious crime:

It is such a horrible feeling though [being flashed at]. It is so traumatic and dreadful. You don’t want to bother risking it for the sake of having a walk.

It is imposing on you and your space.

Participants, female, 21–35 years, Burgess

And these boys started asking me my name, and I didn’t want to tell them. And then they started following me around. [. . .] You know, girls get raped, mugged, harassed, stuff like that.

Participant, female, 13/14 years, Asian, Jones

On the other hand, a few participants, particularly in the study by Farrall et al. (Farrall), say that experiences of crime have not made them fearful.

Mass media

Several participants describe television or newspaper reports as a source of fear. Sensational reports of violent crime are particularly a factor. Reporting of crimes that happen locally, in areas that people use on a regular basis, are especially likely to impact on behaviour (Farrall, Valentine a): ‘Every time ye lift a paper or pit on the telly, yer hearing something. That’s aw . . . it’s just kept alive in the brain as for as I’m concerned’ (participant, female, 58 years, Farrall).

However, some participants explicitly criticise the sensationalism of media crime reporting and its role in perpetuating feelings of fear (Crime Concern a, Farrall, Valentine a). There is also a recognition that the media reporting of crime is skewed in certain important respects. Participants critique the focus on sexual violence against women (Valentine a; see Gender) and the stigmatisation of certain areas and groups as stereotypical perpetrators of crime (Parry; see Social stigma and area reputation).

Individual factors

Participants identify a number of individual-level factors that may make a difference to fear. Several participants discuss people’s pre-existing psychological dispositions and attitudes (Crime Concern a, Farrall, Pain a). This emphasis on individual psychological factors appears to be particularly characteristic of older participants: ‘I’ve always been a fatalist in my life. And er I’ve never been afraid of anything. Not a single thing’ (participant, male, older, Pain a).

Along similar lines, Hollway and Jefferson’s330 findings emphasise how different people may respond differently to ‘objectively’ similar risk situations depending on the personal values and anxieties that inform their reading of environmental factors.

Relatedly, several participants describe the importance of maintaining a confident appearance (regardless of one’s emotional state) as a strategy to reduce risk (Crime Concern a, Pain b). Another important individual factor is one’s perception of one’s own vulnerability and resources to resist attack (Farrall, Squires): ‘I’ve got a gun cabinet back there, you know. Two shotguns and a .22 rifle. And if anybody breaks in here, they can have some of that’ (participant, male, 30s, Squires).

Crime as a social symptom

A theme already mentioned is the sense that crime represents not only the object of individual fear but also a symptom of social problems more broadly, often seen in terms of a historical narrative of decline over the last several decades (Burgess, Farrall).

I think everyone needs to have a certain respect and for law, the police, for parents, for elder people. And I think that has disappeared. I don’t think there’s that same caring community spirit that there used to be.

Participant, Farrall

Psychological stress

Relatively few participants report serious psychological stress as a result of fear of crime. However, a number of participants report less severe psychological impacts of fear (Airey, Dixey, Taylor, Valentine a):

[Y]ou can’t live your life in hiding, you’ve just got to think ahead, just be prepared . . . you have to think it out. It’s a terrible life, it’s a stressful life, very stressful. [. . .] It’s a dreadful way to live.

Participant, Taylor

Those who report serious psychological stress as a result of fear generally fall into two groups: people who have been the victims of serious crimes, particularly sexual or hate crimes, and possibly repeated crimes (see Experiences of victimisation); and people who have a pre-existing mental health condition, who are the focus of one study:

The worst thing? The violent people, the fear they are coming around and breaking and burning and hitting. [. . .] people are scared here because you don’t know who is coming next to the door, people living in fear.

Participant, Whitley

Other participants argue that fear of crime may impact on mental health as part of a broader nexus of individual and social stress and disadvantage (Airey).

Restricted movement and activities

A much more widely expressed consequence of fear of crime is to limit one’s daily activities (Alexander, Crime Concern b, Pain b, Parry, Seabrook, Taylor, Valentine a, Whitley). Although participants from all population groups say that their activities are limited by fear to some extent, the effect appears to be considerably greater for particular groups: women, older people, people with physical disabilities and people with learning disabilities.

So what do you get up to on a night time?

I just stay in.

Why is that?

Y’canna gan out at night, y’get hassled off the Charvers and Ragies . . . they’re really scary like.

Participants, 16–20 years, with learning disabilities, Alexander

I’ll go to my nana’s after school but most of the time just go in, ‘cos you don’t feel safe hanging around the streets.

Participant, 13 years, Seabrook

I just don’t go places or do things where I’ll be at risk.

Participant, female, Valentine a

Parents’ restrictions on children’s activities are a particular concern (these are discussed in Parental restrictions on children).

Of particular concern from the point of view of the health impacts of fear of crime are potential limitations on physical activity. Many participants report not participating in physical activity as a result of fear of crime (Burgess, Farrall, Koskela, Valentine a). In particular, several participants say that they will drive somewhere rather than walk because of the perceived risk of crime. Again, this is particularly the case for women.

I used to sort of cycle and everything, but I just don’t anymore. You can’t really can you? You can’t cycle anywhere, you can’t walk anywhere on your own, not when it’s dark, it’s not worth the risk.

Participant, female, Valentine a

Several female participants report that restrictions to activities become routinised in such a way as to become unconscious. That is, daily activities may be restricted or reorganised to avoid areas or situations perceived as dangerous, without this being a consciously expressed perception (Farrall, Koskela, Valentine a):

[I]n my mind I must be aware that I’m taking precautions, but you know I’m not aware of it. I’ve never really thought about it till we’ve had this conversation. [. . .] You’re doing it and you’re not aware of it.

Participant, female, Koskela

Parental restrictions on children

As described earlier (see Parents and children), many parents report vicarious fear on behalf of their children. This fear is often serious and pervasive. As a result, many parents report restricting their children’s activities (Davis, Dixey, Jones, Koskela, Squires, Valentine a, Valentine b). Again, outdoor physical activity may particularly be reduced: ‘I haven’t got any places to play because even if I do play outside my Mam says it’s too dangerous’ (participant, male, 9–11 years, Davis).

Several parents express regret that such restrictions are necessary, and recognise the potential negative effects of such limitations on children’s independent mobility (Dixey, Koskela, Valentine a, Valentine b). In some cases they recognise that they allow less freedom to their children than they themselves enjoyed at the same age (Valentine a, Valentine b).

I worry that I’m not telling her enough and that I’m not letting her play out enough. I’m stopping her from doing a lot of things that she should be doing, because of my worries, and I feel guilty about that.

Participant, Dixey

I don’t let the kids out late, like as soon as it’s dark I call them in from the streets. They have to sit in from six o’clock, which isnae fair on them really. It’s like an added worry because you’re worried about yourself anyway.

Participant, female, Koskela

Children and young people themselves also describe the negative effects of these restrictions. Two studies (Davis, Jones) describe how young people perceive a contradiction between public health messages encouraging physical activity and the strict rules that govern their behaviour. ‘If you go to the park then you are obviously going to get raped or something. If you go to the gym then you are going to get chucked out’ (participant, female, 13–14 years, Davis).

As noted in Parents and children, parents’ fears appear to frequently be gender specific, with several participants expressing greater fears for girls than for boys. Several participants report that greater restrictions are placed on girls’ activities than boys’ activities (Davis, Goodey, Jones, Valentine a). Media reports (see earlier section on mass media) may play a role in motivating parental restrictions, and, again, this appears to be more of a factor for girls (Jones, Valentine a):

When I ask my Mum can I go out with my friends she like says no. [. . .] if you were a boy you would have been OK, because then you’ll be able to look after yourself more. Because you’re a girl you might get raped, and all these things.

Participant, female, Asian, 13–14 years, Jones

Parental restrictions may create conflict within families, as children and parents disagree over perceptions of safety and allowable activities (Davis, Jones, Pain c). Several children and young people describe giving misleading or incomplete information to their parents, to avoid generating worry that will lead to further restrictions (Davis, Pain c).

Social isolation and lack of cohesion

A further consequence of people restricting activities may be to impinge on their social relationships. As described earlier (see Restricted movement and activities), several participants report being unable to engage in social activities as a result of fear (Day, Farrall, Parry, Valentine a).

More broadly, some participants see fear of crime as having a negative effect on the texture of social interaction at a community level. This links to the theme of crime as a symptom of social decline (see Crime as a social symptom), such that fear of crime is seen as both a consequence and a cause of reduced social cohesion:

Community spirit’s different now to what it used to be, because as I say people are afraid to go out. At nights here it’s very quiet, you never see anybody walking about. They just don’t leave the house for fear of being burgled or attacked.

Participant, male, older, Pain a

Social stigma and area reputation

As several themes have already suggested, fear of crime tends to attach to particular areas or neighbourhoods, sometimes as a result of personal experience but more often through social discourses and media reporting about dangerous areas:

Whitley they say isn’t so nice and parts of Tilehurst I’ve heard, and West Reading (Oxford Road area) they say isn’t so good. People just say ‘Oh it’s rough round there I wouldn’t go that way’.

Participant, female, Valentine a

As described earlier (see Area knowledge), although this kind of local knowledge may be protective against fear for individuals, in that it enables them to avoid areas that induce fear, it also leads people to see unfamiliar areas as more dangerous than their own. Thus, when areas are known to have a reputation as dangerous, residents of those areas tend to see this reputation as undeserved and stigmatising. The difference between their perceptions of their own area and outsiders’ views creates a sense of unfairness of the latter (Airey, Alexander, Farrall, Squires, Taylor, Walklate). Moreover, as such reputations go along with material and social disadvantage, they link to broader social prejudices:

You say ‘Kirkhead’, some of the reactions are actually incredible, and uh they therefore tar you with being somebody who comes from a council estate and uh therefore has nae brains and you know no etiquette, [. . .] they just can’t believe that you could be well brought up and educated.

Participant, Airey

It’s a real shame about the reputation of this place because it makes it embarrassing to say where you live. But actually, there is no problem living here.

Participant, male, 20s, Squires

Feelings of unsafety with respect to particular neighbourhoods also tend to reinforce stereotypes of their residents as criminal (Parry, Squires, Valentine a, Whitley), which, as noted earlier (see Young people as threat), may particularly affect young people’s relations with the police.

You tell someone you are from [area] they automatically think you have got something to do with guns or drugs, and you are a part of a public gang.

Participant, Parry

Whitley, you say you’re from Whitley, and people look at you as if you’re going to pull a knife on them, that’s horrible.

Participant, female, Valentine a

‘Functional fear’

In contrast to the other data reported in this section, several participants report that, although they may feel fearful at times, this has no serious impact on their behaviour or well-being (in addition to those who report not feeling fear at all; see Individual factors). ‘I’m conscious it could happen and I’m aware of it and I want to do something to prevent it, if possible, I’m not sitting anxious about it and thinking maybe somebody will do it ye know’ (participant, male, 49 years, Farrall).

Fear and knowledge of the physical environment

A key factor influencing fear is knowledge of one’s surroundings. As we have seen, much of this can be explained by the fear-reducing effects of social belonging or ‘insidership’ and the presence of strong social networks in local areas. However, knowledge may also help to reduce fear even in the absence of strong social networks. We suggest that the construct of ‘cognitive mastery’ links together a number of the themes, particularly those of familiarity and visibility. The sense that one’s physical environment is open to perception and understanding helps to reduce fear in general. As already noted (see Environmental visibility), the theme of visibility or ‘openness’ is linked with a sense of freedom of action: an environment that is cognitively transparent is also seen as clearly understood in terms of the ‘affordances’ or resources for agency it provides, such as alternative routes.

Fear and knowledge of the social environment

Another important dimension of knowledge, which, again, is independent of the findings on social networks, is the adequacy of one’s knowledge of the social environment. The criminology literature has extensively investigated the role of ‘social incivilities’, such as public drinking or drug use, in driving fear, and the qualitative findings bear out much of this research (see further Chapter 7, Incivilities). They also help to explain some of the pathways through which this association functions. The findings on drinkers and drug users (see Alcohol and drug users as threat) and on young people ‘hanging about’ (see Young people as threat) suggest that such groups are a source of fear insofar as they behave unpredictably and depart from social norms, particularly norms about the use of public space. The findings on familiarity and social cohesion (see Area knowledge and Social cohesion and interpersonal networks respectively) indicate that the fear-reducing effects of social belonging are also substantially driven by a sense of having reliable expectations of others’ behaviour. Common to all of these themes, then, is the sense that much fear of crime arises from the breaking of social norms governing everyday behaviour and interactions, especially interactions between strangers in public space.

The pervasiveness of this sense of social predictability allows us to extend the idea of ‘cognitive mastery’ to the social as well as the physical environment. As with the physical environment, knowledge of the social environment reduces fear both directly and by providing means to avoid crime or to minimise its impact. That is, cognitive mastery is linked to the pragmatic mastery of the social environment and an ability to effectively use the resources it offers. It should be noted that this knowledge is relative to context, and not universally associated with the same phenomena. When the fears of different groups play out conflictually – between young people and older people (see Young people as threat), for example, or between residents of disadvantaged areas and outsiders (see Social stigma and area reputation) – this often appears to result from clashes between divergent sets of social norms. In the case of self-policing communities (see ‘Self-policing’ and ‘grassing’), considerable effort goes into maintaining the transparency of the social framework at a community level so that transgressors against norms can be dealt with.

Public and residential areas

The distinction between public areas (such as shopping areas, public streets and squares, parkland or open countryside) and residential areas seems to be important in shaping perceptions. It should be noted that this is not the same as the distinction between public and private, as particularly deployed in the CPTED concept of ‘territoriality’ (examined further in Chapter 7, Crime Prevention through Environmental Design). The latter is a much sharper and finer-grained distinction, which runs, for example, between a house (or garden) and the street, whereas the public/residential distinction operates at a neighbourhood, or at least street, level.

Many of the determinants of fear explored earlier appear to be specific to one or the other type of area. For example, familiarity appears to reduce fear primarily with respect to residential areas and has little impact on the fear of public spaces. In public areas, factors to do with the physical environment appear to predominate as determinants of fear, whereas in residential areas – particularly one’s own area – the social environment appears to be much more important. Relatively few participants report physical factors such as lighting or neglect as major determinants of fear in the neighbourhood in which they live and, when they do, these appear to be ‘read’ mainly in terms of their social meaning rather than directly as cues for fear (see the following section).

Social meanings of the physical environment

One clear implication of the qualitative findings is that the physical environment affects fear more through its social meanings than directly. Several themes show a consistent link between specific environmental determinants (particularly lighting, visibility and neglect) and fear. Nonetheless, social factors appear to be more central and can frequently over-ride the effects of the physical environment. For example, several participants observe that their knowledge of, and sense of socially belonging to, their own areas outweighs the potentially fear-inducing effects of factors such as neglect or poor lighting. Again, this appears to be particularly the case with residential areas; in public areas such as shopping streets or parks, physical factors may have more importance as determinants of fear.

The fear-inducing effect of such socially coded physical factors is partially a question of perceived risk: environmental indicators of disadvantage are an important part of the complex of cues that enable people to judge the riskiness of different neighbourhoods. However, they also have broader symbolic meanings. Characteristics of the built environment such as the use of prefabricated concrete are seen as threatening, even in contexts such as university campuses, because of a residual association with disadvantaged residential areas (Waters). Evidence of physical neglect points to broader social and political narratives of neglect, disadvantage and inequality. The imagery of disadvantaged residential areas – concrete, high-rise blocks, graffiti and so on – becomes associated with fear within a broader complex of associations that link such environmental factors to poverty and thence to high crime rates. As discussed further in Chapter 7 (see Crime Prevention through Environmental Design), these findings cast doubt on theories such as CPTED which suggest that environmental improvements alone are likely to reduce fear. If the fear-inducing effects of environmental factors are primarily indirect, through their role as indicators of social, economic or political factors, interventions that address the former while ignoring the latter are likely to have only limited and short-term impacts on fear.

More generally, these findings highlight the complexity and theoretical sophistication of the discursive and conceptual contexts within which fear is articulated. People’s perceptions of their environments are not simple reactions to stimuli, but are irreducibly linked to their views about the communities and the broader society in which they live. The qualitative findings suggest that much of the available theory and data on fear of crime, particularly in the more positivist research tradition, has underestimated its theory-laden and context-sensitive nature. In particular, views of and reactions to crime are closely bound up with views about politics and policy. In some cases – of which Squires’ data on CCTV provide a clear example – this may lead to strongly negative perceptions of environmental crime prevention interventions. It is not just that such interventions raise awareness of crime, and hence fear (although this may often also be the case), but that they are seen explicitly as inadequate substitutes for more substantive (and expensive) solutions, and hence as indices of the unimportance of disadvantaged neighbourhoods to policy-makers. In addition, the perception of certain neighbourhoods as safe or risky is based on a complex of factors that are sometimes justified in terms of their physical appearance or layout, but which are also centrally informed by social stereotypes about the populations of those neighbourhoods (see Social stigma and area reputation).

Fear and risk

The relation between fear and risk (what we described in the theory review as the ‘affective’ and ‘cognitive’ components of fear of crime; see Chapter 3, Fear of crime: measures and contexts) cuts across many of the qualitative findings. It is clear from the qualitative data that perceptions of risk are often closely linked to fear. Many participants describe extensive and fine-grained assessments of risk, and detailed mental maps of particular neighbourhoods and places. Some sources of information for risk assessment are direct in the sense of constituting information about crime rates: personal experience, others’ reported experiences, media reports and so on. Other sources are more indirect and build on specific contextual cues in the social or physical environments to inform assessments of risk.

However, it is also clear that such processes for assessing risk are themselves often saturated with symbolic ideas and judgements. This is already the case on an individual level but is amplified further at the social level, as discourse about crime and risk taps into, and thereby perpetuates, prevalent social stereotypes. As a result, it is difficult to separate out a cognitive or rational component and an affective or symbolic component, as many theorists have attempted to do. The qualitative data also underline the importance of minor crime and antisocial behaviour, or concrete threats of crime, in the genesis of fear. Such experiences are reported particularly by women, minority ethnic participants and lesbian and gay participants. These findings suggest that the relationship between perceptions of risk may inform fear in complex ways that cannot be captured by simply comparing statistical risk of a given crime with expressed fear of it. We return to these questions in Chapter 7 (see Fear and rationality and Inequalities and spirit injury).

Two further points should be noted here. First, subjective risk may function much of the time unconsciously, for example in shaping avoidance behaviours, without emerging into consciousness as an explicit estimate of risk. Although qualitative data alone can illuminate this question to a limited extent only, it appears that much of the ‘processing’ of risk cues may take place unconsciously, and possibly that the impact on fear may itself be unconscious.

Second, the risk–fear relation appears to vary considerably depending on the type of crime in question. It is clear that some crimes are feared much more than others, particularly rape, sexual assault and child abduction. For several participants, the notion of objective risk appears to have little meaning as applied to these most serious crimes. The mere possibility of any risk at all is seen to outweigh any other considerations, or, more exactly, to mean that it would be inappropriate to attempt to quantify and balance the risk against other risks and benefits. This suggests one reason why reducing fear of crime has proved so difficult: as the risk can never be eliminated altogether, responses to crime are unlikely to track changes in perceived risk (nor, a fortiori, changes in actual risk).

Population subgroups and inequalities

Several differences between population subgroups emerge from the qualitative data. Although it is difficult to say on the basis of these data to what extent fear is more serious in some groups than in others, there is considerable evidence that fear has greater impacts on behaviour for some groups, particularly women but also (although fewer data are available) people with disabilities, people with mental health problems and possibly black and minority ethnic (BME) and lesbian and gay people. The findings in older people appear to be more mixed, with some reporting substantial impacts and others few or none.

There appear to be some congruences between the experiences of women and those of minority ethnic participants and gay and lesbian participants in the way that fear of crime is linked to broader dynamics of social inequality. Participants from all of these groups report a pervasive fear, driven by everyday experiences of discrimination and harassment, which supports an awareness of the possibility of more serious crime. Perceptions of risk also appear to inform their experience of the environment at a more fundamental level, in the sense that they constantly ‘scan’ public spaces for risk, consciously or unconsciously. In addition, for women particularly, their own fears are often reinforced by ‘vicarious’ fears on the part of husbands, boyfriends or parents. These points are taken up in the discussion in the following chapter (see Inequalities and ‘spirit injury’). It should also be noted, however, that parents of young children report a similar kind of pervasive fear but without the same experiential grounding, so this phenomenon cannot be explained in any one-dimensional way.

The perceptions of children and young people are distinct from those of adults in several ways. For younger children in particular there appear to be specific environmental cues for fear that are distinct from those of adults (see Cleanliness and neglect). Older young people (teenagers) also report different determinants of fear. Young people in disadvantaged areas appear to face considerably greater risk of violence when leaving their own areas. This may indicate that the fear of unfamiliar areas is more directly grounded in realistic estimates of risk, and experiences of victimisation, for young people than it is for adults. Finally, children and young people have limited autonomy in responding to risk and their movements are often restricted by their parents (although they have various ways of getting around such restrictions); they are conscious of the dissonances between their own and their parents’ estimates of risk and often aware of ways in which the latter are misinformed.

The perceptions of older people also seem to some extent to be distinct. Two sets of themes appear to be particularly characteristic of this population. The first is the broader meaning of crime as a symptom of social decline, the belief that crime is increasing and the linking of discourses about crime to broader understandings of social trust and cohesion. These perceptions are by no means expressed only by older participants, but they appear to be more strongly expressed by them. The second is the perceived importance of individual-level factors, such as psychological dispositions. Older people appear to be more likely to trace fear back to questions of one’s own attitudes to the world and less likely to situate it in the wider environment. This may arguably reflect a broader long-term shift within British culture, with participants born before the Second World War more likely to explain emotional responses ‘subjectively’, in terms of individual propensities and character, and later generations more likely to explain them ‘objectively,’ in terms of the aspects of reality to which they respond.

People with mental health problems appear to suffer particularly seriously from fear of crime (see Psychological stress). However, the data do not allow any further elaboration of the pathways involved here.

Structure of the chapter

Table 4 outlines the structure of the chapter, showing how each part of the cross-study synthesis draws on material from across the different components of the project. Table 4 is not exhaustive and in particular does not show the connections between the different parts of this chapter itself, but it provides a guide to the main findings discussed in each section.

Home security interventions

The effectiveness review found mixed findings for home security with respect to fear-related outcomes. We speculated that some of the variation might be explained by whether interventions involve raising awareness around crime more generally (the less effective interventions) or are set in a broader context of environmental improvement (the more effective interventions). The qualitative data would seem to lend some support to this in that physical security measures were found to be potentially reassuring but were also associated with feelings of anger and with the view of crime as a symptom of social decline. Thus, the two bodies of data taken together suggest that physical security interventions may have limited potential to reduce fear in the context of strongly crime-focused intervention programmes, but that security enhancements may be promising as part of broader environmental improvements.

Street lighting

The effectiveness review found limited evidence that street lighting reduces fear. There is some evidence of improved fear outcomes as a result of lighting improvement interventions from studies using non-comparative designs. However, studies using more robust controlled designs are more equivocal, with two showing no improvement, one with mixed findings and only one providing clear evidence of effectiveness.

The qualitative findings are arguably encouraging with respect to lighting interventions, given the recurring finding that inadequate lighting increases fear. However, the qualitative review also suggests that, in many cases, the direct impacts of lighting may be outweighed by social factors and that some participants were sceptical about the likely effect of lighting improvements on fear. This suggests that, although improved lighting may have some role to play in reducing fear as part of a broader intervention programme, it is not promising in isolation.

The qualitative findings also point to a number of more specific implications with regard to lighting interventions, such as the importance of consistency in lighting, the need to take into account the layout of the built environment and the overall visibility of areas in planning relighting, and the preference for white lights over yellow or orange ones; this last point is borne out by the effectiveness literature (Knight).

Closed-circuit television

The findings on CCTV appear to be consistent in that there is very little evidence that CCTV improves fear outcomes and few positive perceptions of it to be found in the qualitative data. The findings strongly indicate that CCTV is not a promising intervention for fear of crime (see also Crime Prevention through Environmental Design on ‘natural surveillance’).

Multicomponent crime prevention interventions

The effectiveness review found that multicomponent crime-focused interventions were generally ineffective for fear of crime or social outcomes, although the data are not unequivocal. A number of factors that are often a particular focus of these interventions can be related to the qualitative review. Arguably the most important are ‘physical incivilities’ such as run-down or abandoned buildings, graffiti and litter. The qualitative evidence is to some extent encouraging here, in that the association between such environmental evidence of neglect and fear of crime is a recurring theme; as noted in the theory review (see Chapter 3, Built environment, social environment and fear of crime), this theme is also strongly borne out by the observational quantitative evidence (see further Incivilities). On this basis, one might speculate that the failure of interventions to demonstrate reductions in fear is either because of limitations in the available evidence or because of other components of the interventions negating the positive effects of a reduction in incivilities.

This said, the qualitative review also found that, in many cases, incivilities are associated with fear through their social (and politicoeconomic) connotations, rather than directly. [The quantitative observational evidence included in the review of theory has little to say about this question, although the finding in some studies (see Chapter 3, Built environment, social environment and fear of crime) that participants’ perceptions of incivilities are a better predictor of fear than researcher-rated measures may be relevant.] It could be argued from the qualitative data that the underlying complex of social and material disadvantage is a more important determinant of fear than its physical symptoms, and hence that interventions with a primary focus on the latter are inherently limited. (Although several interventions in this category did also include social components, they were in most cases relatively limited and crime-focused and unlikely to impact on broader community-level problems.) However, the qualitative findings are by no means conclusive on this point and a more optimistic reading is possible; see further in Improvements to public areas.

Another factor is visibility or ‘openness’, which emerges from the qualitative evidence as a potentially strong determinant of fear. This may be addressed in interventions by, for example, removing vegetation or other obstacles to visibility, improving lighting and installing mirrors. Three intervention studies (Baker, Felson, Webb) particularly emphasised such components, all of which found broadly encouraging trends towards reduced fear. Finally, as already noted, the qualitative review found generally negative perceptions of security measures such as fencing and shutters. Interventions with a strong focus on measures such as locks, barriers, entry systems and road closures generally appear to be less likely to show evidence of effectiveness (Arthur Young, Baker, Donnelly, Fowler, Kaplan a). However, although these resonances between the two sets of data are suggestive, they involve sweeping generalisations about complex multicomponent interventions and highly equivocal data, and are very far from demonstrating the differential effectiveness of the intervention components.

Housing improvement

The effectiveness review found mixed evidence for an effect of housing improvement and relocation interventions on fear outcomes. The implications to be drawn from the qualitative evidence are also somewhat mixed. On the one hand, the qualitative data indicate the importance of a pleasant and welcoming physical environment in reducing fear, which would suggest that housing interventions are promising in reducing fear. On the other hand, the qualitative data also indicate the importance of social familiarity and interpersonal networks in protecting against fear (although, as the data on self-policing indicates, these factors are not invariably positive in their effects). This latter body of data suggests that interventions such as relocation may have adverse effects with respect to fear if they disrupt social networks, although this implication is not clearly borne out by the detail of the effectiveness data.

Regeneration and area-based initiatives

Taken in overview, the qualitative data strongly encourage an integrated, area-based approach that integrates crime prevention activity within broader environmental and social programmes. However, the small body of effectiveness data on such interventions is equivocal and does not strongly indicate that they are effective in reducing crime. As previous research has found,308 it is very difficult to evaluate these broad interventions with respect to health or socioeconomic outcomes; our findings bear out this difficulty. There is a clear need for more robust evaluations of regeneration and area-based initiatives in general, and with respect to fear of crime outcomes in particular.

It is possible that the very substantial improvements in fear of crime outcomes seen in both the intervention group and the control group of the New Deal for Communities evaluation (Beatty) indicate the effect of programmes targeted at deprived areas in general, and operating outside the New Deal for Communities framework, but, of course, this cannot be demonstrated from the available evidence. As already noted, the findings of the New Deal for Communities evaluation are consistent with evidence from the British Crime Survey showing a steady improvement in fear outcomes across the UK population over the last 15 years or so (see Implications of the findings for research and policy).

Improvements to public areas

The effectiveness review found that non-crime-focused environmental improvements in public areas can produce small but significant improvements in perceived safety outcomes. This again bears out the importance of physical incivilities as found in both the qualitative review and the theory review. Relatively small-scale and low-cost interventions appear to be promising when they address common environmental cues for fear such as graffiti, although it is unclear whether or not the improvements in cognitive outcomes (perceived risk) will feed through to improved affective and well-being outcomes.

More specifically, we may be able to relate the effectiveness findings to the distinction between public and residential areas with respect to determinants of fear that we tentatively proposed on the basis of the qualitative evidence. If we look again at categories 4 and 7 together (i.e. environmental improvements with and without a crime prevention focus respectively) and then divide by public or residential area, those with a greater focus on public areas, such as commercial areas or public transport (Baker, Cohen, Felson, Kaplan b, Palmer, Webb), are arguably, although not invariably, more likely to be effective than those with a main focus on residential areas (Arthur Young, Donnelly, Fowler, Mazerolle). [It is unclear on which side of the divide we should place schools, as we have little qualitative data on the school setting; arguably, the relatively persistent and substantive social relationships found in schools make them more like residential than public areas. This would make Kaplan et al. ’s study of schools (Kaplan a) an exception to the general trend.] If this interpretation is tenable, it may suggest that environmental improvements generally (whether crime focused or not) are more promising in public areas than in residential neighbourhoods, where the effect of the physical environment in isolation is likely to be outweighed by social factors.

Crime, fear of crime and well-being revisited

The theory review found limited evidence for a strong link between crime, considered in isolation, and health and well-being at the population level (as distinct from the health impacts of crime on individual victims). Although there is a robustly established area-level association between higher crime and poorer health, it is unclear whether or not we need to posit a direct causal link to explain this association rather than an indirect one driven by shared determinants of the two outcome types, particularly socioeconomic disadvantage.

To the extent that such a link exists, most existing measures of fear of crime do not appear to be very useful in exploring it. Although there is some evidence for a modest association between fear of crime and health outcomes, fear of crime itself is only weakly related to crime and hence not a reliable mediator between crime and health. Any direct area-level link between crime and well-being is probably more likely to be explained by a general concept of psychological stress. The social impacts of crime are likely to be more accurately measured using existing measures of social well-being (e.g. social cohesion, trust) rather than measures of fear of crime.

The review of theory suggests that the true value of fear of crime as an explanatory mechanism is probably to be found elsewhere than in the link from crime to well-being. The associations observed between fear of crime and health and well-being outcomes on the one hand and environmental determinants on the other suggest that fear may be a mechanism for mediating the effects of environmental factors more generally, such as environmental neglect or poor housing conditions, on health and well-being. However, the findings of the qualitative review indicate that these factors need to be understood in a holistic way that integrates their social meanings, and not treated as purely physical determinants, to understand their impact on well-being. In addition, there remain a number of unanswered questions about the link between fear of crime and health outcomes, particularly with regard to the direction of causality underlying it and differences in its strength among different population subgroups.

Because of the unclarity of the crime–well-being link, it is difficult to reach any substantive conclusions on the extent to which crime reduction may be a promising way to improve health and well-being more broadly. This is exacerbated by the near-total absence of evaluation data measuring the health effects of crime reduction interventions. We located only a single study of a crime reduction intervention in the general community-living population that measured health status outcomes (Halpern). This study, although showing some promising findings, suffers from some methodological limitations. In addition, as the intervention evaluated included a substantial amount of more general, non-crime-focused environmental improvement, it could plausibly be argued that the positive mental health effects observed were a result of the generally improved housing stock and public environment, rather than anything to do with crime. Further research along these lines is clearly needed. The effectiveness findings regarding social outcomes such as social cohesion, although not quite as exiguous as those on health outcomes, are also highly inconclusive.

Thus, the available evidence base does not allow us to recommend any particular crime-related intervention or policy from the point of view of mental health, or of health and well-being more broadly considered. In general, the findings of the theory review and the effectiveness review do not indicate that narrowly focused crime reduction interventions are likely to be an effective or cost-effective way to improve well-being. However, the findings of the effectiveness review on reducing the fear of crime do provide some pointers as to how crime-related policies and interventions might impact on well-being, with the caveat that this implication relies on the links between fear of crime and well-being reported in the theory review, which are not based on a fully robust systematic synthesis of the evidence.

Environmental improvements

Arguably the single most promising category of interventions in the effectiveness review consisted of small-scale environmental improvements in public areas. It should be remembered that the effectiveness evidence here consists only of two, relatively low-quality studies (Cohen, Palmer), with methodological limitations in a number of domains including selection bias and study design, and which demonstrate effectiveness only for perceived risk and not affective fear outcomes. Nonetheless, it is suggestive that their findings are congruent with the qualitative evidence and with the relevant theory. These findings are particularly encouraging as such interventions are generally low cost and relatively easy to implement. In addition, the quality of life impact of minor environmental improvements is already on the policy agenda, in the shape of initiatives such as the Community Payback scheme (see www.gov.uk/community-sentences/community-payback; accessed 9 December 2013).

Further research on these interventions would be of value to more rigorously evaluate effectiveness, identify which components contribute to the intervention impact and establish whether or not there are impacts on broader well-being outcomes. As already noted, the most promising locations appear to be heavily used public areas, such as transport hubs, rather than primarily residential areas. The evidence suggests that addressing visible ‘eyesores’ such as graffiti and litter should be the main priority; it may also be of value to consider factors such as lighting, visibility and signage. The qualitative evidence also indicates that a number of more specific factors in particular environments may have an impact on fear. Hence, the design of interventions may benefit from integrating qualitative evidence, both the general evidence base considered in this review and, if possible, further formative research conducted with reference to the specific location where the intervention is planned.

Integration of fear of crime reduction with broader initiatives

Although the effectiveness evidence is inconclusive – and leaving aside the challenges identified in the previous section – the qualitative and theory reviews suggest that narrowly focused interventions to reduce fear of crime (such as intensified policing, or the provision of information about crime) are less likely to be effective than interventions that address either general problems in the environment (e.g. environmental improvements, housing renewal, and urban regeneration more generally) or the social, economic and political determinants of fear. The latter might include, for example, interventions to promote social cohesion, reduce alcohol and drug use, address racism and other forms of discrimination and promote the empowerment and decision-making capacity of communities. The absence of a clear message from the available data on the effectiveness of these broader interventions is problematic, but the other data considered in this project indicate that such strategies are promising ways to reduce fear of crime, and further evaluation research would be valuable.

Inequalities and targeting of interventions

The qualitative review and elements of the theory review indicate that different groups within the population vary substantially in the determinants of their fear of crime and in the extent to which the latter affects well-being. There is little reason in the effectiveness evidence to think that any of the included interventions increase inequalities in fear of crime outcomes. It is unclear whether this is the case with respect to health and well-being outcomes. However, there are some promising (if inconsistent) effectiveness findings regarding avoidance behaviours. Given that the qualitative evidence (see Fear and well-being) supports the latter more than any other putative mechanism linking fear of crime and health and well-being – and suggests that it may generate inequalities, particularly by gender – this would be a valuable focus of future research.

Many interventions included in the review were targeted at high-crime and/or low SES areas, broadly defined. Although there are obviously good reasons for this, the qualitative evidence suggests some potential drawbacks. First, the social stigma attached to areas with a bad reputation is substantial and may have damaging impacts in its own right, and it is possible that intensive crime-focused interventions could exacerbate this. Second, the points made earlier about the relation of fear of crime to crime are also relevant to the question of targeting. Many participants recognise that their local areas suffer from high crime rates but express low levels of fear, because of the protection offered by their social insidership. Conversely, many residents of low-crime areas express considerable fear, which impacts substantially on their behaviour. Although the dynamics here are complex – for example, low-fear but high-crime areas might be accounted for in terms of residents ‘normalising’ their circumstances in a potentially problematic way – it is clear that areas or communities at high risk of crime are not necessarily those at high risk of fear of crime. Third, many participants are explicitly critical of what are perceived to be cosmetic fixes that do not address the underlying causes of crime and fear of crime. This perception particularly emerges from the data on CCTV, but there seems to be a more general point that superficial or ill-thought-out attempts at intervention can reinforce a sense that a community’s problems are unimportant to decision-makers. (This should also be entered as a caveat against our recommendations on environmental improvement made earlier; see Environmental improvements.)

One population subgroup of particular interest is people with mental health problems. Although the data are not conclusive, there is reason to suspect that a substantial part of the observed population-level association between fear of crime and well-being may be accounted for by a strong association among the relatively small subgroup of the population who have clinical or subclinical mental health problems. In addition, and still more speculatively, people subject to repeat victimisation may suffer particularly strong well-being effects. If so, interventions addressing these populations specifically might be a more effective and cost-effective way to improve well-being through reducing fear of crime than interventions aimed at the whole population.

The policy context of fear of crime

As already noted at various points, there appear to be strong suggestions in all four components of the project that a more integrated approach is promising from the point of view of addressing fear of crime. To date, policy and research on fear of crime have tended to be strongly bound up with that on crime; the great majority of research has been carried out by criminologists, and the policy agenda on fear of crime has been almost exclusively driven by agencies with a main interest in crime and policing. Hence, the reduction of fear of crime has usually been pursued as a beneficial side effect of reducing crime, rather than as a primary goal in its own right.

There are obviously good practical reasons why fear of crime has remained largely within the orbit of crime. However, this situation is potentially problematic, as many policy-makers and researchers working in crime-related fields are well aware, because of the absence of any reliable link between crime and fear of crime. Both the observational quantitative data and the qualitative evidence, as well as the experiences of practitioners, illustrate how fear of crime does not closely track crime rates. Because of this, it is prima facie questionable whether or not a strongly crime-focused approach, led by police forces or by other stakeholders with a main focus on crime reduction, is the best way to reduce fear, and this scepticism tends to be borne out by the effectiveness evidence. (Indeed, there are serious questions as to whether or not currently widely used approaches to environmental crime prevention have been rigorously shown to be effective even with respect to crime outcomes; however, these questions lie beyond the scope of this review.)

More specifically, as Farrall et al. 49 observe, the centre of gravity of the fear-of-crime agenda in the UK – at least as far as evaluation is concerned – has shifted away from academic research and towards policy stakeholders since the mid-1990s. One side effect of this shift has been to dissociate evaluation research on fear of crime from broader sociological perspectives, which have largely been left to the critical qualitative tradition, and to shape it more in accordance with policy agendas. This raises a potential concern about the independence of this research, particularly given the findings of the effectiveness review that very few studies use robust designs that can minimise bias.

Some commentators on policy have suggested that this has made fear of crime research somehow complicit with punitive and authoritarian policy agendas on crime. This is a real concern, and the history of policy presents a number of examples of how research and theory on the well-being impacts of crime have been used to support such agendas, such as the ‘broken windows’ idea (see Chapter 3, Built environment and crime). On the other hand, our interviews with stakeholders (see Appendix 12) suggest that the opposite may often be the case, in that punishment-centred approaches to crime tend, in practice, particularly when there is pressure on resources, to crowd out any concern with fear, or more generally with the broader impacts of crime on communities. The interview data suggest that this has particularly been the case in the UK over the last couple of years, such that fear of crime has become a less central concern for decision-makers than in the 2000s, when, for example, fear of crime was explicitly included as a target in public service agreements.

Even where fear of crime remains a concern, the shift towards more policy-oriented research has arguably obscured its specific nature and significance. The goal of reducing fear of crime has often not been clearly distinguished from other goals to do with changing public attitudes, in particular promoting more realistic views about crime rates and risks, raising awareness of successes in crime reduction, and promoting more positive attitudes to the police and the criminal justice system. This has led to a certain amount of confusion about how research findings may inform policy. Of course, there are links between these broader attitudinal factors and fear of crime, as several studies have shown (see Chapter 3, Individual attitudes). Nonetheless, such factors are not the primary determinants of fear of crime. To the extent that attitudes to crime are linked to fear of crime, the broader social meanings of crime appear to be more important than views about its prevalence or the effectiveness of policies to deal with it, or about police–community relations. In addition, although these latter attitudinal outcomes may have some value for evaluation purposes, they seem prima facie unlikely to have any well-being implications. The widespread conflation of fear of crime with such broader factors has not been conducive to a focused attention on what reduces fear.

Thus, the messages implied by research and theory on fear of crime – including here not only the more critical work conducted since the early 1990s but also the predominantly quantitative observational studies of the 1970s and 1980s – have not been widely taken up in a policy and practice context. Although academic researchers have continued to advocate a dissociation between crime and fear of crime at the conceptual level, and to see the latter as a broad complex of environmental perceptions that may not be closely related to crime, the policy and practice implications of such a dissociation have not been thought through in detail. The available evidence and theoretical reflections on fear of crime suggest that there is potentially value in making fear of crime a more mainstream part of the agenda in other policy fields, particularly public health, but also urban planning, social policy and social inclusion, so that the broader social and policy determinants of fear can be more effectively addressed.

There are some indications in the literature (including the existence of this project) that such a shift has already begun, and that fear of crime is becoming more recognised as a dimension of the well-being agenda in its own right, rather than as a subsidiary dimension of crime reduction. Combined with the suggestions that fear is becoming a less central issue for policy-makers and practitioners in crime-related fields, this suggests that a more general reorganisation of the evidence and policy landscape may be under way. Thus, this would seem to be an opportune moment to consider how policy and research on fear of crime might develop to take account of this changed context.

Implications of the findings for research and policy

Within research and policy on crime, the findings of this review indicate that – to the extent that fear of crime remains a policy concern – a broader recognition that reducing crime and reducing fear of crime often do not go hand-in-hand, and may even conflict, would be valuable. Further policy guidance and research to determine the relative importance of the two goals would be valuable here. This might, for example, include further work on costs and cost-effectiveness along the lines of Dolan et al. ’s work,122,234 or, more generally, further research to enable comparison of the well-being consequences of crime and fear of crime on a common metric. As discussed earlier, it is also important to distinguish fear of crime from more general attitudinal measures to do with crime and policing.

In terms of public health and social policy research more generally, the consequence of seeing fear of crime as a dimension of well-being would seem to be to encourage its use as part of a battery of measures examining people’s perceptions of the social and physical environments, along with measures such as social cohesion, perceived trust, satisfaction with the environment and so on. Indeed, several of the studies of non-crime interventions in the effectiveness review do exactly this. Further uptake of fear of crime as an outcome measure in evaluations of social and environmental interventions – particularly along with measures of health behaviours and health status – would be a valuable contribution to the evidence base.

However, such an increased uptake of fear of crime as an outcome measure is likely to face a number of challenges. One is the difficulty of translating between distinct disciplinary ‘framings.’ As discussed in the review of theory, the irreconcilability of the different academic discourses required to understand the impacts of crime on health and well-being corresponds to serious differences between the conceptual frameworks used by different groups of stakeholders. In particular, the relatively narrow range of approaches that predominate in crime- and policing-related fields may pose a problem for communication with decision-makers in other policy and practice areas. This may partly account for the issue identified by Perry130 that approaches widely recognised as valuable within public health, such as universal primary prevention, have generally not been taken up within crime-related fields (see Chapter 3, Crime and health).

Moreover, as discussed earlier (see Chapter 3, Fear of crime: measures and contexts), there are serious unanswered questions as to how to interpret fear-of-crime outcomes. Arguments for the basic incoherence of the concept have been repeatedly made, and do not seem to have become substantially less valid over time. The meaning of even strong trends in the empirical data on fear of crime, such as the substantial and steadily maintained year-on-year decline in worry about crime in the UK over the last 15 or 20 years, remains fundamentally unclear. British Crime Survey data310 show worry about burglary declining steadily from a high of 26% in 1994 to 10% in 2010–11, and worry about violent crime declining similarly from 25% in 1998 to 13% in 2010–11. As noted earlier, the utilisation of the concept in the context of policy and practice has also been beset by ambiguities. These apparently insoluble problems have given rise to a widespread sense that the fear-of-crime agenda no longer provides an adequate theoretical framework, and that a transition to a more inclusive concept is required. 71

Thus, it may be challenging to integrate fear of crime into a more general picture of environmental impacts on well-being, as it is still unclear what the environmental determinants of fear of crime really are, and what aspects of fear impact on well-being. Although we make a few suggestions below, it seems unlikely that the controversies around this issue will be settled in the foreseeable future. In other words, although there is reason to think that measuring fear of crime can help to access some dimensions of the environment–well-being link, it is far from clear exactly which dimensions these are. To some extent, these questions can be addressed only by further empirical research. However, it must be admitted that the long but inconclusive history of fear-of-crime research to date does not inspire confidence in the outcome of such a process.

Crime Prevention through Environmental Design

As discussed earlier, CPTED theory on fear of crime has rarely been developed in its own right, distinct from the theory on crime. The assumption has usually been that measures effective in reducing crime will also be effective in reducing fear. Moreover, the extent to which CPTED theory is borne out by the evidence with respect to crime outcomes is itself questionable. Nonetheless, CPTED theorists do put forward a coherent and detailed theory of environmental influences on fear of crime. The findings of this project bear out the predictions of CPTED theory in some areas, but indicate its inadequacy in others. In the following sections we set out the relevant findings according to the three foundational concepts of CPTED theory: access control, surveillance and territoriality.

Fear and rationality

As discussed in the theory review, some form of dualistic distinction between ‘rational’ and ‘irrational’ or ‘symbolic’ fear has been widely utilised as a fundamental distinction in the theoretical literature. Rational fear would be fear that responds to actual risk (i.e. in which the affective component is purely driven by the cognitive component), whereas all other fear – which has been hypothesised to be driven by a wide range of factors (see following section) – would then be irrational. This framing of the issue has perhaps been a factor in some researchers’ and policy-makers’ reluctance to separate questions about fear of crime from questions about crime (as we argue is necessary above; see The policy context of fear of crime), from an understandable desire to avoid labelling people’s fears as irrational or ‘merely’ symbolic. However, many theorists have argued that this distinction is itself an oversimplification, and fails to take account of how estimations of risk and affective reactions actually inter-relate in the context of lived experience.

Some thinkers on fear of crime have attempted to produce revised versions of the distinction between the kind of fear of crime that responds to immediate risk and the kind that expresses broader symbolic resonances, which can take account of these critiques, such as Jackson’s111 distinction between ‘experience’ and ‘expression’. However, it seems clear that such revisions still face problems in integrating existing theories premised on an absolute distinction between rationality and irrationality with a more critical perspective that would see rationality in more pluralistic or context-sensitive terms.

We would argue that the qualitative evidence tends to show the impossibility of isolating the rational dimension of fear from the symbolic dimension, for several reasons. One has to do with the complexity of the causal pathways, as discussed in the theory review. Even taking into account the various distortions that may affect cognitive assessments of risk, the latter do not explain more than a small amount of the variation in affective responses, and a wide range of other factors may impact on people’s emotional experiences. Some factors (e.g. conditions in the physical environment) may be linked to both the cognitive and the affective aspects of fear in distinct ways, and interact in unpredictable ways depending on contextual or individual-level determinants. The practical difficulty of generalising about these pathways, such that the ‘rational’ core of fear, represented by the impact on affective fear from cognitive risk perceptions, could be isolated from the ‘irrational’ components represented by the impacts of other factors, is highly challenging.

More fundamentally, however, the identification of the cognitive component of fear with the perceived risk of victimisation is itself conceptually questionable. To the extent that fear responds to a range of environmental factors other than objective risk, its cognitive dimension cannot be limited to the estimation of the latter, but includes a wide range of situated knowledges about the conditions in the social and physical environment, at a range of scales. A particularly important dimension of this, as discussed later (see Inequalities and ‘spirit injury’), is that the objects of the cognitive dimension of fear include the prevailing structural inequalities and relations of political and socioeconomic domination within a society. Women’s fear of rape, for example, may in many cases be a rational, cognitively grounded response to social conditions, in ways that are obscured if it is simplistically juxtaposed to objective rates of rape. As argued by the ‘risk society’ theorists, then, what counts as risk – and hence what counts as a rational response to the risk – is open to negotiation in the context of particular experiential regimes, and cannot be defined a priori by statistical generalisations.

Moreover, as the theory review suggests, it is primarily affective outcomes that are of importance from the point of view of well-being outcomes. Given the absence of a straightforward conceptual or empirical link between affective fear and perceived risk, it seems theoretically implausible that the latter is a major determinant of well-being in its own right, and the qualitative data tend to confirm this. Actual victimisation certainly has negative impacts on well-being, sometimes very serious impacts. The mere fact of risk, however, appears to have little impact on well-being unless it is made experientially salient. Salience may be increased by a number of factors, including features of the physical and social environments, but is largely independent of risk: the most experientially salient crimes in terms of fear (rape and child abduction) are usually recognised to be very rare occurrences. (This may depend on background risks being relatively low, such that most individuals are unlikely to be the victims of serious violent crime over timescales of years. In contexts with higher rates of serious violent crime, statistical risk may impact on well-being in the sense that more individuals suffer the direct effects of victimisation.) Thus, we would argue that statistical risk is not a useful way to think about fear with respect to well-being outcomes.

Fear as symbolic attractor

The absence of a strong link between risk and fear has led many theorists to search for the ‘missing’ drivers of fear: some have focused on specific determinants such as vulnerability, whereas others have cast their net much more broadly. Here we consider three such broader theories: first, the theory that fear reflects anger about the symbolic social meanings of crime, associated particularly with Stephen Farrall and colleagues; second, the idea that fear reflects broader insecurities about status, associated particularly with Ian Taylor; and, third, Hollway and Jefferson’s theories on the psychodynamics of fear.

Social cohesion and collective efficacy

Social cohesion and collective efficacy have been posited as fundamental determinants of fear of crime by a number of researchers. Our findings strongly bear this out on a general level in that social connectedness and interaction, and a more general sense of being ‘at home’ socially within a given community, is a substantial protector against fear of crime. Familiarity with one’s own social environment, and particularly the sense of rootedness in it which comes from long-term residency, help to reduce fear considerably. Some theorists have raised the concern that social interaction could exacerbate fear by spreading concerns about crime; our data suggest that this possibility is largely outweighed by the value of such interaction in increasing cohesion, although several participants do mention talk about crime as potentially increasing fear.

However, a number of points should be raised against any notion of social cohesion as a panacea in dealing with crime. The protective effect of social cohesion appears to be limited in its spatial extent, and to operate usually only in one’s immediate residential neighbourhood (it also appears to rely on long-term residency in an area). Because there is some reason to think that fear with reference to the immediate local neighbourhood has more substantial impacts on well-being than fear with reference to other areas, this is a relatively minor point, but should be borne in mind.

There is also a ‘dark side’ to social cohesion, as manifest in the data about self-policing communities. Even where cohesive communities do protect against fear for insiders, they may have exactly the opposite effect for outsiders. More generally, strong social cohesion depends on a shared set of values at a community level, which may conflict with the values officially espoused by policy-makers and other outsiders, and which may in many contexts provide social capital to offenders (again, the self-policing data are the most dramatic illustration of this). As a number of theorists have observed, then, social cohesion should not be regarded uncritically as a good.

Incivilities

The role of both physical and social ‘incivilities’ as drivers of fear is clearly substantial. Physical incivilities such as graffiti and litter, and social incivilities such as public drinking or drug use, are frequently cited in the qualitative data as determinants of fear; more tentatively, the importance of physical incivilities appears to be borne out by the effectiveness data.

However, our findings strongly suggest that physical and social incivilities are two entirely different phenomena, and the mechanisms of their effect on fear of crime are very different, such that it is misleading to include them both under the same term. As incivilities theorists argue, physical incivilities appear to drive fear as indicators of neglect at a community level: they show that an area is not well cared for and that social norms are weak. They also – a point less elaborated in the theory – function as symbolic indicators of low SES, high-crime, ‘rough’ areas; as such, they operate similarly to other features of the built environment such as high-rise housing, which usually do not constitute ‘incivilities’ at all, but simply serve to provide (correct or incorrect) information about the character of a neighbourhood.

By contrast, social incivilities drive fear mainly because they involve people seen to be unpredictable and threatening in themselves. We postulated earlier that drinkers, drug users and young people ‘hanging about’ are seen as threatening because they do not conform to norms concerning the use of public space. Whether this explanation is accepted or not, it is clear that these groups are feared directly, and specifically regarded as likely to commit crimes; their role as indicators of environmental conditions appears to be of much less, if any, importance. It should also be noted that, although perceptions of what constitute physical incivilities seem to be fairly consistent across the population, the perception of social incivilities is relative to expectations that may vary substantially between different population groups. (Most incivilities research has ignored this point, using a purely etic concept of ‘incivilities’ that does not claim to directly translate emic categories.) Young people, in particular, appear to have different norms about the use of public space from older adults, and labelling one side of this conflict ‘incivilities’ detracts attention from the conflict of norms itself as a driver of fear. We would thus suggest that, although the theory that incivilities generate fear is well grounded in many respects, the concept itself conflates distinct phenomena and is probably not useful.

Inequalities and ‘spirit injury’

Theories of ‘spirit injury’ posit that the well-being effect of crime at a community level is mainly due to its effect in maintaining social inequalities (e.g. of gender, ethnicity, SES or age). The symbolic resonances of crimes involving discrimination thus amplify risk and contribute to a pervasive sense of unsafety. Thus, according to these theories, the impacts of crime cannot be understood without referring to the broader social structures in which crime takes place, particularly structures of domination.

A substantial amount of qualitative evidence appears to support these theories. Again, the idea of salience introduced earlier (see Fear and rationality) helps to clarify the mechanism here: the well-being effects of fear operate less through the recognition of an abstract statistical risk than through other factors that maximise the salience of the risk. Spirit injury theory can then help to illuminate how the latter are socially patterned, such that the risk of crime resonates with a broader set of mechanisms that maintain the structural violence of social inequalities. For example, qualitative data from women, BME people and lesbian and gay people indicate that the everyday mechanisms through which inequality is perpetuated – from minor harassment or discriminatory remarks, to crimes such as indecent exposure and the threat of violence – function as reminders of the risk of crime in a way that does not form part of the experience of other groups. Spirit injury theory thus points to the way in which these experiences drive fear as part of a systemic apparatus of inequality. Other theories focus on factors such as differences in perceived vulnerability, or biases in media reporting, to explain differences in groups with respect to fear (particularly gender differences). Although there is some evidence that these play a role, there does also seem to be a need to take into account the social practices that reinforce such differences and reproduce them as inequalities.

Fear and well-being

The theory review distinguished four pathways through which fear of crime might be related to health and well-being outcomes: (1) fear has direct impacts on mental health; (2) mental health has direct impacts on fear; (3) fear has impacts on behaviour, leading to (a) reduced social interaction and (b) reduced physical activity; and (4) fear may reduce social cohesion at a community level and thus well-being.

The qualitative evidence is equivocal with respect to pathway (1). On the one hand, few participants explicitly report serious mental health problems linked to the fear of crime (there is more evidence for the mental health impacts of actual victimisation). On the other, there are methodological challenges in using qualitative evidence to illuminate impacts on subclinical mental health outcomes, particularly as few of the included qualitative studies took this as a main focus. It is clear that many people have a pervasive concern about the risk of crime and it is plausible that this concern has impacts on well-being.

Pathway (2) is supported by Whitley and Prince’s study (Whitley), which indicates that people with mental health problems are more seriously affected by fear of crime than others; however, we located no other studies of this population. Pathway (3), particularly (3b), is strongly supported by the qualitative data, with many participants reporting that they limit outdoor physical activity as a result of fear of crime. This is particularly true for women, and also for children and young people, whose activities are limited by their parents’ vicarious fears. Pathway (4), as discussed in the theory review, is by nature hard to evaluate on the basis of the quantitative observational data. It is also hard to relate to the qualitative data, as the salience of crime and perceptions of social cohesion are to some extent bound up with prevailing social norms in a community, and it is hard to specify their effect independently of the latter. The findings of this project, overall, do not strongly support pathway (4), but also do not rule it out.

The effectiveness data are not very illuminating here. Only two studies measure the effect of an intervention on both fear of crime and mental health outcomes: one shows an improvement in both (Halpern), the other an improvement in fear of crime but not mental health (Brownsell). Several measure avoidance behaviours as well as fear of crime, which might help to illuminate pathway (3), but the results are very mixed and inconclusive (see Appendix 8). In general, therefore, although the systematic reviews support the idea that fear of crime may be a barrier to physical activity, they otherwise add little to what was already established in the theory review, namely that fear of crime does have some impact on well-being, but its extent is probably limited in most populations, and the pathways through which it operates are complex and unpredictable.

Using theory to inform the review of effectiveness

Much of the value of the review of theory in the context of the systematic reviews was to indicate where important conceptual distinctions needed to be made in interpreting the findings of the effectiveness review. Two areas were particularly important here. First, the substantial body of theoretical and empirical work around the measurement of fear of crime drew attention to the potential issues involved in synthesising research using distinct outcome measures (although not all of these issues could be fully addressed within the review of effectiveness itself). Second, the theory review helped to clarify the existence of at least two largely independent pathways from environmental determinants or interventions to well-being outcomes, one that goes via environmental perceptions and thence the fear of crime, and one that goes via actual crime rates. Thus, the theory review facilitated a more adequate picture of the underlying causal dynamics that may be affected by interventions. Both of these clarifications helped to guard against an overconfident interpretation of the effectiveness findings.

In addition, the review of theories helped to provide a more explicit basis for identifying important questions that have not been adequately addressed by the empirical research (e.g. the effectiveness of crime-related interventions in improving health and well-being). This helped us to formulate more focused and practical recommendations for research (see Implications for future research).

As discussed in Chapters 2 and 3, the theoretical map should be regarded as indicative only, and the methods used are of limited validity in some respects. We were unable to test the findings of the theory review against the effectiveness data because of the limitations of the latter. Future work on the use of reviews of theory might seek to test whether they can be validated against empirical quantitative evidence. For example – along the lines of the EPPI-Centre methodology for mixed-methods synthesis368 – one might use the theory review to generate hypotheses about the differential impact of aspects of intervention delivery and then test these by conducting subgroup analyses of the effectiveness studies; alternatively, data from the intervention studies on mediating variables identified as important in the theory review could be used to test hypotheses about causal pathways.

These difficulties aside, the theory review was invaluable in identifying where robust conclusions could or could not be drawn from the effectiveness evidence. Overall, our experiences support the value of reviews of theory in evaluating the effectiveness of complex interventions, although they also identify a number of potential difficulties. Further methodological work would be valuable here. In particular, it would be helpful to look at reviews of higher-quality effectiveness data, to explore how more sophisticated methods of quantitative synthesis – including standard meta-analysis, but also other techniques, such as mediator analysis – might inform and be informed by reviews of theory.

Using theory to inform the review of qualitative evidence

The relation of the review of theories to that of qualitative evidence was quite different. One reason for this is that, although we maintained a strict separation between the qualitative review and the theory review at the methodological level, there is considerable overlap in content, as many qualitative studies are theoretically informed and many theories are primarily based on qualitative data.

We adopted a two-stage approach: first the qualitative data were synthesised in a thematic analysis and then the findings of this analysis were brought together with the theoretical material in a cross-study synthesis. This contrasts, for example, with meta-ethnography, in which both ‘first-order’ data (participants’ views) and ‘second-order’ data (primary study authors’ interpretations) enter into the ‘third-order’ synthesis. A meta-ethnographic approach to this project, then, would not have included a separate review of theory, but would have relied on drawing out the theoretical implications from the analyses reported in the qualitative studies. [To some extent, these comments also apply to other qualitative synthesis methodologies incorporating a stage of ‘meta-theory’ construction (e.g. Paterson et al.’s ‘meta-study’369); we focus here on meta-ethnography as the most widely used such method.]

Our approach has certain advantages relative to meta-ethnography. First, it allows a broader synthesis, including not only the theoretical constructs that happen to be reported in the included qualitative studies but also a considerably wider range of theories, as well as other kinds of data (e.g. quantitative observational studies). This allowed us to consider how the qualitative evidence might productively engage with broader theoretical debates (e.g. in risk psychology or CPTED theory) and not just with those that have been pursued most enthusiastically by qualitative researchers themselves. Second, our method provides a way to combine the strengths of different methodologies and reconcile the ‘positivistic’ goals of transparency and reproducibility with the richness of more theory-led approaches. Finally, it arguably allows for a greater appreciation of fundamental theoretical conflict than do meta-ethnographic methods, and does not require an ultimate reconciliation of divergent views. (This is a controversial point. Noblit and Hare37 emphasise the idea of ‘refutational synthesis’, but this has received little attention in the subsequent development of meta-ethnographic methods, as Noblit himself recognises. 370) More generally, it is unclear whether the idea of refutational synthesis can encompass some of the fundamental conflicts found in our review of theories. On the other hand, it is possible that a meta-ethnographic analysis of our data would bring out richer or more detailed connections between theory and data that were missed by treating the two separately.

Again, a number of methodological questions remain open. However, unlike the situation with effectiveness data, for which methodologies for the integration of theory remain at a relatively early stage, existing debates in qualitative methodology have already addressed many relevant points. We would welcome a critical engagement with our approach from researchers involved in qualitative synthesis or primary qualitative research to clarify the value of reviews of theory in this context.

Primary research

The findings of the review identify a number of areas in which further primary research would be valuable. Most obviously, there is a clear need for robust primary studies of the health status and health behaviour impacts of environmental crime reduction interventions.

It may be possible to address this gap to some extent on the basis of the existing data through the use of modelling approaches, either conventional epidemiological or economic models, or newer methods such as agent-based modelling. Although we have been able to reach some tentative conclusions by informally combining effectiveness data on fear of crime with observational evidence on health and well-being outcomes, more rigorous quantitative work on this would be valuable.

The evidence base on interventions to reduce fear of crime is of limited reliability. More robust studies, using matched control groups and appropriate data analyses, are required. Studies of all of the intervention types covered in the review would be of value. A further improvement to the evidence base would be the greater use of validated measures for fear of crime outcomes. However, given the ongoing controversy over what fear of crime outcomes actually measure, it is rather unclear what this would mean in practice.

Systematic reviews

A systematic review of observational research linking fear of crime with health and well-being outcomes would be valuable. There is a substantial body of evidence here and this is currently an active field of research. A review would be valuable in establishing what is already known on this topic and, in particular, in providing a robust clarification of the evidence on the pathways through which these outcomes are connected, to inform future research. However, as with any question regarding fear of crime, such a review would need to be carefully conducted to ensure that distinct measures of fear are kept separate and not conflated. (Reviews of observational data on the social and physical environmental determinants of fear of crime, or on the crime–fear relation, would probably not be useful, for two reasons: the heterogeneity of measures is such that it is unclear whether any useful conclusions could be reached; and these are generally less active research agendas, so the practical value of the reviews is less obvious.)

The environmental determinants of crime outcomes would also be a potentially useful research question for a systematic review. This said, our findings suggest that the available evidence base is mostly of low quality and may not support substantive conclusions. Nonetheless, with appropriate safeguards (e.g. a reasonably stringent quality threshold), such a review might make a useful contribution.

Systematic reviews of the effectiveness for fear outcomes of other intervention types not included in our review of effectiveness would be valuable to complement our findings. In particular, there is a substantial body of research on the fear of crime impacts of interventions concerned with changing policing practice (e.g. community-oriented policing), and a certain amount on the provision of information (e.g. crime maps), and to our knowledge no systematic evidence synthesis has been undertaken on these questions. Such reviews would help to extend and complement the findings of this review, although they would also face similar challenges in terms of the synthesis and interpretation of findings. In principle, it would also be useful to know about the evidence on the health and well-being impacts of such interventions, but it seems likely that, as for our review, very little such evidence exists.

Finally, there is scope for other systematic reviews of qualitative evidence. Most obviously, our qualitative review covered evidence from the UK only, and experiences in other countries may differ, particularly countries with higher rates of violent crime. In addition, the qualitative review suggests a number of other potential questions for systematic reviews, including the experiences of victims of crime; perceptions of the urban environment and disadvantage; perceptions of policing; the social meanings of crime and disorder; and parents’ and children’s views of safety (in general, that is, not limited to crime). In general, the methodology of the systematic review of qualitative evidence has not been taken up within crime-related disciplines nearly as much as it has in health-related disciplines. Thus, there are numerous areas in which substantial bodies of good-quality primary qualitative evidence exist, but in which no systematic reviews appear to have been conducted. In addition to the substantive value of such reviews, the increased uptake of qualitative synthesis methods in criminology and the sociology of crime would also make a valuable contribution to the methodological literature.

Selection bias

Study design

Confounders

Blinding

Data collection

Withdrawals and dropouts