Fear of crime and health in residential tower blocks

EUROPEAN JOURNAL OF PUBLIC HEALTH 2002; 12: 10–15

Fear of crime and health in residential tower blocks A case study in Liverpool, UK GEOFF GREEN, JAN M. GILBERTSON, MICHAEL F.J. GRIMSLEY *

Background: Though it is often assumed that fear of crime erodes mental health, research evidence is limited. Our study seeks to assess the relationship between these attributes in residents of the city of Liverpool. Method: Evidence is drawn from a sample survey of 407 adults living in 21 tower blocks. A number of social and psychosocial attributes linked with feelings of safety are compared with self-reported health status using logistic and multiple regression techniques. Possible reciprocal relationships were investigated using two-stage least squares. Results: Fear of crime in this sample is generally much lower in the home than in Britain as a whole and much higher out on the neighbouring streets at night, but there are sub-group variations. We find significant associations between fear of crime and health status. Feelings of safety when out alone after dark is the most consistent predictor of health status. Those feeling safe score significantly higher on all five dimensions of the SF-36 measure which cover mental and social well-being. Mental health is the strongest correlate and is probably a consequence rather than cause of feelings of safety. Conclusion: The evidence suggests elderly residents believe tower blocks provide safe accommodation. However, feelings of safety in these ‘fortresses’ do not generally extend to walking in neighbouring streets. Fear of crime erodes quality of life and is associated with poorer health. Keywords: fear of crime, housing, psychosocial determinants of health, SF-36, stress

I

10

n a wide-ranging review of evidence commissioned by the UK Government, the Report of the Independent Enquiry into Inequalities in Health asserts that ‘Crime and the fear of crime affect profoundly the quality of people’s lives … Fear of crime can also be a cause of mental distress and social exclusion’.1 According to a report of the UK Public Health Alliance2, ‘The effects of the fear of crime, rather than crime itself, on the health of individuals and communities has been largely underestimated’ but ‘Available research into the impact of crime and fear of crime on public health is limited’. Linda East contrasts populist perceptions of such links in deprived communities with academic silence on the issue, finding ‘little published literature within the health domain which explicitly explores the connections between crime, fear of crime, anti-social behaviour and health’.3 This paper provides new evidence drawn from a study of Housing Action Trust (HAT) residents, which aimed to investigate the impact of new housing design on their perception both of personal safety and health. We hypothesised that a relatively deprived population, after moving from tower block flats in Liverpool into new houses built on ‘secure by design’ principles, would feel * G. Green1, J.M. Gilbertson1, M.F.J. Grimsley2 1 Centre for Regional Economic & Social Research, Sheffield Hallam University, UK 2 Division of Applied Statistics, Sheffield Hallam University, UK Correspondence: Geoff Green, Centre for Regional Economic & Social Research, Sheffield Hallam University, City Campus, Sheffield S1 1WB, UK, tel. +44 114 225 4524, fax +44 114 225 2197, e-mail: [email protected]

less fearful of crime and, in turn, experience better mental and emotional wellbeing. This route to health is shown schematically in figure 1 as pathways 3 then 1. The linkages between housing estate design and feelings of security, pathway 3, are well rehearsed by Jane Jacobs4 and Oscar Newman5 in the USA and in the UK by Alice Coleman,6 Barry Poyner and Barry Webb.7 Their impact on health is more difficult to establish because of a number of confounding variables – ‘it must be like trying to accurately identify all the ingredients in a very thick soup’.8 Figure 1 shows some of these other ingredients, hypothesizing complex pathways (2 then 1) from social cohesion to health, particularly mental health. Research evidence, primarily from the USA, has established a relationship (pathway 2) between community cohesion and crime (Hirschfield and Bowers,9 Wilkinson et al.10). In the UK Michael Hough’s11 specialist study of anxiety about crime draws out the correlates of fear from the 1994 British Crime Survey. Fear of crime is highly correlated with crime itself but also linked to age, sex, class, neighbourhood and race. Though there is little systematic evidence linking fear of crime and health (pathway 1) a number of commentators link fear with stress and regard stress as mediating socioenvironmental determinants and health outcomes. Richard Wilkinson12 argues that health is affected as much by psychosocial processes associated with material deprivation as by the physical manifestation of such deprivation. Broadly agreeing, Jon Elstad states that

Relationship between fear of crime and health

Figure 1 Schematic links between fear of crime and health

‘socio-environmental demands – stressors – engender psychological stress, i.e. a troubled state of mind which can surface in many ways, as anxiety, fear, hopelessness, or anger’.13 East points to some evidence14 that fear of crime is a major cause of stress and Elstad has ‘few doubts that psychological stress, generated by despairing circumstances, insurmountable tasks, or lack of social support, can influence disease parameters’. These studies use multivariate analyses of large data sets to identify significant linkages. When the larger Liverpool study is complete it should also be possible to assess the impact of significant change in estate and house design by applying similar statistical methods to a smaller data set. The purpose of this paper is to explore pathways 1, 2 and 3 and identify other significant correlates of fear and crime by undertaking a cross-sectional analysis of baseline data of residents’ attitudes prior to their move to new accommodation. All of these residents live in similar tower-block apartments where security had been improved with ‘access and control’ measures such as intercoms and CCTV. We hypothesize that health – especially mental health – co-varies with fear of crime; and fear of crime in turn co-varies both with the degree to which residents can rely on neighbours and their level of satisfaction with their living accommodation. Pathway 1 is particularly complex. Although the main focus is on measures of mental health and wellbeing as outcomes (bold arrow), it can be argued that a reciprocal relationship is likely to exist. Anxiety and stress may influence mental condition but those suffering from poor mental health may be more fearful of crime because of their condition. It is also plausible that a reciprocal relationship may exist between fear of crime and social cohesion. In essence, figure 1 schematically models social and psycho-social patterns within a socially and educationally homogeneous sample population living in very similar tower-block accommodation. METHOD

Background and sample In 1993, Liverpool Housing Action Trust took responsibility for 66 of the 67 residential tower blocks owned by the municipality and occupied by over 3000 residents.

The larger study is designed to establish living conditions and assess perceptions of health both before and after some of these residents move to new purpose built houses and bungalows. It is a longitudinal study tracking 200 tower block residents into their new homes and comparing any changes in their health and quality of life with a control group of 207 residents who stay put for the relevant period. This smaller study analyses data from the baseline survey of all 407, generally elderly, residents undertaken in the winter and early spring of 1997. The sample was obtained from the population of all the adult residents living in 21 tower blocks on four estates in Liverpool who either expected to move within a year – ‘the experimental group’ or expected to remain in their current accommodation for at least three years – ‘the control group’. Data were collected using an intervieweradministered questionnaire. The response rate was 58% and the refusal rate less than 5%, with no contact made with the remaining 37% of residents. Contingency table analyses, binary logistic and multiple regression approaches, including two-stage least squares, were used to explore key pathways outlined in figure 1. The relevant operational study variables are summarized in table 1. Where possible, items were drawn from previous validated studies and designed to facilitate comparison of results. Some percentage responses are also given. So, for example, of the 407 who responded to the question, 68% reported suffering from limiting long-term illness. Outcome measures – pathway 1 The principal outcome measures from pathway 1 are four dimensions of health and a mental component summary measure, largely based on these, derived from the Medical Outcomes Survey Short Form 36 (SF-36) questionnaire. The SF-36 originated in the USA15 and has been anglicized for use in the UK.16 It measures health perceptions via 35 items across eight correlated dimensions and one item measuring health change. Responses to each item within a dimension are combined to generate a score from zero to 100, where 100 indicates ‘good’ self-reported health or wellbeing on each dimension. In order to make statistical analyses and interpretation more straightforward, the eight dimensions can be summarized into mental and physical functioning components (MCS and PCS) by a method based on principal component factor analysis.17 The MCS is most heavily weighted on i) the SF-36 Mental Health Index – a measure of nervousness and depression, ii) the SF-36 Energy/Vitality scale – a measure of tiredness, iii) the SF-36 Emotional Role scale – a measure of difficulties with daily activities because of emotional problems and iv) the SF-36 Social Functioning score, which measures interference with normal social activities because of physical or emotional problems. As these five measures of self-reported mental health and well-being were the main outcomes of interest, relationships between the physical components summary and the four remaining SF-36 physical and general health dimensions, though investigated, are not reported on here.

11

EUROPEAN JOURNAL OF PUBLIC HEALTH VOL. 12 2002 NO. 1

The reliability of the SF-36 instrument in terms of its internal consistency and stability over time has been tested and found to be satisfactory in meeting psychometric criteria.18–20 The instrument was also sensitive to differences in the health status of residents of improved and unimproved tower blocks as measured in a comparative study21 undertaken prior to the Liverpool investigation. The SF-36 was supplemented by the British General Household Survey question on limiting longterm illness.

12

Measures of social cohesion and physical structure – pathways 2 and 3 Social cohesion was operationalized using the BCS questions ‘Would you say in general your neighbourhood (or tower block) is one in which people do things together and try and help each other or one in which people go their own way?’ More residents (36% compared with 21%) responded ‘help each other’ when questioned about their ‘block’ rather than their neighbourhood. Physical structure was operationalized by three variables – those living alone (59%), satisfaction with their housing (24% were dissatisfied) and satisfaction with their domestic heating (43% were dissatisfied). Stress was identified from answers to a question specifically about the renewal or redevelopment process being undertaken by HAT and 18% of respondents reported that stress was considerable. Levels varied according to where residents were in the redevelopment programme but we assumed levels also reflected an underlying predisposition. All pathway models also incorporated the demographic characteristics of age and gender of subjects and their residential area of the city as a contextual factor. Ethnicity and social class, however, were not used as analytical variates. Some 98% of respondents described themselves

Measures of psychosocial aspects – pathways 1 and 2 ‘Fear of crime’ can be a misnomer. The conceptual issues and confusions in research on fear of crime have been well reviewed. Hough distinguishes perceptions of risk from fear, and fear itself from commonplace anxiety. In the Liverpool study, stomach-churning fear or corrosive anxiety was expected to have more influence on wellbeing than a resident’s cool assessment of risk. Ideally the survey should encourage residents to differentiate between the three. But this would have overloaded the questionnaire and taken the study further into uncharted territory. So measures of fear of crime were derived by ranking responses to validated questions from the Table 1 Model concepts and operator variables British Crime Survey (BCS) Concept Variable which Hough admits is ‘un- Physical setting deniably a blunt instrument’. a) Satisfied with domestic heating (noa/yes) Implicit fear of crime (as an b) Satisfied with housing (noa/yes) ‘all encompassing shortc) Living alone (yesa/no) hand’ for worry, anxiety and Social cohesion risk) was ascertained by d) Kind of neighbourhood - where people: (Help each asking residents ‘How safe othera/mixture/go own way) do you feel i) in your home e) Kind of tower block - flats where people: (Help each othera/mixture/go own way) alone at night or ii) walking alone in this area after dark?’ Psycho-social aspects f) Perception of crime change (more yesa/no) And, following the BCS Fear indicators: convention, for those who g) Safe out after dark (implicit) (noa/yes) never go out, interviewees h) Safe at home (implicit) (noa/yes) were asked ‘how safe would i) Worried about burglary (explicit) (yesa/no) you feel?’ Of study respondj) Worried about mugging (explicit) (yesa/no) ents, around 61% felt (or k) Stress (yesa/no) would feel) unsafe when out after dark whereas only 6% Mental health and wellbeing l) SF-36 Mental Health Index (MHI) felt unsafe at home. For the m) SF-36 Energy/Vitality (Energy) more explicit fear of crime n) SF-36 Emotional Role (Em R) aspects, 24% of the sample o) SF-36 Social Functioning (Soc F) expressed anxiety about p) Mental Component Summary (MCS) being burgled whilst 46% Other health were worried about being q) SF-36 General Health Perception (GHP) mugged. Such responses r) SF-36 Pain Index (Pain) were consistent with those s) SF-36 Physical Functioning (Phys F) for implicit fear of crime and t) SF-36 Physical Role (Phys R) also with the 43% who felt u) Limiting Long-term Illness (LLTI) (yesa/no) that neighbourhood crime v) Physical Component Summary (PCS) had increased over the previous two years. a: Answer to the question corresponding with given percentage

%

N

43

402

24

405

59

407

21

405

36

402

43

397

61

401

6

405

24

404

46

404

18

388

68

407


effects. After demographic and contextual factors had been accounted for, the only significant relationship was that between the implicit fear on the streets after dark (fear dark) and feelings about people in the neighbourhood. Even this was not clear-cut, however, as both those who felt people helped each other and those feeling neighbours went their own way were less fearful of being out after dark than those who thought their neighbourhood was a mixture. On a bivariate level, those who felt that people in their own tower blocks and those in the neighbourhood went their own way tended to be more fearful of being mugged or burgled. Thorough analysis of the property design effect on fear of crime (pathway 3) awaits comparative data from the second wave interviews of residents after they have moved to completely new physical surroundings. However, we undertook a modest cross-sectional analysis of residents’ levels of satisfaction with their apartments (housing) in general and their heating systems in particular homes and also of whether they lived alone as indicators of physical setting. Though there were more significant predictors for this pathway, relationships were linked to perceptions of crime change and stress levels rather than implicit and explicit fear of crime. Those who were not satisfied with their housing were more likely to perceive a recent increase in crime (odds ratio 1.97) and report greater stress (odds ratio 2.94). Those dissatisfied with their heating also reported more stress (odds ratio 2.06). In general, logistic analyses confirmed the demographic and contextual correlates of fear highlighted by Hough. Younger people were more inclined to fear burglary and the elderly were more fearful of being out alone after dark. Fear of mugging, however, was not related to age. On average, women were nearly three times as likely as men to express implicit fear of crime and fear of being mugged (odds ratios were 2.90 and 2.74 respectively). Finally, neighbourhood is another significant boundary in Hough’s analysis of British Crime data. It might have been expected to be significant in this study, since one

as white British and most were retired after a lifetime in manual-related occupations. RESULTS

National comparisons The headline results of the survey are that fear of crime for this sample of the tower block population of Liverpool is much lower in the home than in Britain as a whole22 and much higher out on the neighbouring streets at night. Table 2 summarizes the position of residents and compares it with results of an earlier MORI23 survey of HAT residents in 1993 (before certain security measures were introduced into the tower blocks) and with the picture in Britain as a whole. Analyses of pathways 2 and 3 A summary of results for the pathway between social cohesion and psychosocial aspects (pathway 2) is given in table 3. Simple bivariate analyses are used, with chisquare tests of independence and binary logistic models controlling for demographic, contextual and pathway Table 2 Fear of crime compared Very safe %

Fairly safe %

A bit unsafe %

Very unsafe %

Liverpool HAT 1993

55

34

6

3

Liverpool HAT 1997

70

24

4

2

Britain 1996

54

36

8

2

Feelings of safety home alone at night

Feelings of safety walking alone in neighbourhood after dark Liverpool HAT 1997

11

28

25

36

Britain 1996

25

43

21

11

Sources: HAT 1993 = MORI (n=2,032), HAT 1997 = CRESR (n=407), Britain 1996 = British Crime Survey (n=16,348). Comparable figures not available for HAT 1993 ‘walking alone after dark’.

Table 3 Summary of bivariate associations and binary logistic main effects for pathways 2 and 3 Demographic and contextual Area

Physical setting Heating satisfaction

Housing satisfaction

Social cohesion Living alone

Neighbourhood

Age

Gender

**

**

*

(*)

**

**

Tower block

Psychosocial Fear dark Crime change Fear burglary

(*) (–)*

**

**

*

**

(–)*

*

**

(*)

* **

(*)

*

*

(–)* Fear mugging

(–)*

** Stress

*

**

**

**

**

*

**

*

** Significant at 0.01; * significant at 0.05; (*) significant at 0.10 In each cell, upper indicators show the significance for bivariate associations, lower indicators show significance for binary logistic main effects.

13

EUROPEAN JOURNAL OF PUBLIC HEALTH VOL. 12 2002 NO. 1

estate is in a poor neighbourhood of Liverpool and the other three are located within relatively prosperous districts. Once other factors were taken into account, however, areas only differed with respect to perceived crime change; residents from the poorest neighbourhood were less inclined to perceive a recent crime increase, and reported lower stress levels. Respondents from an area where uncertainty about the redevelopment process was greatest, were over twice as likely to report considerable levels of stress. Analyses of pathway 1 The key question is whether fear of crime has independent association with health, specifically mental health and the SF-36 Mental Health Index (MHI) in particular. The picture is complicated by the likelihood of both a more direct route to health which skips the intermediary variable of fear and a reciprocal relationship between fear and mental well-being. There is a possibility, therefore, that fear of crime may simply be a dependent variable, contingently linked both to upstream factors and to health. In order to identify the independent effect of fear, multiple and logistic regression modelling was performed on the five relevant SF-36 indices and final models developed after initial stepwise procedures. All operator variables were investigated together with age, gender and area of residence. The significant main effects for the three appropriate multiple regression and two binary logistic models are summarized in table 4. This also includes, where relevant, the adjusted R-squared values. The direction of associations between the operationalized concepts and the MHI score, the main outcome of interest, were as anticipated. The psychosocial predictors showed the highest coefficients, with those expressing implicit fear of crime and higher levels of stress having lower scores on the mental health index, by, on average, 11 points and 10 points respectively. Clearly, stress has a significant effect on mental well-being independent of fear of crime. With regard to social cohesion, lower MHI scores were associated with those feeling that immediate, tower block neighbours do not tend to help each other. A two-stage least squares model was employed to investigate the likely reciprocal relationship between mental health and implicit fear of crime. The implicit fear variable (fear dark) was first modelled on the explicit fear variates which had weak associations with mental wellbeing indicators. This first-stage predicted implicit

fear variate was then employed in the full models as a replacement for the actual fear dark variable. The effects remained significant and these results are not given here. Of the three physical setting variates (a non-direct pathway), two showed independent significant effects. Those living alone and those unhappy with their domestic heating manifested lower levels of self-reported mental health. After allowing for the other predictors, increasing age appeared to be associated with better mental health. When all six elements were combined in our multivariate model, they accounted for approximately 15% of the variation in the MHI score. The psychosocial variables of stress and implicit fear of crime, together with age, were also significant predictors in the multiple regression models for Energy/Vitality and the mental component summary, with multiple correlation coefficients (adjusted R2) of R=0.32 (0.10) and R=0.37 (0.13) respectively. Taking age into account, those reporting high stress averaged nearly eight points lower on the energy scale and over five points lower on the mental component summary than those reporting low stress. Those reporting fear averaged respectively around 12 points and five points lower than those not so reporting. Because of their limited number of possible values and highly skewed distributions, the SF-36 dimension scores for emotional role and social functioning were both binarized. A score of 100 indicated no perceived problems and a score of less than 100 some problems. Again implicit fear of crime was the most consistent associate, with those feeling unsafe when outside after dark scoring lower on both dimensions. Those recording fear were, on average, 2.94 times as likely to report poor emotional role capability than those who were not afraid. Such individuals were also twice as likely to score low on social functioning. Stress had an independent effect on both emotional role and social functioning; while those who perceived a worsening of the crime rate and those living alone also scored low on social functioning. CONCLUSION

In many European cities tower blocks provide inadequate accommodation, with many residents perceiving them as insecure. In the city of Liverpool, substantial investment in security measures has greatly enhanced residents’ feelings of safety in such accommodation. However, turning tower blocks into ‘fortresses’ may increase

Table 4 Summary of multiple regression main effects coefficients and binary logistic odds ratios: SF-36 (mental) dimensions

MHI Energy MCS

Age (years)

Heating

Living alone

Tower block

0.20**

–5.24*

–4.80*

–2.96*

Crime change

–0.17* 0.15**

Emotional role (Bin) Social func. (Bin)

(1.57)*

** Significant at 0.01; * significant at 0.05; (*) significant at 0.10. Odds ratios in parentheses

14

(1.80)**

Stress

Fear dark

Adjusted R2

–10.08**

–11.01**

0.15

–7.73*

–11.73**

0.10

–5.53**

–4.81**

0.13

(2.59)**

(2.94)**

NA

(2.55)**

NA


residents’ alienation from the typically bleak landscape which surrounds them. About two-thirds of tower block residents in our study felt unsafe walking out alone after dark in neighbouring streets and this fear of crime is highly correlated with poorer mental health, after adjusting for age and gender. Residents who did feel safe out alone scored significantly higher on composite indicators of mental and social wellbeing. These findings suggest that measures to improve security should encompass both home and neighbourhood. In Liverpool the solution is to replace the tower blocks with completely redesigned estates of low-rise accommodation. However, this process of renewal can itself be stressful and may lead to poorer mental health. Liverpool residents experiencing these higher levels of stress report lower levels of mental health independent of any link with fear of crime. Urban renewal is a complex business. Achieving a more secure neighbourhood may lead to better mental health in the longer term, but short term the development process may undermine it. Evidence for this article was gained from a wider study funded by The North West Regional Executive of the UK National Health Service, The Housing Corporation and Liverpool Housing Action Trust.

REFERENCES 1 Acheson D. Independent Enquiry into Inequalities in Health. London: HMSO, 1998:53, 54. 2 McCabe A, Raine J. Framing the Debate: Crime and Public Health. Birmingham: UK Public Health Alliance, 1997. 3 East L. The quality of social relationships as a public health issue: exploring the relationship between health and community in a disadvantaged neighbourhood. Health Soc Care Community 1998;6(3):189-95. 4 Jacobs J. The death and life of great American cities. New York: Random House, 1961. 5 Newman O. Defensible space: people and design in the violent city. London: Architectural Press, 1973. 6 Coleman A. Utopia on trial: vision and reality in planned housing. London: Shipman, 1985. 7 Poyner B, Webb B. Crime free housing. London: Butterworth, 1991.

8 McCabe A, Raine J. Framing the Debate: Crime and Public Health. Birmingham: UK Public Health Alliance, 1997. 9 Hirschfield A, Bowers KJ. The effect of social cohesion on levels of reported crime in disadvantaged areas. Urban Studies 1997;34,8:1275-95. 10 Wilkinson RG, Kawachi I, Kennardy BP. Mortality, the social environment, crime and violence. Sociol health illness 1998;20,5:578-97. 11 Hough M. Anxiety about crime: findings from the 1994 British Crime Survey. Home Office Research Study. London: Home Office, 1995. 12 Wilkinson RG. Unhealthy societies: the afflictions of inequality. London: Routledge, 1996. 13 Elstad JI. The psycho-social perspective on social inequalities in health. Sociol health illness 1998;20,5:602. 14 East L. The quality of social relationships as a public health issue: exploring the relationship between health and community in a disadvantaged neighbourhood. Health Soc Care Community 1998;6(3):189-95. 15 Ware JE, Sherbourne CD. The MOS 36 item short-form health survey (SF-36): I. Conceptual framework and item selection. Med Care 1992;30:473-83. 16 Brazier JE, Harper R, Jones NMB, O’Cathain A, Thomas KJ, Usherwood T, Westlake L. Validating the SF-36 health survey questionnaire: new outcome measure for primary care. BMJ 1992;305:160-4. 17 Jenkinson C, Layte R, Wright L, Coulter A. The UK SF-36: an analysis and interpretation manual. Health Services Research Unit, University of Oxford, 1996. 18 Stewart AL, Hays RD, Ware JE. The MOS short-form general health survey: II. Reliability and validity in a patient population. Med Care 1988;26(7):724-35. 19 McHorney CA, Ware JE, Raczek AK. The MOS 36 item Short-Form Health Survey (SF-36): II. Psychometric and clinical tests of validity in measuring physical and mental health constructs. Med Care 1993;31:247-63. 20 McHorney CA, Ware JE, Lu JFR, et al. The MOS 36 item Short-Form Health Survey (SF-36): tests of data quality assumptions and reliability across diverse patient groups. Med Care 1994;31:247-63. 21 Green G, Ormandy D, Brazier JE, Gilbertson JM. Tolerant building: the impact of energy efficiency measures on living conditions and health status. In: Rudge J, Nichol F, editors. Cutting the cost of cold: affordable warmth for healthier homes. London: E&FN Spon, 2000. 22 Special table supplied to authors by Research and Statistics Directorate, extracted from Mirlees-Black C, Mayhew P, Percy A. The 1996 British crime survey, England and Wales. Home Office Statistical Bulletin 1996;19. 23 Market and Opinion Research International. Liverpool HAT Social Survey. Liverpool Housing Action Trust, 1993.

Received 2 December 1999, accepted 10 October 2000

15