An examination of social capital and social disorganisation in neighbourhoods in the British household panel study

Social Science & Medicine 56 (2003) 1425–1438

An examination of social capital and social disorganisation in neighbourhoods in the British household panel study Andrew McCulloch* Institute for Economic and Social Research, University of Essex, Wivenhoe Park, Colchester, CO4 3SQ, UK

Abstract Recent developments in social science research suggest that social environmental factors may be important for explaining community variations in health. We investigate the structural sources of two mechanisms that produce community variations in health. Using survey data collected from a representative cross-section of British households we examine variations in neighbourhood social capital and neighbourhood social disorganisation across a sample of British neighbourhoods. Adjusting for respondent’s attributes, we assess the effects of neighbourhood characteristics measured by the 1991 census in Britain. The results show that concentrated affluence, residential instability and ethnic heterogeneity predict social capital for women. Population density is the only neighbourhood characteristic to predict social capital for men. For both men and women concentrated disadvantage and population density are associated with social disorganisation. Residential instability is additionally associated with social disorganisation for women. For women it was found that neighbourhood characteristics interact with individual social class in accounting for variations in social capital, the effects of neighbourhood characteristics being larger for those in professional and managerial and skilled non-manual occupations. The results show that neighbourhood structural characteristics influence social organisation processes. This helps establish a link between the structural characteristics of neighbourhoods and individual health outcomes. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Social capital; Social disorganisation; Neighbourhood characteristics; British household panel survey

The study of neighbourhood environments and their influence on population health has recently increased among public health researchers. Neighbourhood research has focused primarily on the effects of concentrated poverty. In various studies, adolescent health behaviour, smoking, drinking, long-term illness, standardised mortality ratios, female heart disease and infant mortality rates were outcome measures affected independently by a context of economic disadvantage which was measured at the neighbourhood or small area level. Contextual effects remained after controlling for a variety of individual risk factors, like social class, household income, age, ethnicity, car ownership or pre-existing health conditions (Diez-Roux et al., 1997; Kleinschmidt, Hills, & Elliott, 1995; LeClere, Rogers, & Peters, 1998; Robert, 1998; Shouls, Congdon, & Curtis, *Fax: +44-1206-873151. E-mail address: [email protected] (A. McCulloch).

1996; Sooman & Macintyre, 1995; Waitzman & Smith, 1998). Much of this research sought only to confirm and document the existence of contextual effects. Much less understood are the precise processes through which neighbourhoods exert their effects. If numerous and seemingly disparate outcomes are linked together empirically across neighbourhoods and are predicted by similar structural characteristics, there may be a common underlying cause or mediating mechanism. Hence, rather than seeking specific or unique theories for each and every different outcome, another strategy is to develop a more general theory. Sampson, Raudenbush, and Earls (1997) and Sampson, Morenoff, and Earls (1999) present an attempt to specify an ecological model of how neighbourhood contexts are caused and the consequences they have for people and families, focusing particularly on the mechanisms that mediate between neighbourhood characteristics and

0277-9536/03/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 9 5 3 6 ( 0 2 ) 0 0 1 3 9 - 9

1426

A. McCulloch / Social Science & Medicine 56 (2003) 1425–1438

individual- or family-level outcomes. The theoretical framework of Sampson et al. (1997, 1999) centres on community social organisation, that is, the ability of a community structure to realise the common values of its residents and maintain effective social controls. Sampson et al. (1997, 1999) identify several characteristics of community social organisation which link neighbourhood-level structural characteristics with associated individual-level outcomes. The most important of these are social capital and social disorganisation. Neighbourhood-level structural characteristics have no hypothesised direct link to individual outcomes. Rather, neighbourhoods with disadvantaged social composition are more likely than advantaged neighbourhoods to lack environments conducive to the development of social organisation and social capital. The presence of social disorganisation and lack of social capital will help explain the statistical link between social composition and individual outcomes. In this paper we examine the structural sources of neighbourhood-level variations in both social disorganisation and social capital by combining census data on the social composition of neighbourhoods with individual reports of social capital and social disorganisation from the British Household Panel Study (BHPS). The next section goes beyond economic and demographic factors to discuss the social mechanisms relevant to individual health outcomes. We then examine the connection between structural characteristics of neighbourhoods and neighbourhood social mechanisms. Finally, we discuss the implications of the link between the social composition of neighbourhoods and neighbourhood social organisation.

Social organisation within neighbourhoods Properties of the social organisation of neighbourhoods can be just as, if not more, important in determining health as actual environmental exposures. As Wilkinson (1996, p. 215) notes, ‘‘It is the social feelings which matter, not exposure to a supposedly toxic material environment. The material environment is merely the indelible mark and constant reminder of one’s failure, of the atrophy of any sense of having a place in a community, and of one’s social exclusion and devaluation as a human being.’’ Neighbourhoods, therefore, can be sources of stress or a source of resources, depending on their social characteristics, and this can in turn influence health. It has been suggested that the development of social capital within a community can benefit health (Wilkinson, 1996; Lochner, Kawachi, & Kennedy, 1998). Social capital is not a unidimensional concept but the different authors on this topic share a common interest in how networks of relationships aid individual’s

actions. Coleman (1988) argues that social capital consists of ‘‘obligations and expectations, which depend on trustworthiness of the social environment, information-flow capability of the social structure and norms accompanied by sanctions.’’ Bourdieu and Wacquant (1992) discuss ‘‘the sum of resources, actual or virtual, that accrue to an individual or a group by virtue of possessing a durable network of more or less institutionalised relationships of mutual acquaintance and recognition.’’ Putnam (1995) defines social capital in a more general manner as ‘‘features of social life—networks, norms and trust— that enable participants to act together more effectively to pursue shared objectives.’’ Social capital is therefore a resource that is produced through relationships; whereas physical capital has a material form, and human capital consists of the skills and abilities gained by an individual. Two perspectives can be identified relative to the level at which the returns from social capital are conceived; whether they are accrued for the individual or for the individual. In the first perspective, the focus is on the use of social capital by individuals, how individuals access and use resources embedded in social networks. The focal points in this perspective are how individuals invest in social relations and how individuals capture the embedded resources in the relations. Lin (2001) provides an integrative review of such research through a focus on networks as a resource for status attainment. In the second perspective, the focus is on social capital at the group level, with discussions dwelling on how certain groups develop and maintain social capital as a collective asset and how such a collective asset enhances group member’s life chances. Bourdieu and Coleman have discussed this perspective extensively. The central interest of this perspective is to explore the elements and processes in the production and maintenance of the collective asset (Coleman 1988, 1990). Social capital may be developed in different ways. Social contact, particularly in public areas, and informal mutual assistance, or neighbouring behaviour allow residents to become better acquainted and discuss shared problems. Interactions such as the exchange of advice, material goods, and information lead to social networks that link adults in the community and that can provide benefits ranging from baby-sitting to job opportunities and emergency credit (Portes & Sensenbrenner, 1993). This sort of exchange may be facilitated by, but does not require, the presence of strong personal ties such as those found in tightly bounded friendship and kinship networks. Social capital is considered to be important for health for two reasons (Campbell, 1999; Wilkinson, 1996). The first takes into account the role of peer influences. In communities that are rich in social capital influences that stem from social ties may promote

A. McCulloch / Social Science & Medicine 56 (2003) 1425–1438

health-enhancing behaviours. Secondly, residents of communities with high levels of social capital are more likely to have high levels of perceived control over their everyday lives than individuals in communities with low levels of social capital. This is important for health, given that people who feel in control of their lives in general are more likely to take control of their health, through health-enhancing behaviours, or through the speedy and appropriate accessing of health services. The second dimension of neighbourhood-level social organisation studied here is social disorganisation. Social disorganisation refers to the inability of residents of an area to regulate everyday public behaviours and physical conditions within the bounds of their community. Indicators of a lack of order and social control in the community are both social and physical (Skogan, 1990). Visible signs of social disorganisation include trouble among neighbours and the presence of people hanging around on the streets and drinking alcohol. Physical disorder refers to the overall physical appearance of a neighbourhood. Places with high levels of physical disorder are noisy, dirty, and run down; many buildings are in disrepair or abandoned; and vandalism and graffiti are common. Physical disorder such as litter, graffiti, and vandalism also indicates that social control has broken down. The consequences of social disorganisation for individuals are great (Garofalo & Laub, 1978; Ross, Reynolds, & Geis, 2000). In contrast to those who live in safe, clean neighbourhoods, people who live in neighbourhoods characterised by disorganisation and a lack of social control may have high levels of fear, mistrust, perceived powerlessness, isolation, restricted outdoor activity, anger, anxiety, depression and poor health. In summary, social capital refers to the resource potential of personal and organisational networks, whereas social disorganisation relates to the degree to which residents are able to realise common goals and exercise social control. Social disorganisation and social capital may be conceived as overlapping, rather than competing explanations of the social mechanisms hypothesised to account for the effects of neighbourhood structural characteristics. Communities high in social capital are better able to realise common values and maintain social controls. They each hypothesise links between the socioeconomic composition of neighbourhoods and health via intervening social processes (Mayer & Jencks, 1989). The differences among them are largely differences of emphasis in the particular compositional features considered important and the roles of various intervening mechanisms. But they are each based on a causal framework that considers the socioeconomic composition of neighbourhoods the ultimate source of neighbourhood influences.

1427

Neighbourhood structural features Local neighbourhoods are important for an understanding of the distribution of economic resources and social stratification in Britain (Green, 1996; Noble & Smith, 1996). Physical capital and human capital (e.g., income, education, housing) are unevenly distributed across neighbourhoods. The importance of geographical characteristics is critical to an understanding of what communities provide to their residents. Residential stability allows for the development of social capital and social organisation via the formation of dense social networks. A high rate of residential turnover, especially substantial population loss, weakens interpersonal ties. Residential instability not only hinders the formation of new social networks, but the severing of existing social ties initiates a disruptive process that affects the entire system of social networks. Homeownership creates incentives to improve one’s neighbourhood because homeowners have a significant asset, the value of which is tied to the quality of the community. Homeownership also creates barriers to mobility. Lower levels of mobility also create incentives to invest in social capital. When someone expects to live longer in a community, the incentives to invest in that community become stronger. Thus, we expect that homeownership will promote collective efforts to maintain neighbourhood social networks and social organisation. Socioeconomic disadvantage is a second characteristic that distinguishes neighbourhoods. The geographical concentration of low socioeconomic status residents stems from macroeconomic changes related to the decline in manufacturing and good paying jobs for those with lesser skills. Economic stratification by residence thus increases the neighbourhood concentration of cumulative forms of disadvantage, intensifying the social isolation of low income and single-parent residents from resources that could support collective social control. Sampson et al. (1997) argue that such extreme resource deprivation acts as a force that hinders mutual trust and the shared willingness to intervene for the common good. Even when personal ties are strong in areas of concentrated disadvantage, daily experiences with distrust, fear of strangers, uncertainty and economic dependency are likely to reduce expectations for taking effective collective action. Thus, neighbourhoods characterised by concentrated disadvantage are expected to face multiple barriers to generating social order. The concentration and isolation of families and households characterised by multiple forms of disadvantage, for example, of lower-income, single-parent families, may also lead to less diverse social networks. We also consider the related but conceptually distinct factor of concentrated affluence (Massey, 1996). Poverty has not only become more concentrated in recent years, but so has the geographical sorting of residents by

1428

A. McCulloch / Social Science & Medicine 56 (2003) 1425–1438

resources such as education and occupation. Recent research suggests the importance of separating the upper end of the socioeconomic distribution from the lower end. Brooks-Gunn, Duncan, Klebanov, and Sealand (1993), for example, show that it is the positive influence of concentrated socioeconomic resources, rather than the presence of low-income neighbours, that enhances adolescent outcomes. We therefore explicitly assess the role of concentrated affluence in generating social capital and social organisation. Of course other characteristics affect the ability of local communities to develop social capital and social organisation. Other dimensions of neighbourhood efficacy (such as political ties) may be important, too.

Data and methods Neighbourhood-level research is dominated by studies of poverty rates and other sociodemographic characteristics based on census data or other government statistics that do not provide information on the social organisation of neighbourhoods. To examine how neighbourhood contexts are caused we use data from the BHPS, an annual panel survey of a representative cross-section of British households (Taylor, 1999). In the BHPS households were selected for inclusion based on an equal probability sample of the population of Great Britain (England, Wales, and Scotland south of the Caledonian canal), using a twostage stratified cluster design with postcode sectors as primary sampling units. Within selected households

information was gathered on all household members, and efforts were made to interview all those aged 16 and over. The first wave of BHPS interviews took place between September and December 1991. Our analysis is based on the seventh and eighth waves of the BHPS, and our sample consists of 2392 men and 2807 women who were resident at the same address at each of these waves. Measures of social organisational processes At wave eight respondents were asked eight questions about the neighbourhood in which they lived: ‘‘I feel like I belong to this neighbourhood’’, ‘‘the friendships and associations I have with other people in my neighbourhood mean a lot to me’’, ‘‘if I needed advice about something I could go to someone in my neighbourhood’’, ‘‘I borrow things and exchange favours with my neighbours’’, ‘‘I would be willing to work together on something to improve my neighbourhood’’, ‘‘I plan to remain a resident of this neighbourhood for a number of years’’, ‘‘I like to think of myself as similar to the people who live in this neighbourhood’’, ‘‘I regularly stop and talk with people in my neighbourhood’’. These eight questions indicate varied aspects of the social capital in the individual’s neighbourhood (Lochner et al., 1999). The term neighbourhood was not defined explicitly in the survey questionnaire. However, individual’s understanding of neighbourhood in response to similar questions on the trustworthiness of neighbours or the extent to which they know people is generally equivalent to the level of street, road or block (Campbell, 1999).

Fig. 1. Histograms of social capital and social disorder for men and women.

A. McCulloch / Social Science & Medicine 56 (2003) 1425–1438

Each response was coded on a five-point scale from strongly disagree (score=1) to strongly agree (score=5). The responses were summed yielding an index with a mean of 28.1 for men and 29.2 for women (Fig. 1). Higher values indicate higher levels of social capital. The reliability measures obtained from the individual reports using Cronbach’s alpha were 0.83 for men and 0.82 for women. At wave seven respondents were asked eight questions from the scale developed by Buckner (1988) concerning the severity of various community problems. The problems are: ‘‘graffiti on walls or buildings’’, ‘‘teenagers hanging around in streets’’, ‘‘drunks or tramps on the streets’’, ‘‘vandalism and deliberate damage to property’’, ‘‘insults or attacks to do with someone’s race or colour’’, ‘‘homes broken into’’, ‘‘cars broken into or stolen’’, and ‘‘people attacked on the streets’’. These measures focus on individual perceptions of community problems, and on perceptions about the physical environment (Ross & Mirowsky, 1999). To residents these conditions of disorganisation or incivility, both physical and social, symbolise not only a superficial neglect of the community but also an underlying breakdown in both local norms of behaviour and informal and formal social controls. Each response was coded using a four-point scale, ranging from very common (score=1) to not at all common (score=4). The responses were summed yielding an index with a mean of 24.4 for men and 24.1 for women (Fig. 1). Higher values indicate higher levels of social organisation. The reliability measure obtained from the individual reports using Cronbach’s alpha was 0.86 for both men and women. To assess the construct validity of these two measures, we examine their relationship with each other and with five measures that tap related but conceptually distinct dimensions of neighbourhood social organisation. Local organisations are a five-item index of involvement in reported local organisations and programmes (parent’s association, tenants or residents group, religious group, voluntary service group, other community group). Local services is a five-item inventory of the standard (1=poor to 4=excellent) of local services (schools, medical, transport, shopping, leisure). Voluntary activities taps involvement (1=never to 5=at least once a week) by residents in evening classes, local groups and voluntary work. A single item measured trust of other people: ‘‘Generally speaking, would you say that most people can be trusted, or that you can’t be too careful in dealing with people?’’. A single item measured concern about crime: ‘‘Do you ever worry about the possibility that you, or anyone else who lives with you, might be the victim of crime?’’. We expect that social capital and social organisation will correlate positively with these five measures of neighbourhood social life.

1429

Structural antecedents We examine five indexes of neighbourhood structural differentiation that build on the theoretical framework above and prior analyses of census data. In this analysis the definition of neighbourhood is dictated by the geographical units for which data are available. Wards are the constituency for local government elections. Being based on political boundaries, it is arbitrary, but nevertheless represents a useful framework for analysis. Their average population is around 5500, but this varies considerably. The men in our sample come from a total of 1053 wards and the women from a total of 1159 wards. The maximum number of men in one ward is 12 and the maximum number of women is 16. Concentrated disadvantage was assessed by the Townsend index of four components; each measured for the ward of residence at the 1991 census (Townsend, Phillimore, & Beattie, 1989). These components were: the proportion of the labour force unemployed, the proportion of households with no car access, the proportion of households with one person per room and over and the proportion of households not owning their own home. Consistent with Sampson et al. (1997), the second scale captures neighbourhood residential instability, defined as the percentage of residents who resided in a different house 1 year earlier, and the percentage of nonowner-occupied homes. Both scales are based on the summation of equally weighted Z-scores. To these basic dimensions of social structure we introduce three additional measures. Concentrated affluence is defined by the percentage of economically active household heads employed in professional or managerial and technical occupations. Ethnic concentration is defined as the percentage of household heads born in New Commonwealth. Finally, population density is defined as persons per hectare. Multilevel models The clustered nature of the BHPS sample design is addressed by using multilevel linear models that account for the non-independence of observations within neighbourhoods. The multilevel model has been described elsewhere and extensively applied in social science and medicine (Bryk & Raudenbush, 1992; Snijders & Bosker, 1999; Duncan, Jones, & Moon, 1996). The multilevel model simultaneously estimates within-neighbourhood and between-neighbourhood equations. The withinneighbourhood model regresses the two measures of social mechanisms (social capital and social organisation) on a core set of individual and household-level characteristics that have been shown in prior research to influence both perceptions and behaviour relevant to neighbourhoods. Specifically, we examine eight person-level (or household-level) attributes: education;

1430

A. McCulloch / Social Science & Medicine 56 (2003) 1425–1438

deprivation; social support; household type (composed of separate indicators for couple with no dependent children, couple with dependent children, lone-parent, single and other); residence in social housing (rented from local authority or housing association); residential tenure in years; age and social class based on the 1990 Registrar General’s Classification of Occupations (OPCS, 1991). Women were classified according to their own occupation (current or most recent) and not that of their partner, where married or cohabiting. The original six social classes were collapsed into four, largely to avoid small numbers in extreme groups: classes I and II (professional and managerial/technical), IIInm (skilled non-manual), IIIm (skilled manual) and IV and V (semi and unskilled manual). The between-neighbourhood model regresses each neighbourhoods mean social capital or social organisation score against the set of neighbourhood characteristics. Males and females were modelled separately to allow for any gender differentiation of relationships. Using social capital as an example, the model for individual i in neighbourhood j is Social capitalij ¼ Xij b þ U0j þ eij ;

ð1Þ

where Xij is the set of covariates and b is a vector of parameters associated with Xij : The error term, eij ; is the unique contribution of each individual, which is assumed to be independently and normally distributed with constant variance s2 : U0j is the neighbourhoodlevel error term, assumed to be normally distributed with a variance of t: Models were also estimated to assess whether the effect of neighbourhood characteristics was different for different types of people. This was done through the inclusion of cross-level interaction variables, namely interactions between individual social class and neighbourhood characteristics. The underlying argument here is that, irrespective of any overall effect of a neighbourhood characteristic, there may be some groups for whom the effect is larger or smaller.

Results Table 1 examines construct and discriminant validity. The results are similar for men and women. The scales measuring social capital and social organisation are significantly and strongly related to each other in a positive direction. Conforming to extant neighbourhood theory, social capital and social organisation are also significantly and positively related to more general components of neighbourhood social life: involvement in local organisations, local services, participation in voluntary activities and trust. Also as expected social organisation is significantly related to concern about crime. Social capital is significantly lower in disadvantaged, residentially unstable, ethnically heterogeneous

and high population density neighbourhoods, but is higher in areas of concentrated affluence. Social organisation is significantly higher in disadvantaged, residentially unstable, ethnically heterogeneous and high population density neighbourhoods, but is lower in areas of concentrated affluence. Table 2 presents bivariate correlations among the neighbourhood structural predictors measured by 1991 census data. As expected, concentrated disadvantage and concentrated affluence are inversely related, but do not present severe multicollinearity problems. Concentrated disadvantage is positively related to ethnic concentration, residential instability and population density. Areas of high ethnic concentration tend to be low in residential stability, and affluent areas have a low population density. Overall, the correlations in Tables 1 and 2 support construct and discriminant validity for census measures as well as for the survey-based measures of social capital and social organisation. We now turn to the multilevel results. The main question is, once individual correlates of social capital and social organisation are controlled, what is the relative predictive power of exogenous structural characteristics? Table 3a and b shows the partition of the variance within and between neighbourhoods. For social capital the results reveal an intraclass correlation of 0.090 for men and 0.113 for women meaning that 9.0% and 11.3%, respectively, of the scales variance is between neighbourhoods, with the remainder attributable to random error and individual-level variation. The estimated intraclass correlations for social order are appreciably higher at 0.212 for men and 0.255 for women. Table 4a shows that controlling for socioeconomic status and other individual and household predictors, men and women who are long-term residents tend to report high levels of social capital. Contrary to expectation, however, respondents who live in owner occupation and private rented tenures do not report significantly higher levels of social capital in comparison to those resident in social housing. Men in skilled manual occupations and women in semi and unskilled manual occupations report higher levels of social capital than those in professional and managerial occupations. The results highlight age and household-type differences in men and women’s experiences of community life: respondents who are older tend to report higher levels of social capital as do respondents living in households with dependent children. Adjusting for these variables Table 4a shows that concentrated affluence, residential instability and ethnic concentration predict neighbourhood social capital among women. Population density is the only statistically significant predictor of neighbourhood social capital for men. The results for social organisation (Table 4b) show that controlling for socioeconomic status and other

A. McCulloch / Social Science & Medicine 56 (2003) 1425–1438

1431

Table 1 Mean values and bivariate correlation coefficients between neighbourhood-level measures of social organisation and structural differentiation Measure

Mean value

Correlation coefficients Social capital

Social order

Men Neighbourhood social organisation Social capital Social organisation Concern about crime Voluntary activities Local organisations Local services Trust

28.17 24.44 0.62 4.55 0.26 12.31 0.40

— 0.14** 0.02 0.09** 0.13** 0.2** 0.09**

0.14** — 0.19** 0.02 0.03 0.11** 0.12**

Neighbourhood structural differentiation Concentrated disadvantage Concentrated affluence Residential instability Ethnic concentration Population density

0.20 0.38 0.10 0.04 24.00

0.14** 0.05* 0.13** 0.12** 0.17**

0.33** 0.21** 0.22** 0.15** 0.23**

Women Neighbourhood social organisation Social capital Social organisation Concern about crime Voluntary activities Local organisations Local services Trust

29.19 24.16 0.67 5.43 0.38 12.50 0.38

— 0.14** 0.02 0.10** 0.13** 0.19** 0.12**

0.14** — 0.19** 0.02 0.04* 0.05* 0.10**

Neighbourhood structural differentiation Concentrated disadvantage Concentrated affluence Residential instability Ethnic concentration Population density

0.30 0.37 0.06 0.04 24.15

0.16** 0.10** 0.15** 0.11** 0.11**

0.35** 0.22** 0.22** 0.19** 0.23**

**po0:01; *po0:05 (two-tailed tests). Table 2 Intercorrelations between measures of neighbourhood structural differentiation Structural differentiation

(1)

(2)

(3)

(4)

(5)

Men Concentrated disadvantage Concentrated affluence Residential instability Ethnic concentration Population density

1 0.6 0.74 0.46 0.52

1 0.29 0.1 0.22

1 0.23 0.36

1 0.48

1

Women Concentrated disadvantage Concentrated affluence Residential instability Ethnic concentration Population density

1 0.61 0.75 0.43 0.5

1 0.29 0.07 0.2

1 0.23 0.37

0.48

1

individual and household predictors, men and women who are long-term residents, as well as those living in social housing, tend to report lower levels of social organisation. The results again highlight age-related differences: respondents who are older tend to report higher levels of social organisation. Men in semi and unskilled manual occupations report lower levels of social organisation than those in professional and managerial occupations. The results reveal a clear pattern of neighbourhood economic stratification. Controlling for concentrated affluence and ethnic concentration, the level of social organisation is closely associated with concentrated disadvantage and population density for both men and women. Residential instability is additionally associated with social disorganisation for women. Because person-level measures of stability (home ownership and residential tenure) and

1432

A. McCulloch / Social Science & Medicine 56 (2003) 1425–1438

Table 3 Men

Women

b (a) Decomposition of variance of scales measuring social capital Intercept 28.04 Between-neighbourhood variance (t) 2.66 Within-neighbourhood variance (s2 ) 26.76 Intraclass correlation 0.090 (b) Decomposition of variance of scales measuring social organisation Intercept 24.56 Between-neighbourhood variance (t) 4.30 Within-neighbourhood variance (s2 ) 16.01 Intraclass correlation 0.212

deprivation all significantly influence social organisation, the estimated neighbourhood effects of concentrated disadvantage are noteworthy. Residents of disadvantaged neighbourhoods report significantly lower levels of social organisation than their counterparts in more advantaged neighbourhoods. While social capital (personal ties) may be little affected by the concentration of disadvantage, social organisation is attenuated. For both outcomes the random part of the model results show that a large part of the variance remains unexplained, particularly at the level of the individual. For social capital more variance is explained within neighbourhoods although it remains low overall. The results indicate that considerable variation remains in responses from respondents within the same neighbourhood. Another source of evidence from the survey supports the hypothesis that social capital and social organisation are important factors associated with poor neighbourhood quality. Residents were asked whether their neighbourhood was a good or bad place to live and what made it a good or bad place to live. For each respondent up to six responses were coded using 96 codes. Table 5 lists aspects of neighbourhood mentioned by respondents giving their neighbourhood as either a good or bad/mixed place to live (all six responses combined). When residents are asked about neighbourhoods they refer to the behaviour or qualities of the people. When responses are combined the most frequently mentioned aspects of neighbourhood among those who rated their neighbourhood quality as bad or mixed were ‘‘problems with young people’’, ‘‘people unfriendly’’, ‘‘crime rate high’’, ‘‘problems with drugs’’ and ‘‘problems with vandalism’’. Among respondents who rated their neighbourhood quality as good there was substantial conformity in responses. The most

t

b

t

227.66

29.06 3.15 24.83 0.113

251.01

220.37

24.39 5.53 16.13 0.255

216.79

frequently mentioned aspects were ‘‘quiet area’’ and ‘‘people friendly’’. There were minor differences between responses from men and women. Models showing the effects of cross-level interactions are shown for social capital in Table 6a and for social organisation in Table 6b. For men neighbourhood characteristics tend to have a general rather than a socially specific effect. For both social capital and social organisation the effect of neighbourhood characteristics are uniform across different types of people. For women the results are more complicated. The effect of neighbourhood characteristics on social organisation is consistent across individual social class. However, the effects of area characteristics on social capital vary according to individual social class. The relationship of neighbourhood characteristics to social capital being stronger for women in professional and managerial and skilled non-manual occupations than for women in manual occupations.

Discussion The results reveal that concentrated disadvantage has no significant effect on the level of neighbourhood social capital. We found that among women concentrated affluence, residential stability and ethnic heterogeneity are linked to higher levels of neighbourhood social capital whereas for men population density was the only neighbourhood characteristic linked to such levels. High residential turnover restricts the ability of women to develop informal ties and friendship networks with the other residents of their neighbourhood. Similarly, feelings of belonging, social ties and social networks are stronger the greater the ethnic homogeneity of an area. Concentrated affluence has been found to be positively related to children’s cognitive ability

A. McCulloch / Social Science & Medicine 56 (2003) 1425–1438

1433

Table 4 Men b

Women t

b

(a) Coefficients and t-statistics from the regression of social capital on individual and neighbourhood predictors Intercept 17.94 18.67 18.32 Person-level 0.488 3.46 0.101 Deprivationa 0.449 9.54 0.419 Social supportb Age 0.078 8.78 0.056 Residential tenure 0.035 3.14 0.053 Social housing 0.603 1.79 0.006 Couple no dependent childrenc 0.05 0.14 0.398 Couple-dependent children 0.756 2.12 1.532 Lone parent 0.6 0.94 0.936 Other 1.27 1.9 0.982 Single — — — Educationd 0.125 1.61 0.052 Semi and unskilled manual occupation 0.445 1.34 1.125 Skilled manual occupation 0.61 2.25 0.384 Skilled non-manual occupation 0.023 0.07 0.403 Professional and managerial occupation — — — Neighbourhood level Concentrated disadvantage 0.036 0.49 0.04 Concentrated affluence 0.471 0.45 2.294 Residential instability 0.109 0.93 0.228 Ethnic concentration 1.67 0.91 4.96 Population density 0.025 4.37 0.001 Between-neighbourhood variance (t) 1.18 1.82 Within-neighbourhood variance (s2) 23.79 22.86 Intraclass correlation 0.047 0.074 Percentage of variance explained Between neighbourhoods 55.6 42.2 Within neighbourhoods 11.1 7.9 (b) Coefficients and t-statistics from the regression of social organisation on individual and neighbourhood predictors Intercept 24.33 30.39 24.21 Person-level Deprivation 0.328 2.82 0.411 Social support 0.061 1.57 0.05 Age 0.022 2.98 0.031 Residential tenure 0.02 2.12 0.034 Social housing 1.36 4.88 0.886 Couple no dependent children 0.319 1.1 0.751 Couple-dependent children 0.361 1.23 0.636 Lone parent 1.23 2.35 0.539 Other 0.905 1.64 0.971 Single Education 0.004 0.07 0.051 Semi and unskilled manual occupation 0.643 2.36 0.301 Skilled manual occupation 0.339 1.52 0.163 Skilled non-manual occupation 0.147 0.51 0.032 Professional and managerial occupation Neighbourhood level Concentrated disadvantage 0.389 6.02 0.429 Concentrated affluence 0.117 0.13 0.143 Residential instability 0.155 1.51 0.232 Ethnic concentration 2.327 1.45 1.63 Population density 0.021 4.33 0.017 Between-neighbourhood variance (t) 1.99 2.45

t 18.58 0.82 8.41 6.43 5.3 0.02 1.29 4.58 2.19 1.39 — 0.7 3.81 1.08 1.63 — 0.56 2.23 2 2.72 0.21

28.78 3.96 1.2 4.21 4 3.67 2.88 2.26 1.49 1.63 0.82 1.21 0.54 0.16

6.76 0.15 2.27 1 3.44

1434

A. McCulloch / Social Science & Medicine 56 (2003) 1425–1438

Table 4 (continued) Men b 2

Within-neighbourhood variance (s ) Intraclass correlation Percentage of variance explained Between neighbourhoods Within neighbourhoods

Women t

15.22 0.116 53.7 4.9

t

b 15.59 0.136 55.7 3.3

a Respondents were asked about the enforced absence of several items. The absence and affordability elements were incorporated in one question, as follows: ‘‘Would you like to be able to but must do without because you cannot afford it?’’. The following six items were administered in this fashion: ‘‘keep your home adequately warm’’; ‘‘pay for a week’s annual holiday away from home’’; ‘‘replace worn out furniture’’; ‘‘buy new, rather than second hand, clothes’’; ‘‘eat meat, chicken or fish at least every second day’’; ‘‘have friends or family for a drink or meal at least once a month’’. The respondent’s deprivation score is the number of items which they want but are unable to afford. b Individual’s social support was assessed using responses to five items (e.g. ‘‘Is there someone who will listen?’’). Each response takes the value 1 if the individual had no one to offer support, 2 if the individual had one person and the value 3 if they had more than one person. Responses to the five questions were summed and the total score divided into three groups. c Children aged under 16 years or aged 16–18 years and in full-time higher education. d Highest educational qualification: 0=none to 5=degree.

Table 5 Men Aspect of Neighbourhood

Women %

Aspect of Neighbourhood

%

(a) Most frequently mentioned aspects of neighbourhood from respondents rating their neighbourhood as bad or mixed Problems with young people 22.48% Problems with young people 26.29% People unfriendly 14.01% Problems with drugs 15.72% Noise problems 13.68% Problems with vandalism 14.91% Problems with drugs 12.38% People unfriendly 14.63% Problems with vandalism 12.05% Noise problems 14.36% Traffic problems 12.05% Pollution problems 11.65% Good shopping facilities 10.75% Area becoming worse 11.38% Crime rate high 10.42% Crime rate high 11.11% Area becoming worse 9.77% Other negative aspect 10.30% Pollution problems 9.45% Traffic problems 10.30% Number of respondents 307 369 (b) Most frequently mentioned aspects of neighbourhood from respondents rating their neighbourhood as good Quiet area 44.40% Quiet area People friendly 31.52% People friendly No crime 25.95% Neighbours friendly Neighbours friendly 22.11% No crime Rural surroundings 19.85% Rural surroundings Good shopping facilities 16.53% Good shopping facilities Like the area 14.99% Town centre accessible Good local facilities 14.75% Like the area Town centre accessible 13.74% Has open spaces Area feels safe 12.40% Good local facilities Number of respondents 2082

(Brooks-Gunn et al., 1993). In terms of scarcity of resources, affluent neighbourhoods may have more institutional resources. The presence of affluent neigh-

43.34% 36.80% 23.73% 23.07% 20.15% 20.11% 14.06% 13.98% 13.78% 13.24% 2433

bours may also increase residents’ feelings that they could gain through involvement and commitment to the neighbourhood.

A. McCulloch / Social Science & Medicine 56 (2003) 1425–1438

1435

Table 6 Interactions

Men b

Women t

b

t

(a) Coefficients and t-statistics from the regression of social capital on the interaction of individual social class and each neighbourhood predictor Concentrated disadvantage  Semi and unskilled manual occupation 0.052 0.32 0.207 1.67 Skilled manual occupation 0.099 0.85 0.02 0.11 Skilled non-manual occupation 0.088 0.42 0.002 0.01 Professional and managerial occupation 0.027 0.22 0.029 0.24 Concentrated affluence  Semi and unskilled manual occupation 1.468 0.58 2.31 1.17 Skilled manual occupation 3.61 1.84 1.424 0.48 Skilled non-manual occupation 0.688 0.22 3.161 1.89 Professional and managerial occupation 2.228 1.47 4.778 2.95 Residential instability  Semi and unskilled manual occupation 0.101 0.36 0.035 0.16 Skilled manual occupation 0.201 1.06 0.079 0.26 Skilled non-manual occupation 0.487 1.39 0.496 2.58 Professional and managerial occupation 0.078 0.41 0.189 0.99 Ethnic concentration  Semi and unskilled manual occupation 0.069 0.02 3.57 1.13 Skilled manual occupation 1.9 0.61 2.946 0.51 Skilled non-manual occupation 2.642 0.53 6.79 2.32 Professional and managerial occupation 3.46 1.12 5.72 1.74 Population density  Semi and unskilled manual occupation 0.028 2.32 0.002 0.15 Skilled manual occupation 0.013 1.28 0.022 1.4 Skilled non-manual occupation 0.026 1.74 0.006 0.73 Professional and managerial occupation 0.034 3.85 0.007 0.7 (b) Coefficients and t-statistics from the regression of social organisation on the interaction of individual social class and each neighbourhood predictor Concentrated disadvantage  Semi and unskilled manual occupation 0.24 1.75 0.512 4.82 Skilled manual occupation 0.42 4.27 0.307 1.91 Skilled non-manual occupation 0.423 2.42 0.478 4.69 Professional and managerial occupation 0.412 3.99 0.273 2.61 Concentrated affluence  Semi and unskilled manual occupation 0.611 0.29 0.841 0.5 Skilled manual occupation 0.683 0.41 3.575 1.41 Skilled non-manual occupation 0.157 0.06 1.29 0.88 Professional and managerial occupation 0.076 0.06 0.887 0.63 Residential instability  Semi and unskilled manual occupation 0.262 1.13 0.138 0.74 Skilled manual occupation 0.403 2.53 0.449 1.77 Skilled non-manual occupation 0.025 0.08 0.273 1.64 Professional and managerial occupation 0.165 1.03 0.149 0.9 Ethnic concentration  Semi and unskilled manual occupation 3.571 1.12 4.62 1.7 Skilled manual occupation 0.308 0.12 0.638 0.13 Skilled non-manual occupation 3.389 0.81 0.854 0.34 Professional and managerial occupation 2.725 1.06 3.27 1.17 Population density  Semi and unskilled manual occupation 0.03 3 0.001 0.01 Skilled manual occupation 0.021 2.54 0.02 1.48 Skilled non-manual occupation 0.016 1.24 0.025 3.32 Professional and managerial occupation 0.017 2.22 0.021 2.57 Each model also contains the individual and family level predictors in the models in Table 4. The coefficients of the individual and family level predictors are similar in magnitude and statistical significance to those in Table 4 and are not presented.

1436

A. McCulloch / Social Science & Medicine 56 (2003) 1425–1438

The results for social organisation reveal a clear pattern of neighbourhood economic stratification. Concentrated disadvantage emerges as an important predictor: regardless of person-level covariates, disadvantaged neighbourhoods exhibit considerably lower levels of social organisation than do advantaged neighbourhoods. Low population density also appears instrumental in allowing many neighbourhoods to achieve an efficacious environment. Apparently, the concentration of multiple forms of disadvantage erodes neighbourhood order and social control. The distressing aspect of neighbourhood disadvantage is not a lack of informal social ties with neighbours. Different response to neighbourhood characteristics by gender may be due to different roles, stressors, and experiences. It is suggested that women’s social capital is more heavily influenced by neighbourhood characteristics, because they rely on local social networks in their multiple roles as mother and worker. Women are generally more actively concerned with local facilities for children and child care. Women of all ages play an important role in maintaining the social structure of neighbourhoods and in being actively involved in both formal and informal organisations. Social networks are centred around children in relation to schooling, child care and general concerns about safety and security. Furthermore, while women’s participation in community organisations can be linked to their traditional roles and responsibilities in bringing up children, the social position of men in a deprived community has become more problematic. The loss of position as the principal earner may result in a general lack of interest, and be reflected in a lack of involvement in the daily running of voluntary and community organisations. Additionally, older adults appear to have stronger feelings of commitment to their neighbourhoods and so may be an important source of community action and involvement. Cross-level interactions in which the effect of neighbourhood characteristics on social capital varies according to individual social class, were found for women but not for men in this analysis. For women in professional occupations increases in neighbourhood disadvantage may lead to changes in lifestyle, for example, strategies involving spatial restrictions to avoid potential harassment or other factors such as social isolation possibly resulting from status inconsistency. Our study suggests that the association of neighbourhood characteristics with social organisation is uniform across both social class and gender. That is, areas of concentrated neighbourhood disadvantage and high population density are characterised by lower levels of social organisation for all types of people. It may be the case, therefore, that neighbourhoods may matter through a contextual effect. Thus, living in deprived areas has an impact on individual’s experience of crime, vandalism, grafitti and

other signs of social disorganisation not simply because of the presence of more deprived people but also due to living in unpleasant, undesirable and unsafe environments. Our research shows the importance of distinguishing social organisation from social ties. Some researchers see these as almost interchangeable because alliances among neighbours provide the mechanism by which informal social control is enforced. However, close family ties, mutual aid and voluntarism are often strong features of poor areas. It is these qualities which may enable people to cope with poverty, unemployment and wider processes of social exclusion. Portes and Landolt (1996) point out that ‘‘there is considerable social capital in ghetto areas, but the assets attainable through it seldom allow participants to rise above their poverty.’’ The general pattern of relationships found in this study is similar to that found by Sampson et al. (1997, 1999) in their community-based study of juvenile delinquency and crime in Chicago neighbourhoods. Sampson et al. (1997, 1999) surveyed 8782 residents of 343 Chicago neighbourhoods in 1995 to ask about their perceptions of social cohesion and trust in the neighbourhood. In both our studies concentrated poverty is negatively related to indices of social organisation. Sampson et al. (1997) found that an index of mutual trust and social control was significantly inversely associated with reports of neighbourhood violence, violent victimisation, as well as homicide rates. Similarly, Sampson and Groves (1989) using the 1982 British Crime Survey found that areas with high rates of deviant behaviour were also characterised by low economic status, ethnic heterogeneity, and high population turnover. They argued that these conditions led to social disorganisation, which in turn led to deviant behaviour among individuals. These discussions can be likened to current debates on poverty and social exclusion, which focus on people’s ability to participate in activities that others take for granted (Townsend, 1979). The question is to what extent does social disorganisation contribute to exclusion from participation in social activities? Few studies have examined the impact of social disorganisation on people’s ability to participate in society. As with all cross-sectional research, this study is limited in its efforts to establish causal direction. Undoubtedly, there are reciprocal feedback effects of social disorganisation on social capital that fail to be captured by cross-sectional analyses. In other words, high rates of disorganisation may themselves result in a decline in community social capital. Skogan (1990) identified several feedback processes that contribute to declining social capital, including: fear of crime leading to physical and psychological withdrawal from community life; deteriorating conditions leading to the exit of businesses, with accompanying loss of jobs (and norms of labour market attachment); and further change in the

A. McCulloch / Social Science & Medicine 56 (2003) 1425–1438

composition of localities. If people avoid their neighbourhoods due to fear of crime, fewer opportunities exist for local networks and associations to develop. The resulting disorganisation of community structure in turn fuels further crime, producing a cycle of declining social capital, followed by rising crime, followed by a further decline in social capital. Future research should examine the meaning and sources of variation within neighbourhoods in survey respondent’s perceptions and behaviours. The results show considerable heterogeneity in responses from respondents within the same neighbourhood. The analyses controlled for person-level covariates such as age, sex and socioeconomic status, but very little of the variation was explained. What is the source of this unexplained variation? Between neighbourhood differences in reports of social capital probably arise in part from within-neighbourhood differences in unmeasured factors such as local friendship or kinship ties and organisational affiliations (Tienda, 1991). Neighbourhoods are much less homogeneous than commonly suggested in the literature, suggesting the need to study smaller ecological units. Using the same data as this study, McCulloch (2001) examines the consequences of neighbourhood social organisation and social capital for individuals self-rated health problems. The results indicate that the neighbourhood mechanisms identified are important for health, and this helps establish a link between the structural characteristics of neighbourhoods and individual health outcomes. Left unanswered, though, is another set of why and how questions: how do the structural characteristics of neighbourhoods influence the development and maintenance of social capital and social order? Exploring such qualitative aspects of the neighbourhood measures employed is necessary to advance our understanding of why neighbourhoods matter. Such an investigation may require integrating both neighbourhood-level quantitative measures, the subject of quantitative sociologists and demographers, and qualitative measures more commonly explored by ethnographers and psychologists. Finally, the notion of local residents working together to produce social organisation and develop social capital is not the whole picture. As shown, what takes place within neighbourhoods is influenced by socioeconomic factors linked to the wider economy. In addition to encouraging communities to combat crime through strategies of informal social control and develop social networks through local voluntary organisations, policies to address the social and ecological changes that have affected many communities need to be considered. Housing-based neighbourhood stabilisation (through renovation of existing low income housing) and dispersing the concentration of new public housing are two examples of a bottom-up approach to supporting social

1437

cohesion and safer neighbourhoods (Sampson, 1995). At the same time, Government should not ignore top-down approaches, such as policies to reduce income inequality, as a possible way of promoting social capital. Knowing that what happens within neighbourhoods is important does not imply that inequalities between neighbourhoods can be ignored.

Acknowledgements Funded under the ESRC Cities and Competitiveness Research Programme Grant Number L130251010. Thanks are due to the referees for helpful comments.

References Bourdieu, P., & Wacquant, L. (1992). Invitation to reflexive sociology. Chicago: University of Chicago Press. Brooks-Gunn, J., Duncan, G. J., Klebanov, P. K., & Sealand, N. (1993). Do neighborhoods affect child and adolescent development? American Journal of Sociology, 99, 353–395. Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models, applications and data analysis methods. Newbury Park, CA: Sage. Buckner, J. C. (1988). The development of an instrument to measure neighbourhood cohesion. American Journal of Community Psychology, 16, 771–791. Campbell, C. (1999). Social capital and health. London: Health Education Authority. Coleman, J. (1988). Social capital in the creation of human capital. American Journal of Sociology, 94(Supplement), S95–S120. Coleman, J. (1990). Foundations of social theory. Cambridge, MA: Harvard University Press. Diez-Roux, A. V., Nieto, J., Muntaner, C., Tyroler, H. A., Comstock, G. W., Shahar, E., Cooper, L. S., Watson, R. L., & Szklo, M. (1997). Neighbourhood environments and coronary heart disease: A multilevel analysis. American Journal of Epidemiology, 146, 48–63. Duncan, C., Jones, K., & Moon, G. (1996). Health-related behaviour in context: A multilevel modelling approach. Social Science & Medicine, 42, 817–830. Garofalo, J., & Laub, J. (1978). The fear of crime: Broadening our perspective. Victimology, 3, 242–253. Green, A. (1996). Aspects of the changing geography of poverty and wealth. In J. Hills (Ed.), New inequalities: The changing distribution of income and wealth in the United Kingdom. Cambridge: Cambridge University Press. Kleinschmidt, I., Hills, M., & Elliott, P. (1995). Smoking behaviour can be predicted by neighbourhood deprivation measures. Journal of Epidemiology and Community Health, 49(Suppl.2), S72–S77. LeClere, F., Rogers, R., & Peters, K. (1998). Neighbourhood social context and racial differences in women’s heart disease mortality. Journal of Health and Social Behavior, 39, 91–107.

1438

A. McCulloch / Social Science & Medicine 56 (2003) 1425–1438

Lin, N. (2001). Social capital: A theory of social structure and action. Cambridge: Cambridge University Press. Lochner, K., Kawachi, I., & Kennedy, B. P. (1999). Social capital: A guide to its measurement. Health and Place, 5, 259–270. Massey, D. S. (1996). The age of extremes: Concentrated affluence and poverty in the 21st century. Demography, 33, 395–412. Mayer, S. E., & Jencks, C. (1989). Growing up in poor neighborhoods: How much does it matter? Science, 243, 1441–1445. McCulloch, A. (2001). Social environments and health: Cross sectional national survey. British Medical Journal, 323, 208–209. Noble, M., & Smith, G. (1996). Changing patterns of income and wealth in two contrasting areas. In J. Hills (Ed.), New inequalities: The changing distribution of income and wealth in the United Kingdom. Cambridge: Cambridge University Press. OPCS (1991). Standard occupational classification, Vol. 3. London: HMSO. Portes, A., & Landolt, P. (1996). The downside of social capital. The American Prospect, 26, 18–21. Portes, A., & Sensenbrenner, J. (1993). Embeddedness and immigration: Notes on the social determinants of economic action. American Journal of Sociology, 98, 1320–1350. Putnam, R. (1995). Tuning in, tuning out: the strange disappearance of social capital in america. PS: Political Science and Politics, 28, 664–683. Robert, S. (1998). Community-level socioeconomic status effects on adult health. Journal of Health and Social Behaviour, 39, 19–37. Ross, C. E., & Mirowsky, J. (1999). Disorder and decay: The concept and measurement of perceived neighbourhood disorder. Urban Affairs Review, 34, 412–432. Ross, C. E., Reynolds, J. R., & Geis, K. J. (2000). The contingent meaning of neighbourhood stability for residents’ psychological well-being. American Sociological Review, 65, 581–597.

Sampson, R. J. (1995). The community. In J. Q. Wilson, & J. Petersilia (Eds.), Crime (pp. 193–216). San Francisco: Institute for Contemporary Studies. Sampson, R. J., & Groves, W. B. (1989). Community structure and crime: Testing social disorganization theory. American Journal of Sociology, 94, 774–802. Sampson, R. J., Morenoff, J. D., & Earls, F. (1999). Beyond social capital: Spatial dynamics of collective efficacy for children. American Sociological Review, 64, 633–660. Sampson, R. J., Raudenbush, S. W., & Earls, F. (1997). Neighborhoods and violent crime: A multilevel study of collective efficacy. Science, 277, 918–924. Shouls, S., Congdon, P., & Curtis, S. (1996). Modelling inequality in reported long term illness in the UK: Combining individual and area characteristics. Journal of Epidemiology and Community Health, 50, 366–376. Skogan, W. (1990). Disorder and decline. Berkeley, CA: University of California Press. Snijders, T., & Bosker, R. (1999). Multilevel analysis. London: Sage. Sooman, A., & Macintyre, S. (1995). Health and perceptions of the local environment in socially contrasting neighbourhoods in Glasgow. Health and Place, 1, 15–26. Taylor, M. F. (Ed.) (1999). British household panel survey user manual volume A: Introduction. Technical Report and Appendices. Colchester: University of Essex. Tienda, M. (1991). Poor people and poor places: deciphering neighborhood effects on poverty outcomes. In J. Huber (Ed.), Macro-micro linkages in sociology (pp. 244–267). Thousand Oaks CA: Sage. Townsend, P. (1979). Poverty in the United Kingdom. Harmondsworth: Penguin. Townsend, P., Phillimore, P., & Beattie, A. (1989). Health and deprivation. London: Routledge. Waitzman, N., & Smith, K. (1998). Phantom of the area: Poverty-area residence and mortality in the United States. American Journal of Public Health, 88, 973–976. Wilkinson, R. G. (1996). Unhealthy societies. The afflictions of inequality. London: Routledge.