IJE Advance Access originally published online on February 28, 2006
International Journal of Epidemiology 2006 35(3):607-613; doi:10.1093/ije/dyl010
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Article |
Does area-based social capital matter for the health of Australians? A multilevel analysis of self-rated health in Tasmania
1 Key Centre for Women's Health in Society, University of Melbourne, Melbourne, Australia
2 School of Public Health, Queensland University of Technology, Brisbane, Australia
3 Department of Society, Human Development and Health, Harvard School of Public Health, Boston, MA, USA
* Corresponding author. Key Centre for Women's Health in Society, School of Population Health, University of Melbourne 0310, Australia. E-mail: a.kavanagh{at}unimelb.edu.au
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Background Material circumstances and collective psychosocial processes have been invoked as potential explanations for socioeconomic inequalities in health; and, linking social capital has been proposed as a way of reconciling these apparently opposing explanations.
Methods We conducted multilevel logistic regression of self-rated health (fair or poor vs excellent, very good, or good) on 14 495 individuals living within 41 statistical local areas who were respondents to the 1998 Tasmanian Healthy Communities Study. We modelled the effects of area-level socioeconomic disadvantage and social capital (neighbourhood integration, neighbourhood alienation, neighbourhood safety, social trust, trust in public/private institutions, and political participation), and adjusted for the effects of individual characteristics.
Results Area-level socioeconomic disadvantage was associated with poor self-rated health (ß = 0.0937, P < 0.001) an effect that was attenuated, but remained significant, after adjusting for individual characteristics (ß = 0.0419, P < 0.001). Social trust was associated with a reduction in poor self-rated health (ß = –0.0501, p = 0.008) and remained significant when individual characteristics (ß = –0.0398, P = 0.005) were included. Political participation was non-significant in the unadjusted model but became significant when adjusted for individual characteristics (ß = –0.2557, P = 0.045). The effects of social trust and political participation were attenuated and became non-significant when area-level socioeconomic disadvantage was included.
Conclusion Area-based socioeconomic disadvantage is a determinant of self-rated health in Tasmania, but we did not find an independent effect of area-level social capital. These findings suggest that in Tasmania investments in improving the material circumstances in which people live are likely to lead to greater improvements in population health than attempts to increase area-level social capital.
Keywords Social capital, socioeconomic disadvantage, multilevel analysis
Accepted 10 January 2006
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
There now exists a large and growing literature documenting an association between socioeconomic position (SEP) and health1–4: socioeconomically disadvantaged groups have higher mortality rates for most major causes of death1–3 and their morbidity pattern indicates that they experience more ill health.2,4
Explanations for socioeconomic inequalities in health are frequently presented as being consistent with one of two potential hypotheses: neo-material and psychosocial.
The neo-material perspective views the relationship between SEP and health as being a consequence of how society is structured and organized, and the extent to which governments invest in economic resources and human capital.5–7 Socioeconomic health inequalities are seen to originate primarily from the differential distribution of resources and capital, variations in exposure to adverse material and economic conditions at critical periods (e.g. infancy and childhood), and accumulated negative experiences and exposures across the lifecourse.5–8
In contrast, proponents of the psychosocial perspective argue that socioeconomic inequality results in less trusting and reciprocal relations among individuals, and lower levels of civic and political participation, leading to the erosion of social capital and a loss of social cohesion. Authors have postulated that social cohesion has its effects on health through three pathways. First, lack of social cohesion produces negative individual psychological responses, such as stress and anxiety, which influence health via complex neuroendocrine and immunological mechanisms.9–11 Second, social cohesion facilitates the formation of shared norms that promote healthy behaviours and negatively sanction unhealthy practices.12 Third, social cohesion provides the context for political action that can be mobilized to prevent activities that might have negative impacts on communities (such as private land development) or to lobby for more resources (such as public transport) that may make areas more health-promoting.13 The last two pathways are often regarded as social capital resources that flow from high levels of social cohesion.14
Studies of social capital and mortality conducted in the US, Canada, and Europe have found that areas with high stocks of social capital have lower mortality rates, an effect that appears to be independent of area-level social disadvantage.9,15–18 Multilevel studies have found associations between area-level social capital and mortality,19 self-rated health,20–24 health service use,25 walking,26 women's use of anxiolytic-hypnotic drugs and hormone replacement therapy,27,28 and homicide.29 These effects have largely been independent of individual-level and area-level social disadvantage.
In a recent issue of this Journal Szreter and Woolcock suggested the neo-material and psychosocial perspectives might be reconciled by theoretical refinements to the concept of social capital. To date, social capital researchers have drawn on the notions of bonding and bridging social capital, which refer to relationships between socially similar (e.g. class, ethnicity) and dissimilar individuals and groups, respectively. Szreter and Woolcock suggest there is a need for an additional form of social capital to be considered, that of linking social capital, which refers to the quality of vertical ties that exist between individuals and groups who are explicitly recognized as unequal (such as between local governments and citizenry). If communities are rich in bonding and bridging capital but weak in linking capital this may enable individuals and groups to access some resources (such as emotional support) but may exclude them from resources that flow from vertical ties such as information about jobs.30
In this paper we provide further empirical evidence for the debate regarding the importance of neo-material and psychosocial explanations in a multilevel study of self-rated health in Tasmania, Australia. We test the relative importance of the different components of social capital posited by Szreter and Woolcock using six social capital variables, which we have called: social trust (bonding), trust in public and private institutions (linking), neighbourhood integration (bonding and bridging), neighbourhood alienation (lack of bonding and bridging), and two variables—neighbourhood safety and political participation—that we conceptualize as social capital resources.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Setting
The study is based in Tasmania, an island state of Australia, which in 1996 had a population of 459 658.31 Tasmania is divided into administrative units known as Statistical Local Areas (SLAs), which broadly correspond with council boundaries defined by Local Government Areas. Tasmania comprises 43 SLAs that range in geographic area (28.5–9575 km2) and in population size (in 1996) from 914 to 59 618 persons. Two pairs of contiguous SLAs (Hobart inner and Hobart remainder; Launceston inner and Launceston Part B) were combined owing to very small population sizes.
With the exception of a measure for area-based socioeconomic disadvantage all variables were derived from individual responses to the Tasmanian Healthy Communities Survey (HCS). Details of the survey have been published elsewhere.32 Briefly, the HCS was a postal survey of 25 000 Tasmanian adults aged 18 years or older selected from the state electoral roll (voting is compulsory for Australian citizens). The electoral roll was divided into age-gender strata within each SLA to ensure representation from all Tasmanians. The survey achieved a 60% response rate (n = 15 112).
Outcome variable
Respondents were asked if 'In general, would you say your health is excellent, very good, good, fair or poor? And this was subsequently recoded as a dichotomous outcome (1 = fair/poor; 0 = excellent, very good, good). Respondents who did not answer this question (n = 295) were excluded from the analysis.
Compositional variables
Individual-level variables that may be important predictors of self-rated health were included: age, Indigenous status, marital status (married, separated/divorced, widowed, single), education (highest level achieved), most recent occupation (white collar, sales and clerical, blue collar, never in the paid workforce), household income (divided into quintiles), and smoking status (current smoker, past smoker, non-smoker. We fitted missing value categories for the variables measuring Indigenous status, occupation, education, and household income, which had between 2.3 and 19.0% missing values.
Area-level social capital
We used questions from the HCS to create individual summary scores for the six social capital variables. These were aggregated within each SLA to produce a mean value for each SLA for each of the variables. We have described in detail the methods we used to construct these measures33; below we provide a brief explanation.
Using principal components analysis (PCA) with varimax rotation we obtained factor solutions to questions about trust and neighbourhood perceptions. The trust question was a 12 item stem question that asked ’In general, how often can you trust each of the following to act in your best interests?‘ with options ranging from 1 = Never to 5 = Always, and 9 = Not Applicable. Five items were removed because of high levels of non-applicability or because they had low component loadings and/or cross-loaded on more than one component. We obtained a two-factor solution. Trust in public and private institutions loaded on (from highest to lowest) items about public servants, government, large corporations, local council, and small business. Social trust loaded highly on items about relatives and friends.
The neighbourhood question was a 13 item stem question that asked ’What do you think about the neighbourhood that you live in? How much do you agree with the following statements?‘ Response options ranged from 1 = Strongly Disagree to 5 = Strongly Agree, and 9 = Not Applicable. Three items were removed owing to low loading and/or cross-loading. The final solution resulted in three factors that were interpreted as ’neighbourhood integration‘, ‘neighbourhood alienation’, and ‘neighbourhood safety’. Neighbourhood integration loaded highly on items about: sorry to move away, have a lot in common with people in neighbourhood, treated with respect, like living in neighbourhood, and trust neighbours to look after property. Neighbourhood alienation loaded on: no one would notice if I no longer lived here and I have little to do with people in neighbourhood. Neighbourhood safety loaded on: children are safe to walk around neighbourhood during the day and it is safe to walk around neighbourhood at night.
Standardized scoring coefficients were calculated for the items forming the five factors (two trust and three neighbourhood) and these were used to derive factor scales for each of the constructs. These scales were re-scored to range from 0 to 10, with higher scores indicating greater average levels of the factor (e.g. higher levels of social trust) in the SLA.
A measure of political participation was derived from the stem question ‘In the past 12 months have you done any of the following?’ with seven possible political activities such as signed a petition or contacted a state or federal MP (coded 0 = No and 1 = Yes). The responses to these seven activities were summed (potential range 0–7).
Area-based socioeconomic status
Area-level SES for each SLA was measured using its index of relative socioeconomic disadvantage (IRSD) score. IRSD scores are derived by the Australian Bureau of Statistics using PCA, and they reflect the overall level of socioeconomic disadvantage of an area measured on the basis of attributes such as low income, low educational attainment, high levels of public sector housing, high unemployment, and jobs in relatively unskilled occupations.34 The IRSD scores used in this present study were calculated using data from the 1996 Australian Census and were scaled across the 41 SLA to have values from 0 to 10, with higher scores indicating greater levels of socioeconomic disadvantage.
Analysis
We applied multilevel statistical procedures,35–38 to model the individual and contextual variation in self-rated poor health. Specifically we calibrated a two-level weighted (for sampling and response probability) binary logistic model with a nested structure: 14 495 individuals (Level 1) nested within 41 SLAs. Fixed and random parameter estimates and their standard errors are quasi-likelihood based with second order Taylor series expansion,38 as implemented within the MLwiN program.39
We built the models in complexity to test for the independent effects of area-SES and social capital. First, we present the effects of the Level-1 fixed effects (age, gender, marital status, education, occupation, income, and smoking) and second, we assessed the effects of area-level variables (as fixed effects at Level 2) with and without adjustment for the Level-1 fixed effects. Finally, to test whether there was an independent effect of social capital, we assessed the effects of each of the social capital variables after adjusting for area-SES and the Level-1 fixed effects. The results for all models are presented as regression coefficients and the ratio of the coefficient to its standard error: ratio values greater than ±2 are significantly different from 0 at P 
0.05.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Table 1 shows the data definition and structure and details of the individual and area-level variables.
|
In a variance components model there was significant variation between areas in self-rated health [
2(between areas) = 0.0367; SE = 0.0136; P < 0.001]. Table 2 shows the effects of the individual-level predictors on self-rated health. Poor self-rated health was more likely in widowers, Indigenous Australians, those with lower levels of education (primary school or Year 10), blue collar workers, and those never in paid work. There is a clear gradient of increasing risk of rating health as poor or fair from the highest to the lowest income quintile categories. After adjustment for individual-level predictors there was no significant variation between areas in self-rated health (P = 0.16).
|
Table 3 shows the association between area-level disadvantage, social capital, and self-rated health prior to (Model 2) and after adjustment (Model 3) for the individual-level fixed effects. Area-SES is associated with an increased risk of reporting fair or poor health even after adjustment for compositional factors (P < 0.001). Living in areas with higher levels of social trust reduces the risk of having poor or fair self-rated health (P = 0.005), an effect that remains significant after adjustment for compositional factors (P = 0.008). After accounting for compositional factors political participation became significant; in areas where high levels of political participation individuals were less likely to rate their health as fair or poor (P = 0.045). Trust in public and private institutions is associated with poor self-rated health (P = 0.037), however, after adjustment for individual variables it became non-significant.
|
Adjusting for area-SES attenuated the effect of social trust and political participation so that they became statistically non-significant while the effects of area-SES remained statistically significant (results not shown in table).
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
The findings of our study suggest that in Australia, health is more likely to be influenced by neo-material than psychosocial pathways. We found strong evidence for the effect of area-SES, with individuals living in more socioeconomically deprived areas being more likely to rate their health as fair or poor even after adjustment for individual characteristics. There was some evidence for the effects of two social capital variables—social trust and political participation—with people living in areas with higher stocks of these factors being less likely to have poor self-rated health. However, the effects of social trust and political participation became statistically non-significant, although they remained protective, after adjustment for area-SES.
These findings are consistent with our recent limited multilevel study (age and gender at Level 1 only) in Tasmania, where we found that area-SES was associated with higher mortality rates but did not find evidence that there was an independent effect of social capital (using the six measures in this study).33
Neo-material or psychosocial pathways?
Our findings lend support for the neo-material perspective and are consistent with other studies, which have demonstrated a consistent association between area-SES and self-rated health.20,40,41
We did not find empirical support for Szreter and Woolcock's theoretical reframing of social capital to include linking social capital. Trust in public and private institutions captures the quality of the vertical ties that Szreter and Woolcock describe as linking social capital. Theoretically, high levels of linking social capital enables the redistribution of resources,30 however, the effect of this variable appears to be null or, if anything, increases the probability of poor self-rated health (Table 3). In theory, linking social capital can potentially alleviate the deleterious impact of living in socioeconomically disadvantaged areas but we did not find any support for this hypothesis; the effects of area-SES remained strong even after adjustment for trust in public and private institutions. Indeed, it seems likely that people living in more advantaged areas may feel more comfortable operating across institutional power differentials so that we might expect to observe higher levels of linking social capital in more advantaged areas. However, there was no correlation between area-SES and trust, and public and private institutions (r = 0.06).33
It is possible that when investments are made in improving the quality of local environments the benefits of social capital, particularly linking social capital, may be observed. Szreter and Woolcock argue the importance of linking social capital using the case of access to hip replacements as an example; they argue that disadvantaged individuals need to know that they can access hip replacements before they can benefit. This information may flow to them in conditions where there are high levels of linking social capital.30 However, in the absence of services for hip replacements no amount of linking social capital will improve access. Our study provides evidence for the need to improve disadvantaged local environments (equivalent to the provision of services for the hip replacement). It is only then that the theoretical beneficial effects of linking social capital, in terms of enabling the distribution and redistribution of resources, might occur.
While we did not find any evidence for the effects of linking social capital, we found limited evidence for the benefits of bonding social capital, which facilitates the transfer of resources across networks of socially similar individuals and groups. While bonding social capital may be essential for the distribution of emotional, instrumental, and informational support, it does not enable the redistribution of resources from advantaged to the disadvantaged. Political participation was protective for health after adjustment for individual risk factors. Political participation is a resource that theoretically flows from social capital infrastructure, such as high levels of trust and reciprocity.14 An outcome of high levels of political participation may be the procurement of additional social and material resources, at least in the context of a responsive public sector. Like social trust, political participation became statistically insignificant after adjustment for area-SES.
Although we show that area-SES is an important determinant of self-rated health, we are unable to ascertain what it is about living in areas with higher levels of socioeconomic disadvantage that is deleterious for self-rated health. The effect of area-SES may reflect differences in aspects of the environment including physical infrastructure (e.g. access to nutritious food, public spaces for physical activity), the provision of services (e.g. health services), the quality of housing, and levels of pollution. Future studies need to ascertain what it is about living in disadvantaged environments that is bad for health using detailed audits of the local environments.
We believe that our findings are robust because, with the exception of political participation, the social capital variables were derived by linearly combining individual factor-scores obtained in a PCA and these variables had similar distributional properties (mean and standard deviation) as area-SES.33 In addition, we believe that the social capital variables we have used are likely to be valid because: they have good face validity, tapping the multiple facets of social capital (e.g. infrastructure such as trust and resources such as political participation); the social capital variables are moderately correlated so we are probably capturing different dimensions of similar latent constructs33; and, as with other Australian studies42,43 rural areas have higher levels of social capital than urban areas. Furthermore, while most Australian studies have used slightly different questions and have conceived of social capital as an individual-level variable, we were able to compare the individual frequencies of two political activities in the past 12 months—signing a petition and attending a protest meeting—with the frequency of the same activities in a South Australian study.42 We found similar proportions of women and men signed a petition (women 50% and men 45%) and attended a protest meeting (women 6.5% and men 7.6%) as that reported by Baum and colleagues.42 This suggests that we have a similar distribution of social capital as that found in at least one other Australian state.
The fact that we included many individual-level variables reduces the likelihood that unmeasured or poorly measured individual confounders44–49 can explain our findings; it is possible that there is over-adjustment as some of the individual variables may lie on the causal pathway between the area-level SES and social capital, and self-rated health. However, adjusting for compositional factors did not result in much attenuation of social trust, which was the only social capital variable that was significant in the unadjusted models (Table 3). We describe the ‘average’ effect of the area-level variables across individual SES groups; it is possible that the effects of these variables are not uniform across SES groups. However, we tested for cross-level interactions between household income and each of the area-level measures and found no evidence for effect modification of area variables by household income (results not shown) suggesting that our approach is appropriate.
There are some limitations of this study, which should be borne in mind. First, we present neo-material and psychosocial pathways as separate pathways (in line with previous commentators) as a heuristic device for statistical modelling purposes. However it is important to note that these pathways are not entirely separate and may feed into each other. For example, as suggested above, the quality of the material environment may emanate from strong local political participation. Second, SLAs may not be the most appropriate spatial scale for measuring area-level SES and social capital in Tasmania and, if this is the case, we may have underestimated the area-level effects.50 SLAs vary in geographic and population size but mostly do correspond with boundaries for local government. Local governments, who are dependent on local land taxes (taxed as a proportion of a property's value), are responsible for much of the local infrastructure planning and implementation. Therefore, there are plausible reasons as to why these would be the most appropriate spatial scale to measure area-SES. However, SLAs are unlikely to be the most appropriate unit for the exploration of social capital, which may operate at a more micro-scale such as neighbourhoods. This means that it is possible that we have underestimated social capital effects. Third, we had a large proportion of missing data on key variables, particular individual SES, which could potentially bias our results. We fitted models, which only included observations that did not have missing data. These models, conducted on a smaller sample (n = 9878), showed some attenuation of the effects of area-SES and political participation but the direction of the effects was unchanged. Finally, it is possible that the social capital measures are subject to appreciable measurement error because they are based on people's perceptions of their local environments. Area-SES, on the other hand, is captured using more objective measures, such as income, education, housing tenure, and so forth. If the social capital variables are more likely to be affected by measurement error then this will increase our type 2 error for these measures relative to area-SES.
|
Conclusion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
This study finds support for the neo-material pathway linking disadvantage to health.5–7 This suggests that gains in population health in Tasmania are more likely to be made by identifying the characteristics of socioeconomically disadvantaged areas that place residents at risk of poorer health outcomes. Political commitment is needed to facilitate the redistribution of resources to individuals living in disadvantaged areas through investments in physical infrastructure, human capital (such as public education), and financial capital (such as tax transfers).
KEY MESSAGES
|
|
Acknowledgments |
|---|
A.M.K. is supported by a Victorian Health Promotion Foundation (VicHealth) Principal Research Fellowship. G.T. is supported by a National Health and Medical Research Council/National Heart Foundation Career Development Award (CR 01B 0502).
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
1 Shaw M, Gordon D, Dorling D, Davey Smith G. The Widening Gap: Health Inequalities and Policy in Britain. Bristol: The Policy Press, 1999.
2 Adler N, Ostrove J. Socioeconomic status and health: what we know and what we don't. Ann N Y Acad Science 1999;896:3–15.[CrossRef][Web of Science][Medline]
3 Turrell G, Oldenburg B, McGuffog I, Dent R. Socioeconomic Determinants of Health: Towards a National Research Program and a Policy and Intervention Agenda. AusInfo, Canberra: Queensland Univerity of Technology, School of Public Health, 1999.
4 Turrell G, Mathers C. Socioeconomic inequalities in all-cause and specific cause mortality in Australia: 1985–1987 and 1995–1997. Int J Epidemiol 2001;30:231–39.
5 Lynch J, Davey Smith G, Kaplan G, House J. Income inequality and mortality: Importance to health of individual income, psychosocial environment or material conditions. BMJ 2000;320:1200–04.
6 Lynch J, Davey Smith G, Harper S, Hillemeier M, Ross N, et al. Is Income Inequality a Determinant of Population Health? Part 1. A Systematic Review. Milbank Q 2004;82:5–99.[CrossRef][Web of Science][Medline]
7 Muntaner C, Lynch J, Hillemeier M, Lee JH, David R, Benach J et al. Economic inequality, working-class power, social capital, and cause-specific mortality in wealthy countries. Int J Health Serv 2002;32:423–32.[CrossRef][Web of Science][Medline]
8 Turrell G. Income inequality and health: In search of fundamental causes. In: Eckersley R, Dixon J, Douglas R (eds). The Social Origins of Health and Well-Being. Melbourne: Cambridge University Press, 2001, pp. 83–104.
9 Wilkinson R. Inequality and the social environment: a reply to Lynch. J Epidemiol Community Health 2000;54:411–13.
10 Wilkinson R, Kawachi I, Kennedy B. Mortality, the social environment, crime and violence. Sociol Health Illn 1998;20:578–97.[CrossRef]
11 Marmot M, Wilkinson R. Psychosocial and material pathways in the relation between income and health: a response to Lynch et al. BMJ 2001;322:1233–36.
12 Kawachi I, Berkman L. Social cohesion, social capital and health. In: Berkman L, Kawachi I (eds). Social Epidemiology. New York: Oxford University Press, 2000, pp. 174–90.
13 Coleman J. Foundations of Social Theory. Cambridge: Harvard University Press, 1990.
14 Ziersch A, Baum F, MacDougall C, Putland C. Neighbourhood life and social capital: the implications for health. Soc Sci Med 2005;60:71–86.[CrossRef][Web of Science][Medline]
15 Skrabski A, Kopp M, Kawachi I. Social capital in a changing society: cross-sectional associations with middle aged female and male mortality rates. J Epidemiol Community Health 2003;57:114–20.
16 Kawachi I, Kennedy B, Lochner K, Prothrow-Smith D. Social capital, income inequality and mortality. Am J Public Health 1997;87:1491–98.
17 Kennedy B, Kawachi I, Prothrow-Smith D, Gupta V. Social capital, income inequality, and firearm violent crime. Soc Sci Med 1998;47:7–17.[CrossRef][Web of Science][Medline]
18 Veenstra G. Social capital and health (plus wealth, income inequality and regional health governance. Soc Sci Med 2002;54:849–68.[CrossRef][Web of Science][Medline]
19 Lochner K, Kawachi I, Brennan R, Buka S. Social capital and neighbourhood mortality rates in Chicago. Soc Sci Med 2003;56:1797–805.[CrossRef][Web of Science][Medline]
20 Subramanian S, Kawachi I, Kennedy B. Does the state you live in make a difference? Multilevel analysis of self-rated health in the US. Soc Sci Med 2001;53:9–19.[CrossRef][Web of Science][Medline]
21 Subramanian S, Daniel K, Kawachi I. Social trust and self-rated health in US communities: a multilevel analysis. J Urban Health 2002;79:S21–34.[Web of Science][Medline]
22 Kawachi I, Kennedy B, Glass R. Social capital and self-rated health: a contextual analysis. Am J Public Health 1999;89:1187–93.
23 Browning C, Cagney K. Neighbourhood structural disadvantage, collective efficacy, and self-rated physical health in an urban setting. J Health Soc Behav 2002;43:383–99.[CrossRef][Web of Science][Medline]
24 Wen M, Browning C, Cagney K. Poverty, affluence, and income inequality: neighbourhood economic structure and its implications for health. Soc Sci Med 2003;57:843–60.[CrossRef][Web of Science][Medline]
25 Hendryx M, Ahern M, Lovrich N, McCurdy A. Access to health care and community social capital. Health Serv Res 2002;31:85–101.
26 Fischer J, Li F, Cleveland M. Neigbourhood influences on physical activity among older adults. J Aging Phys Act 2004;11:49–67.
27 Johnell K, Merlo J, Lynch J, Blennow G. Neighbourhood social participation and women's use of anxiolytic-hypnotic drugs: a multilevel analysis. J Epidemiol Community Health 2004;58:59–64.
28 Merlo J, Lynch J, Yang M, Lindstrom M, Ostergren PO, Rasmusen NK et al. Effect of neighborhood social participation on individual use of hormone replacement therapy and antihypertensive medication: a multilevel analysis. Am J Epidemiol 2003;157:774–83.
29 Sampson R, Raudenbush S, Earls F. Neighbourhoods and violent crime: a multilevel study of collective efficacy. Science 1997;277:918–24.
30 Szreter S, Woolcock M. Health by association? Social capital, social theory and the political economy of public health. Int J Epidemiol 2004;33:650–667.
31 Public Health Information and Development Unit. Social Health Atlas of Australia. Adelaide: Open Book Publishers, 1999.
32 Tasmanian Department of Health and Human Services. First Results of the Healthy Communities Survey 1998. Hobart: Health and Wellbeing Outcomes Unit, 1999.
33 Turrell G, Kavanagh A, Subramanian S. Area variation in mortality in Tasmania (Australia): the contributions of socioeconomic disadvantage, social capital and geographical remoteness. Health Place in press.
34 Australian Bureau of Statistics. Socioeconomic Indexes for Areas, 1996. Canberra: AGPS, 1998.
35 Subramanian SV. Multilevel methods for public health research. In: Kawachi I, Berkman L (eds). Neighborhoods and Health. New York: Oxford University Press, 2003, pp. 66–111.
36 Subramanian SV. Multilevel methods, theory and analysis. In: Anderson N (ed.). Encyclopedia on Health and Behavior. Thousand Oaks, CA: Sage Publications, 2004, pp. 602–08.
37 Subramanian SV. The relevance of multilevel statistical models for identifying causal neighborhood effects. Soc Sci Med 2004;58:1961–67.[CrossRef][Web of Science][Medline]
38 Goldstein H. Multilevel Statistical Models. London: Arnold, 2003.
39 Rasbash J, Browne W, Goldstein H, et al. A User's Guide to MLwiN. London: Multilevel Models Project, 2000.
40 Stafford M, Martikainen P, Lahelma E, Marmot M. Neighbourhoods and self rated health: a comparison of public sector employees in London and Helsinki. J Epidemiol Community Health 2004;58:772–78.
41 Torsheim T, Currie C, Boyce W, Kalnins I, Overspeck M, Haugland S. Material deprivation and self-rated health: a multilevel study of adolescents from 22 European and North American countries. Soc Sci Med 2004;59:1–12.[CrossRef][Web of Science][Medline]
42 Baum FE, Bush R, Modra C, Murray Cj, Cox EM, Alexander KM, et al. Epidemiology of participation: an Australian community study. J Epidemiol Community Health 2000;54:414–23.
43 Bullen P, Onyx J. Social Capital: Family Support Services and Neighbourhood and Community Centres in NSW. NSW: Family Support Services and Neighbourhood and Community Centres in NSW, 1999.
44 Blakely T, Woodward A. Ecological effects in multi-level studies. J Epidemiol Community Health 2000;54:367–74.
45 Diez-Roux A. Bringing context back into epidemiology: variables and fallacies in multilevel analysis. Am J Public Health 1998;88:216–22.
46 Diez Roux AV. Estimating neighbourhood health effects: the challenges of causal inference in a complex world. Soc Sci Med 2004;58:1953–60.[CrossRef][Web of Science][Medline]
47 Diez-Roux A. Investigating neighbourhood and area effects on health. Am J Public Health 2001;91:1783–89.
48 Diez-Roux AV. Multilevel analysis in public health research. Ann Rev Public Health 2000;21:171–92.[CrossRef][Web of Science][Medline]
49 Sampson R, Morenoff J, Gannon-Rowley T. Assessing ‘Neighbourhood Effects’: social processes and new directions in research. Annu Rev Sociol 2002;28:443–78.[CrossRef][Web of Science]
50 Boyle M, Willms J. Place effects for areas defined by administrative boundaries. Am J Epidemiol 1999;149:577–85.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. van Hooijdonk, M. Droomers, I. M Deerenberg, J. P Mackenbach, and A. E Kunst The diversity in associations between community social capital and health per health outcome, population group and location studied Int. J. Epidemiol., December 1, 2008; 37(6): 1384 - 1392. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Agyemang, C. v. Hooijdonk, W. Wendel-Vos, E. Lindeman, K. Stronks, and M. Droomers The association of neighbourhood psychosocial stressors and self-rated health in Amsterdam, The Netherlands J Epidemiol Community Health, December 1, 2007; 61(12): 1042 - 1049. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I A Jacobs, M T Podobny, and D Bilusich One mechanism underlying contrasting health-economy findings Int. J. Epidemiol., August 1, 2007; 36(4): 929 - 931. [Full Text] [PDF]
|
|
|
|
|
|
T. Blakely and V. Ivory Commentary: Bonding, bridging, and linking--but still not much going on Int. J. Epidemiol., June 1, 2006; 35(3): 614 - 615. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||