Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Social capital and health in rural and urban communities in South Australia

Social capital and health in rural and urban communities in South Australia On almost every indicator of health and social disadvantage residents of rural communities in Australia fare worse than their urban counterparts. Rural Australians have shorter life expectancy, higher death rates, are more likely to have a disability, and are also worse off on a range of indicators of disadvantage. Poorer health in rural areas is also reported in many other developed countries. Despite this health picture, life in rural areas is often seen as being higher in social capital, as expressed through a greater sense of community and social involvement, than in urban areas. But there is relatively little comprehensive research comparing rural and urban areas on measures of social capital, and we do not know whether social capital has similar associations with health in rural areas as those reported in urban studies. This study uses data from a telephone survey to compare the patterns of social capital and their relationship with health for those living in rural and urban South Australia. Social capital and health Social capital is a theoretically contested construct with considerable debate on the best ways of conceptualising it (see refs 6 and 7 ) for a summary of these debates). Key debates include: whether social capital is an individual or a community‐level construct and relatedly whether it is a ‘public good’; whether there are inequities in its distribution; whether it distracts attention from the material basis of health inequities; and whether a consideration of social capital offers an advantage over a consideration of its constituent parts (e.g. trust, reciprocity). We draw on the theories of French sociologist Bourdieu who defined social capital as “the aggregate of the actual or potential resources which are linked to possession of a durable network of more or less institutionalised relationships of mutual acquaintance and recognition” (p 248). This definition focuses on the resources that flow to individuals as a result of their membership of social networks. Bourdieu argues that conflict is a fundamental dynamic of all social life and that this conflict occurs over symbolic assets such as social capital, as well as material assets. He sees social capital as one of the means by which economic capital is reproduced in better off groups in society such that those with more economic capital have greater access to social capital and that this social capital in turn facilitates access to economic capital. We conceive of social capital as comprising both an individual's social infrastructure (which includes people's formal and informal social networks, and values such as trust and reciprocity that might facilitate the exchange of resources), and the potential resources that this infrastructure enables them to accumulate. We conceptualise social capital as a means to consider how individuals may differ in their access to particular components of infrastructure (e.g. trust, social networks) which may then differentially provide individuals access to resources such as help and assistance. By considering these individual components in the one analysis it is possible to build up a picture of access to ‘social capital’, without losing the respective strengths and weaknesses by using an overall measure. We use Bourdieu's approach rather than that proposed by Robert Putnam reflecting the communitarian school of thought. Putnam conceived of social capital as a community‐level resource and defined it as “ features of social organisation such as networks, norms and social trust that facilitate coordination and cooperation formutual benefit” (p 67). This view sees social capital as a public good of communities and does not explicitly consider how particular sub‐groups or individuals within communities with fewer material resources may be less able to access it, and how this can perpetuate inequity. There is a substantial literature linking social capital to both mental and physical health, with relationships found between both individual and community level indicators of social capital (see 11–14 for reviews). However, there are significant debates, both conceptual and methodological, about the nature of this relationship. For example, research findings have differed depending on the measures of social capital used (as outlined in above mentioned reviews), in particular whether it is measured at an individual or community level. It has also been acknowledged that the relationships between social capital and health are likely to be bidirectional. In this paper we conceive of, and measure, social capital at the level of the individual and focus on the relationship of social capital to health, though acknowledge that the direction may also be the reverse. Social capital in rural and urban areas Rural communities are traditionally seen as having higher levels of social capital. It is argued that lower population density encourages connections between residents, and that isolation and lower levels of public services in rural areas facilitates network cooperation and exchange and voluntary activity. However, we do not know if rural areas are higher in social capital across all the indicators or whether there are particular strengths and deficits. There has also been relatively little research into differences between rural and urban communities in terms of social capital and its relationships to health. The limited relevant Australian research on social capital has considered it in terms of its role in agriculture, sustainability and economic development or has generally not compared urban and rural areas. Two studies that have compared rural and urban areas support the view that people in rural communities have higher levels of some forms of social capital. A national study found that participants from rural areas had higher levels of involvement in community organisations, neighbourhood connections and involvement in voluntary community activities. In a study of five rural and urban areas in New South Wales, Australia, Onyx and Bullen found that residents in rural areas had higher levels of participation in community networks, trust and safety and neighbourliness. However, the same study found that rural residents reported lower levels of tolerance of diversity and sense of collective efficacy. International studies have also found higher levels of social capital in rural compared with urban areas, although the findings are not consistent across social capital measures. There is evidence from both the national and international literature that social capital is differentially distributed according to demographic characteristics, such as gender, income and education. The extent to which these demographic differences are mirrored in both rural and urban environments is not known. In terms of rural/urban comparative studies of social capital and health, one study found significant differences in civic involvement between high and low population density areas but did not find that these differences were associated with health outcomes. Nummela et al. found that, after controlling for individual demographic characteristics, a combined measure of trust and participation was associated with self‐assessed health only in urban areas, and not in semi‐urban and rural settings. Our study uses data on social capital and health collected in a South Australian telephone survey to explore three research questions: 1 Are there differences between rural and urban respondents in the socio‐demographic (e.g. age, income) predictors of access to social capital? 2 Do rural respondents have higher levels of all measures of social capital (measured at an individual level) than urban respondents? 3 Are the relationships between social capital and health different for urban and rural respondents? It was anticipated, firstly, that those with greater socio‐economic resources would have more access to social capital regardless of where they lived, and secondly, that people living in rural areas would have greater access to at least some aspects of social capital. Finally, it was anticipated that, regardless of whether or not there were urban/rural differences in social capital, that social capital's importance would differ in urban and rural settings, with social capital potentially more important in rural settings given generally fewer services in rural areas. Methods Participants and procedures Computer assisted telephone interviews (CATI) were conducted with 2,013 individuals in South Australia in July 2003. Three thousand four hundred households were selected from the electronic White Pages and the person over 18 who last had their birthday was asked to complete the survey. Up to 10 separate call backs were made, with an overall participation rate of 75%. The survey questions reported here were commissioned as part of a regular broader survey conducted by the South Australian Department of Health. The survey included the SF‐12 self‐report health measure and questions relating to social capital, neighbourhood life and neighbourhood problems, and demographic questions. Participants were designated as urban or rural on the basis of the Accessibility and Remoteness Index of Australia (ARIA) score of their postcodes. ARIA is a geographical index that defines remoteness as accessibility to service centres based on road distances. ARIA values range from 0 to 12 and in this study urban areas were defined as those which were ‘highly accessible’ (ARIA score 0–1.84) which indicates relatively unrestricted accessibility to a wide range of goods and services. Rural areas were those with ARIA values greater than 1.84. Measures The following measures were used in the analysis 1 Social capital was conceived as comprising infrastructure and the resources available through this infrastructure, seeking to reflect Bourdieu's conceptualisation of social capital, whereby it is not just the connections between people but also the resources available through these connections. Three social capital infrastructure variables were theorised – trust, reciprocity and networks. Networks reflected the connections between people (the structural aspects of social capital infrastructure), and trust and reciprocity were the cognitive aspects reflecting the values likely to facilitate the exchange of resources. Three social capital resource variables – access to help, civic activities and perceptions of neighbourhood cohesion were theorised. Access to help was seen as reflecting access to important practical, emotional and financial assistance; civic activities were seen as an important resource in lobbying for personal and neighbourhood outcomes; and neighbourhood cohesion was seen as an important psychosocial resource. Table 1 outlines the questionnaire items, response format and source of the questions. Summary scores for the six social capital variables of ‘Trust’, ‘Reciprocity’, ‘Networks’, ‘Civic activities’, ‘Help’ and ‘Civic activities’ were calculated and used in the analysis. We have used similar elements of social capital in previous analyses. The methods for calculating these variables are outlined below. 2 Six demographic variables were included in the analysis: Gender (male=1, female=2), Country of Birth (Australian born=1, non‐Australian born=2); Maritial Status (married or defacto=1, other=0); Education (Left school < 15 yrs, left school after 15 /no additional completed qualifications, trade/apprenticeship/certificate/diploma, bachelor degree or higher), Household Income (<$ 12,000, $12,001‐$20,000, $20,001‐$40,000, $40,001‐$60,000, $60,001‐$80,000, >$80,000, with the mid point of each income band used in the analysis to improve the distribution) and Age. 3 Two health variables were included in the analysis: Mental Health was measured using the mental health summary score of the SF‐12 and Physical Health used the physical health summary score of the SF‐12. Models were run separately for mental and physical health. 1 Questionnaire items, source of questions, and factor loadings for social capital variables.* Questionnaire Items Response format Source Factor Loadings SOCIAL CAPITAL INFRASTRUCTURE Networks In the past 12 months how often have you joined in the activities of a community group? (reverse coded) weekly, monthly, occasionally, never (4–1) New 0.754 On average, excluding the people you live with, how often do you socialise with other people (such as friends, neighbours, family or work colleagues)? (reverse coded) every day, most days, once or twice a week, once or twice a month, less often (4–1) New 0.754 Trust Generally speaking, people in Australia can be trusted strongly disagree, disagree, neither agree nor disagree, agree, strongly agree (1–5) Adapted from Ref. 0.593 Generally speaking, you can trust governments As above As above 0.831 Generally speaking, you can trust big business As above As above 0.802 Reciprocity By helping others you help yourself in the long run As above 25 Raw scores used SOCIAL CAPITAL RESOURCES Help If you had a serious personal crisis and you needed help and comfort how many people could you ask for help? none, 1–2, 3–4, 5+ (1–4) 0.807 If you were in financial difficulty and needed to borrow $50 or more how many people could you ask? As above As above 0.841 If you needed to go to a doctor's appointment and you required a lift to get there how many people could you ask? As above As above 0.802 Cohesion This is a close‐knit [neighbourhood/community] strongly disagree, disagree, neither agree nor disagree, agree, strongly agree (1–5) 0.722 People around here are willing to help their neighbours As above As above 0.785 People in this [neighbourhood /community] generally get along with each other As above As above 0.755 People in this [neighbourhood/community] share the same valulues As above As above 0.652 People in this [neighbourhood/community] can be trusted As above As above 0.709 In my [neighbourhood/community] there is a strong sense of coommunity As above As above 0.745 Civic Activities Have you ever picked up other peoples rubbish in a public place? yes/no 0.497 In the past 12 months did you belong to any group that took locaalaction? yes/no 0.724 In the past 12 months have you attended a protest march, meeting or rally? yes/no 0.674 Did you vote in the local government elections in autumn 2003? yes/no New 0.364 * Each item also had a ‘don't know’ and ‘refused’ option . The analysis of physical and mental health was stratified by urban and rural status, so that four models were created. Statistical methods Confirmatory Factor analysis was undertaken in AMOS and tested the theorised model outlined above using the questions outlined in Table 1 . Summary scores were calculated both within AMOS and also using principal components analysis in SPSS. As part of the overall modelling strategy those scores calculated in SPSS had a higher correlation with the outcome variables and so PCA was used to calculate summary scores for the five sub‐factors that had more than one manifest variable, namely: networks, trust, help, civic, and cohesion. Higher values of the summary scores indicate higher levels of the social capital component. Table 1 indicates the loadings for the five social capital variables. Factor loadings can range from ‐1 to +1, with larger values indicating that the item is more strongly related to the latent construct than those with smaller regression estimates. The variable of reciprocity was the raw score answer to one question (see Table 1 ). Using t‐tests, social capital scores were compared for rural/urban respondents and men and women. Structural Equation Modelling Structural equation modelling (SEM) is a collection of statistical techniques that allow the simultaneous examination of a set of relationships between one or more independent variables and one or more dependent variables. It is particularly useful when there is an interest in the interrelationships between the ‘independent’ variables. The MPLUS program was used to undertake the modelling. Four models were undertaken – one each for mental and physical health for both the rural and urban samples. Each model initially considered: • All the paths between each of the demographic variables and all social capital variables to determine if there were demographic differences in access to social capital. • The paths between the three social capital infrastructure variables and the three social capital resource variables in order to examine the ways that particular infrastructure variables may facilitate access to resource variables. • The paths between the demographic and social capital variables and the health variables to consider socio‐demographic predictors of health and the relevance of social capital to health in each setting. Non‐significant paths ( p > 0.05 and coefficient <0.1) were then removed one by one with the least significant paths removed first and the model fit evaluated after each. For variables that directly influenced the final health outcomes, all paths with a p value 0.05 were retained to consider smaller but still potentially interesting relationships. There has been substantial debate about the best measures of model fit to use to evaluate how well the model fits the observed data. Several model fit measures are reported here. We report four relative fit statistics, which compare the proposed model to the null model. The Root Mean Square Error of Approximation (RMSEA) estimates the amount of deviation between the model and the data. RMSEA values below 0.05 are indications of acceptable fit while values larger that 0.07 are indications of important amount of misfit. MPLUS does not provide the more common Goodness of Fit (GFI) value which is a measure of the relative amount of variance and covariance that is jointly explained. Alternatively, Comparative Fit Index (CFI) and TLI were used where a value of >0.9 is considered to be a well fitting model, though TLI statistics are usually lower than CFI figures. We also report the Standardarised Root Mean Square Residual (SRMR) which is the average difference between the predicted and observed variances and covariances in the model, based on standardised residuals. The smaller the SRMR the better the model fit with values of 0.05 widely thought to indicate a good fit. We also include an absolute fit statistic – the Normed Chi‐Square statistic (NCSq) (which is the chi‐square fit index divided by degrees of freedom to make it less dependent on sample size). Opinion differs on acceptable values, though Schumacker and Lomas argue that values less than 5 are adequate. We used these measures to evaluate model fit, and the final models in general met these criteria, though in some cases the TLI scores were below 0.9 (see Figures 1–4 ). 1 Urban mental health model. 2 Urban physical health model. 3 Rural mental health model. 4 Rural physical health model. In this study,‘don't know’ and refusal responses were treated as missing data. There were small numbers of missing data for each of the social capital questions (in most cases well below 3%). These missing cases were imputed five times using NORM software including the other questions for each factor in the model, and the five complete datasets were then combined (averaged) to form a complete single data set. There were 165 cases missing for income (8.2%) and these were allocated to the mean of the midpoint income values as it was a single item measure. Data were weighted to account for an individuals’ probability of selection within a household and also by age and gender in line with the Australian Bureau of Statistics 2001 Census of Population and Housing data for the South Australian population. Results Table 2 illustrates the demographic profile of the respondents and Table 3 shows the social capital and SF‐12 scores for the urban and rural samples. Figures 1 and 2 show the models for mental and physical health respectively for the urban sample and Figures 3 and 4 show the same models for the rural sample. 2 Demographic details of samples (weighted). N (Urban) N (Rural) Total = 1,487 Total: 526 Gender Male 713 (48%) 265 (51%) Female 774 (52%) 260 (49%) Country of Birth Australian 1109 (75%) 457 (87%) non‐Australian 378 (25%) 68 (13%) Marital status Married/defacto 968 (65%) 389 (74%) Other 519 (35%) 136 (26%) Education Left school < 15yrs 227 (15%) 103 (20%) Left school after 15/no additional completed qualifications 477 (32%) 203 (39%) Trade/apprenticeship/certificate/diploma 491 (33%) 167 (32%) Bachelor degree or higher 292 (20%) 53 (10%) Income Up to $12,000 105 (7%) 38 (7%) $12,001 ‐ $20,000 176 (12%) 88 (17%) $20,001 ‐ $40,000 240 (16%) 99 (19%) $40,001 ‐ $60,000 268 (18%) 116 (22%) $60,001 ‐ $80,000 229 (15%) 66 (13%) More than $80,000 337 (23%) 87 (17%) Missing 133 (9%) 32 (6%) Age Mean= 49.6 Mean=51.2 (SD=17.8) (SD=16.5) 3 Social capital factor scores and health summary scores for urban and rural samples (weighted). Urban Mean (SD) Rural Mean (SD) P value Social capital Networks 3.19 (0.97) 3.43 (0.94) <0.000 Trust 3.12 (0.73) 3.06 (0.75) 0.100 Reciprocity 4.25 (0.63) 4.20 (0.61) 0.184 Help 3.98 (0.94) 4.06 (0.90) 0.083 Civic 2.13 (0.82) 2.29 (0.81) <0.000 Cohesion 3.62 (0.64) 3.82 (0.58) <0.000 Health Physical health 49.09 (10.20) 48.73 (10.12) 0.484 Mental health 52.14 (8.92) 53.08 (8.38) 0.031 Social capital Three of the social capital scores were significantly different between rural and urban participants ( Table 3 ). Rural residents reported significantly higher levels of the social capital infrastructure variable of networks and the social capital resource variables of civic activities and cohesion, compared to urban residents. There were no statistically significant differences in ‘Trust’, ‘Reciprocity’ or ‘Help’. In the urban models ( Figures 1 and 2 * ) there were a number of demographic differences in the associations with the social capital variables. Higher trust scores were significantly associated with higher income, indicating that those on higher incomes had greater levels of trust than those on lower incomes. Women had higher levels of ‘Reciprocity’ than men, and those with higher educational achievement had higher levels of ‘Reciprocity’ than those with lower educational achievement. There were no predictors of network involvement. Men and those on higher incomes had higher levels of ‘Help’ than women and those on lower incomes, respectively. Those with higher educational achievement had higher levels of ‘Civic Activities’ than those with lower levels of educational achievement. There were no demographic predictors for ‘Cohesion’. In terms of the relationship between the social capital infrastructure and social capital resource variables those with higher ‘Reciprocity’ scores had higher levels of ‘Help’, ‘Cohesion’ and ‘Civic Activities’; those with higher ‘Networks’ scores had higher levels of ‘Help’, ‘Cohesion’ and ‘Civic Acitivites’; and those with higher ‘Trust’ had higher levels of ‘Cohesion’ In the rural models ( Figures 3 and 4 ) there were also demographic differences in the associations with social capital variables, though these patterns differed to those found in the urban models. There were no demographic predictors of ‘Trust’ or ‘Reciprocity’, and ‘Networks’ decreased with age and was greater for women.‘Help’ was higher for those who had higher incomes and who were married/defacto.‘Cohesion’ increased with income and age. Those who had higher educational achievement had higher levels of ‘Civic Activities’ than those with lower educational achievement. Those who were married/defacto had higher levels of ‘Civic Activities’ than those who were not. The relationship between social capital infrastructure and social capital resource variables were identical to the urban model, with the exception that ‘Reciprocity’ did not predict ‘Civic Activities’ in the rural sample. Health There were no significant differences between the urban and rural samples in terms of ‘Physical Health’ ( p =0.484, Table 3 ). However, rural participants reported significantly better MENTAL HEALTH, compared to urban participants ( Table 3 , p <0.005). In the urban mental health model ( Figure 1 ), GENDER was significantly associated with ‘Mental health’ with men having better mental health than women.‘Marital status’ was also significantly associated with mental health, with individuals who were married or in defacto relationships better off than those of other marital status. Two social capital infrastructure variables,‘Trust’ and ‘Networks’, were associated with ‘Mental health’, with better mental health associated with higher levels of trust and greater network involvement.‘Mental health’ was also associated with the two social capital resource variables of ‘Help’ and ‘Cohesion’, with those with greater access to help and a stronger sense of cohesion having better mental health. In the urban physical health model ( Figure 2 ), physical health was better for those on higher incomes and with higher educational achievement, and was worse with age. Physical health was also positively associated with ‘Trust’ and ‘Help’. In the rural mental health model ( Figure 3 ), men again had better mental health than women, and mental health was better for older people.‘Trust’ was again associated with mental health, as were ‘Help’ and ‘Cohesion’. In the rural physical health model ( Figure 4 ) physical health again increased with income and was worse with age, though there were no differences by education. None of the social capital variables was associated with physical health. Summary of key urban/rural similarities and differences There were different predictors of the three social capital infrastructure variables in the rural and urban models with women having higher levels of reciprocity than men in the urban model, but higher levels of networks instead in the rural model. Network involvement decreased with age only in the rural model. In terms of social capital resource variables, in both the urban and rural models higher income was associated with greater access to help, and higher education was associated with greater involvement in civic activities. The relationships between the infrastructure and the resource variables was almost identical in the rural and urban models. However, there were some rural/urban differences between the models for the social capital resource variables – men had more access to help than women in the urban model but not in the rural model, and higher income was associated with higher cohesion in the rural model but not in the urban model. In terms of health, higher levels of trust, cohesion and help were associated with better mental health in both urban and rural models, and men had better mental health than women in both models, though the relationship was stronger in the rural model. However, networks and marital status were associated with mental health only in the urban model, and mental health increased with age only in the rural model. In both the rural and urban models those on higher incomes had better physical health, and those who were older had worse physical health, and this relationship was stronger in the rural models. In the urban models, trust, education, and help were also associated with physical health, but not in the rural models. Discussion In relation to the three research questions, differences were found between the populations living in rural and urban areas in terms of the predictors and levels of social capital, and the relationship of social capital with health: 1) There were rural/urban differences in the determinants of access to social capital, though in both cases it was generally those with more socio‐economic resources who had greater access; 2) Rural participants had greater access to some, but not all elements of social capital; 3) Social capital was associated with mental health in both areas, but with physical health only in urban areas. Were there rural/urban differences in the socio‐demographic determinants of access to social capital? Income and education were important predictors of levels of social capital. This supports previous research we have undertaken and reinforces the aspects of Bourdieu's theory of social capital which sees it as one of the mechanisms by which advantage is maintained. Gender was also significantly associated with social capital variables though the findings were inconsistent. These and the other demographic differences found highlight the importance of acknowledging the diversity of life in both urban and rural areas and illustrate that access to social capital is not homogenous in either setting. While there were very few differences between the urban and rural samples in terms of the relationships between social capital infrastructure and social infrastructure variables themselves, there were a number of differences in the demographic predictors of these variables, suggesting that the factors that influence access to social capital may vary for urban and rural residents. Compared to the urban sample, in the rural sample there were fewer demographic differences in social capital infrastructure variables and more demographic differences in access to social capital resource variables. In particular in the rural sample marital status appeared more important in predicting access to social capital. Marital status has previously been linked to social capital activities such as volunteering, and social support, and in rural areas can be a particularly important social and economic distinction. Were there rural/urban differences in social capital? As anticipated, respondents from rural areas reported significantly higher levels of a number of elements of social capital, with higher levels of social networks, greater civic participation and more social cohesion. This is consistent with the Australian studies outlined earlier. While the findings relate to individuals within the areas rather than a study of the effects of place (which precludes the positing of direct effects of aspects of area of residence), some aspects of living in rural or urban areas could be used to hypothesise about the findings. For example, higher civic activities in rural areas might be understood in terms of the lesser access to formalised services. Likewise the isolation of rural areas may also help explain a greater importance of social networks, and higher levels of homogeneity in rural areas may contribute to a greater sense of cohesion. Like Beaudoin and Thorson, we found no statistically significant differences in trust or reciprocity. There were also no differences in access to informal help. That there were differences in only some elements of social capital reinforces the need to both consider the elements separately, and in concert (i.e. as part of the total package of social capital), as it adds value in considering strengths and weaknesses in terms of overall access to social infrastructure and related resources. Were there rural/urban differences in the relationship between social capital and health? Despite the well documented differences in health outcomes between rural and urban Australians, there was no significant difference between the rural and urban respondents in terms of self‐reported physical health, a similar finding to Glover et al. There were significant mental health differences, with rural participants faring better than their urban counterparts. In both the rural and urban models, in line with a significant literature, men reported better mental health. Age and marital status were relevant in the rural and urban models respectively. Better physical health was associated with older age and higher income in both the rural and urban models, with education also relevant for the urban model, supporting a significant literature on the social determinants of health. Elements of social capital were more consistently associated with mental health than physical health. This supports findings from previous research The patterns of relationships between social capital and mental health were similar for rural and urban participants. In contrast to mental health, there were rural/urban differences in the relationships between social capital and physical health, with trust and help related to physical health for urban participants, but no social capital variables significant for rural participants’ physical health. This may relate to the particularly important role of trust in heterogenous urban environments where cohesion is often lower. The ‘help’ measure used related to people available to ask for potential help rather than actual help received. It is possible that in rural areas the actual exchange of favours may be more important to health given the generally lower availability of services, but that this is not reflected in terms of who could be asked. Study strengths and limitations This was a large study that drew on a good suite of theoretically reasoned social capital variables to broadly compare populations in rural and urban areas. It had a good response rate and relatively low levels of missing data. The study also had a number of limitations. First, the study aimed to assess rural/urban differences in social capital and its relationship with health and so used cross‐sectional data in order to do this. However, it is not possible to ascertain causal relationships and it is possible that health status may impact on other variables. Likewise, the variance explained by the models was small, suggesting that there are clearly other variables that are relevant to mental and physical health. Second, for parsimony the path analysis only looked at one‐way relationships and did not consider the way that, for example, access to help may lead to increased trust. Likewise, we only examined some social capital resource variables, however the range of variables we considered was more than in most studies of SC and health. Third, as data were collected at the individual level, it is not possible to make direct claims about area‐level effects. This would require a multilevel modelling approach with a much larger sample size of areas, and would be a fruitful area of further research. In addition, we categorised urban and rural very crudely. The ARIA measure used is limited in its ability to capture the heterogeneity within each category, and the categories themselves are broad. A more nuanced consideration these categories would be a useful further extension of this research. Likewise, a lifecourse approach which recognises that people may move between rural and urban areas would also be worthy of further consideration. Social capital may also vary across the lifecourse. Fourth, identical questions were used for both rural and urban participants. However, the meaning of questions may vary between these settings. For example, one of the measures of help relating to travel to the doctor is likely to have measured something slightly different between the two settings given that in sparsely populated rural South Australia the example of a lift to the doctor could be a long trip that would inconvenience the helper significantly. It is a difficult balance between being able to directly compare populations in rural and urban areas on variables measured in the exactly the same way, and a priori using different variables themselves which makes direct comparisons more problematic. Fifth, there may be unknown selection biases in the sample. For example, Aboriginal people are likely to be under‐represented due to the telephone sampling method. There has been little research on patterns of social capital among indigenous peoples and this would be worthy of further examination. Sixth, using Principal Components Analysis to create the social capital summary scores may have led to minor variations between the Confirmatory Factor Analysis undertaken in AMOS, and the subsequent analysis in MPLUS. Seventhly, the relatively low upper band of the household income measure (which was the standard measure collected in particular South Australian Department of Health surveys at the time of data collection) may have led to more conservative estimates of the association of income with other variables. Finally, the quantitative survey can describe associations between variables which are suggestive of possible relationships. Qualitative research is required to obtain a deeper understanding of the postulated relationships in rural and urban areas. Conclusion We found that the particular shape of social capital and its relationship to health differs for respondents in rural and urban areas. While rural respondents had higher levels of some aspects of social capital our findings support the view (which builds on the theories of Bourdieu ) that socio‐economic status is an important correlate of social capital regardless of the setting. While survey data can suggest associations between variables it is limited in what it can tell us about the texture of the social and power relationships that are vital to fully understanding social capital. Consequently detailed qualitative work is required to further unpack these complex relationships, especially in sub‐groups of the population. The different relationships between social capital and health for urban and rural respondents, and the demographic differences within areas, reinforces the need to consider the setting (whether because of the different demographic makeup of an area and/or characteristics of the area itself) when considering how social capital may inform health promotion approaches. Acknowledgements This study was supported by the National Health and Medical Research Council (projects 324724 and 229913). We would like to thank Katy Osborne for her assistance in preparing the paper and the three anonymous reviewers for their helpful comments. Footnotes * For both the rural and urban analyses the results for the relationships between demographic and social capital variables were identical for the mental and physical health models. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Australian and New Zealand Journal of Public Health Wiley

Social capital and health in rural and urban communities in South Australia

Loading next page...
 
/lp/wiley/social-capital-and-health-in-rural-and-urban-communities-in-south-LthQa1IMBW

References (57)

Publisher
Wiley
Copyright
© 2009 The Authors. Journal Compilation © 2009 Public Health Association of Australia
ISSN
1326-0200
eISSN
1753-6405
DOI
10.1111/j.1753-6405.2009.00332.x
pmid
19236353
Publisher site
See Article on Publisher Site

Abstract

On almost every indicator of health and social disadvantage residents of rural communities in Australia fare worse than their urban counterparts. Rural Australians have shorter life expectancy, higher death rates, are more likely to have a disability, and are also worse off on a range of indicators of disadvantage. Poorer health in rural areas is also reported in many other developed countries. Despite this health picture, life in rural areas is often seen as being higher in social capital, as expressed through a greater sense of community and social involvement, than in urban areas. But there is relatively little comprehensive research comparing rural and urban areas on measures of social capital, and we do not know whether social capital has similar associations with health in rural areas as those reported in urban studies. This study uses data from a telephone survey to compare the patterns of social capital and their relationship with health for those living in rural and urban South Australia. Social capital and health Social capital is a theoretically contested construct with considerable debate on the best ways of conceptualising it (see refs 6 and 7 ) for a summary of these debates). Key debates include: whether social capital is an individual or a community‐level construct and relatedly whether it is a ‘public good’; whether there are inequities in its distribution; whether it distracts attention from the material basis of health inequities; and whether a consideration of social capital offers an advantage over a consideration of its constituent parts (e.g. trust, reciprocity). We draw on the theories of French sociologist Bourdieu who defined social capital as “the aggregate of the actual or potential resources which are linked to possession of a durable network of more or less institutionalised relationships of mutual acquaintance and recognition” (p 248). This definition focuses on the resources that flow to individuals as a result of their membership of social networks. Bourdieu argues that conflict is a fundamental dynamic of all social life and that this conflict occurs over symbolic assets such as social capital, as well as material assets. He sees social capital as one of the means by which economic capital is reproduced in better off groups in society such that those with more economic capital have greater access to social capital and that this social capital in turn facilitates access to economic capital. We conceive of social capital as comprising both an individual's social infrastructure (which includes people's formal and informal social networks, and values such as trust and reciprocity that might facilitate the exchange of resources), and the potential resources that this infrastructure enables them to accumulate. We conceptualise social capital as a means to consider how individuals may differ in their access to particular components of infrastructure (e.g. trust, social networks) which may then differentially provide individuals access to resources such as help and assistance. By considering these individual components in the one analysis it is possible to build up a picture of access to ‘social capital’, without losing the respective strengths and weaknesses by using an overall measure. We use Bourdieu's approach rather than that proposed by Robert Putnam reflecting the communitarian school of thought. Putnam conceived of social capital as a community‐level resource and defined it as “ features of social organisation such as networks, norms and social trust that facilitate coordination and cooperation formutual benefit” (p 67). This view sees social capital as a public good of communities and does not explicitly consider how particular sub‐groups or individuals within communities with fewer material resources may be less able to access it, and how this can perpetuate inequity. There is a substantial literature linking social capital to both mental and physical health, with relationships found between both individual and community level indicators of social capital (see 11–14 for reviews). However, there are significant debates, both conceptual and methodological, about the nature of this relationship. For example, research findings have differed depending on the measures of social capital used (as outlined in above mentioned reviews), in particular whether it is measured at an individual or community level. It has also been acknowledged that the relationships between social capital and health are likely to be bidirectional. In this paper we conceive of, and measure, social capital at the level of the individual and focus on the relationship of social capital to health, though acknowledge that the direction may also be the reverse. Social capital in rural and urban areas Rural communities are traditionally seen as having higher levels of social capital. It is argued that lower population density encourages connections between residents, and that isolation and lower levels of public services in rural areas facilitates network cooperation and exchange and voluntary activity. However, we do not know if rural areas are higher in social capital across all the indicators or whether there are particular strengths and deficits. There has also been relatively little research into differences between rural and urban communities in terms of social capital and its relationships to health. The limited relevant Australian research on social capital has considered it in terms of its role in agriculture, sustainability and economic development or has generally not compared urban and rural areas. Two studies that have compared rural and urban areas support the view that people in rural communities have higher levels of some forms of social capital. A national study found that participants from rural areas had higher levels of involvement in community organisations, neighbourhood connections and involvement in voluntary community activities. In a study of five rural and urban areas in New South Wales, Australia, Onyx and Bullen found that residents in rural areas had higher levels of participation in community networks, trust and safety and neighbourliness. However, the same study found that rural residents reported lower levels of tolerance of diversity and sense of collective efficacy. International studies have also found higher levels of social capital in rural compared with urban areas, although the findings are not consistent across social capital measures. There is evidence from both the national and international literature that social capital is differentially distributed according to demographic characteristics, such as gender, income and education. The extent to which these demographic differences are mirrored in both rural and urban environments is not known. In terms of rural/urban comparative studies of social capital and health, one study found significant differences in civic involvement between high and low population density areas but did not find that these differences were associated with health outcomes. Nummela et al. found that, after controlling for individual demographic characteristics, a combined measure of trust and participation was associated with self‐assessed health only in urban areas, and not in semi‐urban and rural settings. Our study uses data on social capital and health collected in a South Australian telephone survey to explore three research questions: 1 Are there differences between rural and urban respondents in the socio‐demographic (e.g. age, income) predictors of access to social capital? 2 Do rural respondents have higher levels of all measures of social capital (measured at an individual level) than urban respondents? 3 Are the relationships between social capital and health different for urban and rural respondents? It was anticipated, firstly, that those with greater socio‐economic resources would have more access to social capital regardless of where they lived, and secondly, that people living in rural areas would have greater access to at least some aspects of social capital. Finally, it was anticipated that, regardless of whether or not there were urban/rural differences in social capital, that social capital's importance would differ in urban and rural settings, with social capital potentially more important in rural settings given generally fewer services in rural areas. Methods Participants and procedures Computer assisted telephone interviews (CATI) were conducted with 2,013 individuals in South Australia in July 2003. Three thousand four hundred households were selected from the electronic White Pages and the person over 18 who last had their birthday was asked to complete the survey. Up to 10 separate call backs were made, with an overall participation rate of 75%. The survey questions reported here were commissioned as part of a regular broader survey conducted by the South Australian Department of Health. The survey included the SF‐12 self‐report health measure and questions relating to social capital, neighbourhood life and neighbourhood problems, and demographic questions. Participants were designated as urban or rural on the basis of the Accessibility and Remoteness Index of Australia (ARIA) score of their postcodes. ARIA is a geographical index that defines remoteness as accessibility to service centres based on road distances. ARIA values range from 0 to 12 and in this study urban areas were defined as those which were ‘highly accessible’ (ARIA score 0–1.84) which indicates relatively unrestricted accessibility to a wide range of goods and services. Rural areas were those with ARIA values greater than 1.84. Measures The following measures were used in the analysis 1 Social capital was conceived as comprising infrastructure and the resources available through this infrastructure, seeking to reflect Bourdieu's conceptualisation of social capital, whereby it is not just the connections between people but also the resources available through these connections. Three social capital infrastructure variables were theorised – trust, reciprocity and networks. Networks reflected the connections between people (the structural aspects of social capital infrastructure), and trust and reciprocity were the cognitive aspects reflecting the values likely to facilitate the exchange of resources. Three social capital resource variables – access to help, civic activities and perceptions of neighbourhood cohesion were theorised. Access to help was seen as reflecting access to important practical, emotional and financial assistance; civic activities were seen as an important resource in lobbying for personal and neighbourhood outcomes; and neighbourhood cohesion was seen as an important psychosocial resource. Table 1 outlines the questionnaire items, response format and source of the questions. Summary scores for the six social capital variables of ‘Trust’, ‘Reciprocity’, ‘Networks’, ‘Civic activities’, ‘Help’ and ‘Civic activities’ were calculated and used in the analysis. We have used similar elements of social capital in previous analyses. The methods for calculating these variables are outlined below. 2 Six demographic variables were included in the analysis: Gender (male=1, female=2), Country of Birth (Australian born=1, non‐Australian born=2); Maritial Status (married or defacto=1, other=0); Education (Left school < 15 yrs, left school after 15 /no additional completed qualifications, trade/apprenticeship/certificate/diploma, bachelor degree or higher), Household Income (<$ 12,000, $12,001‐$20,000, $20,001‐$40,000, $40,001‐$60,000, $60,001‐$80,000, >$80,000, with the mid point of each income band used in the analysis to improve the distribution) and Age. 3 Two health variables were included in the analysis: Mental Health was measured using the mental health summary score of the SF‐12 and Physical Health used the physical health summary score of the SF‐12. Models were run separately for mental and physical health. 1 Questionnaire items, source of questions, and factor loadings for social capital variables.* Questionnaire Items Response format Source Factor Loadings SOCIAL CAPITAL INFRASTRUCTURE Networks In the past 12 months how often have you joined in the activities of a community group? (reverse coded) weekly, monthly, occasionally, never (4–1) New 0.754 On average, excluding the people you live with, how often do you socialise with other people (such as friends, neighbours, family or work colleagues)? (reverse coded) every day, most days, once or twice a week, once or twice a month, less often (4–1) New 0.754 Trust Generally speaking, people in Australia can be trusted strongly disagree, disagree, neither agree nor disagree, agree, strongly agree (1–5) Adapted from Ref. 0.593 Generally speaking, you can trust governments As above As above 0.831 Generally speaking, you can trust big business As above As above 0.802 Reciprocity By helping others you help yourself in the long run As above 25 Raw scores used SOCIAL CAPITAL RESOURCES Help If you had a serious personal crisis and you needed help and comfort how many people could you ask for help? none, 1–2, 3–4, 5+ (1–4) 0.807 If you were in financial difficulty and needed to borrow $50 or more how many people could you ask? As above As above 0.841 If you needed to go to a doctor's appointment and you required a lift to get there how many people could you ask? As above As above 0.802 Cohesion This is a close‐knit [neighbourhood/community] strongly disagree, disagree, neither agree nor disagree, agree, strongly agree (1–5) 0.722 People around here are willing to help their neighbours As above As above 0.785 People in this [neighbourhood /community] generally get along with each other As above As above 0.755 People in this [neighbourhood/community] share the same valulues As above As above 0.652 People in this [neighbourhood/community] can be trusted As above As above 0.709 In my [neighbourhood/community] there is a strong sense of coommunity As above As above 0.745 Civic Activities Have you ever picked up other peoples rubbish in a public place? yes/no 0.497 In the past 12 months did you belong to any group that took locaalaction? yes/no 0.724 In the past 12 months have you attended a protest march, meeting or rally? yes/no 0.674 Did you vote in the local government elections in autumn 2003? yes/no New 0.364 * Each item also had a ‘don't know’ and ‘refused’ option . The analysis of physical and mental health was stratified by urban and rural status, so that four models were created. Statistical methods Confirmatory Factor analysis was undertaken in AMOS and tested the theorised model outlined above using the questions outlined in Table 1 . Summary scores were calculated both within AMOS and also using principal components analysis in SPSS. As part of the overall modelling strategy those scores calculated in SPSS had a higher correlation with the outcome variables and so PCA was used to calculate summary scores for the five sub‐factors that had more than one manifest variable, namely: networks, trust, help, civic, and cohesion. Higher values of the summary scores indicate higher levels of the social capital component. Table 1 indicates the loadings for the five social capital variables. Factor loadings can range from ‐1 to +1, with larger values indicating that the item is more strongly related to the latent construct than those with smaller regression estimates. The variable of reciprocity was the raw score answer to one question (see Table 1 ). Using t‐tests, social capital scores were compared for rural/urban respondents and men and women. Structural Equation Modelling Structural equation modelling (SEM) is a collection of statistical techniques that allow the simultaneous examination of a set of relationships between one or more independent variables and one or more dependent variables. It is particularly useful when there is an interest in the interrelationships between the ‘independent’ variables. The MPLUS program was used to undertake the modelling. Four models were undertaken – one each for mental and physical health for both the rural and urban samples. Each model initially considered: • All the paths between each of the demographic variables and all social capital variables to determine if there were demographic differences in access to social capital. • The paths between the three social capital infrastructure variables and the three social capital resource variables in order to examine the ways that particular infrastructure variables may facilitate access to resource variables. • The paths between the demographic and social capital variables and the health variables to consider socio‐demographic predictors of health and the relevance of social capital to health in each setting. Non‐significant paths ( p > 0.05 and coefficient <0.1) were then removed one by one with the least significant paths removed first and the model fit evaluated after each. For variables that directly influenced the final health outcomes, all paths with a p value 0.05 were retained to consider smaller but still potentially interesting relationships. There has been substantial debate about the best measures of model fit to use to evaluate how well the model fits the observed data. Several model fit measures are reported here. We report four relative fit statistics, which compare the proposed model to the null model. The Root Mean Square Error of Approximation (RMSEA) estimates the amount of deviation between the model and the data. RMSEA values below 0.05 are indications of acceptable fit while values larger that 0.07 are indications of important amount of misfit. MPLUS does not provide the more common Goodness of Fit (GFI) value which is a measure of the relative amount of variance and covariance that is jointly explained. Alternatively, Comparative Fit Index (CFI) and TLI were used where a value of >0.9 is considered to be a well fitting model, though TLI statistics are usually lower than CFI figures. We also report the Standardarised Root Mean Square Residual (SRMR) which is the average difference between the predicted and observed variances and covariances in the model, based on standardised residuals. The smaller the SRMR the better the model fit with values of 0.05 widely thought to indicate a good fit. We also include an absolute fit statistic – the Normed Chi‐Square statistic (NCSq) (which is the chi‐square fit index divided by degrees of freedom to make it less dependent on sample size). Opinion differs on acceptable values, though Schumacker and Lomas argue that values less than 5 are adequate. We used these measures to evaluate model fit, and the final models in general met these criteria, though in some cases the TLI scores were below 0.9 (see Figures 1–4 ). 1 Urban mental health model. 2 Urban physical health model. 3 Rural mental health model. 4 Rural physical health model. In this study,‘don't know’ and refusal responses were treated as missing data. There were small numbers of missing data for each of the social capital questions (in most cases well below 3%). These missing cases were imputed five times using NORM software including the other questions for each factor in the model, and the five complete datasets were then combined (averaged) to form a complete single data set. There were 165 cases missing for income (8.2%) and these were allocated to the mean of the midpoint income values as it was a single item measure. Data were weighted to account for an individuals’ probability of selection within a household and also by age and gender in line with the Australian Bureau of Statistics 2001 Census of Population and Housing data for the South Australian population. Results Table 2 illustrates the demographic profile of the respondents and Table 3 shows the social capital and SF‐12 scores for the urban and rural samples. Figures 1 and 2 show the models for mental and physical health respectively for the urban sample and Figures 3 and 4 show the same models for the rural sample. 2 Demographic details of samples (weighted). N (Urban) N (Rural) Total = 1,487 Total: 526 Gender Male 713 (48%) 265 (51%) Female 774 (52%) 260 (49%) Country of Birth Australian 1109 (75%) 457 (87%) non‐Australian 378 (25%) 68 (13%) Marital status Married/defacto 968 (65%) 389 (74%) Other 519 (35%) 136 (26%) Education Left school < 15yrs 227 (15%) 103 (20%) Left school after 15/no additional completed qualifications 477 (32%) 203 (39%) Trade/apprenticeship/certificate/diploma 491 (33%) 167 (32%) Bachelor degree or higher 292 (20%) 53 (10%) Income Up to $12,000 105 (7%) 38 (7%) $12,001 ‐ $20,000 176 (12%) 88 (17%) $20,001 ‐ $40,000 240 (16%) 99 (19%) $40,001 ‐ $60,000 268 (18%) 116 (22%) $60,001 ‐ $80,000 229 (15%) 66 (13%) More than $80,000 337 (23%) 87 (17%) Missing 133 (9%) 32 (6%) Age Mean= 49.6 Mean=51.2 (SD=17.8) (SD=16.5) 3 Social capital factor scores and health summary scores for urban and rural samples (weighted). Urban Mean (SD) Rural Mean (SD) P value Social capital Networks 3.19 (0.97) 3.43 (0.94) <0.000 Trust 3.12 (0.73) 3.06 (0.75) 0.100 Reciprocity 4.25 (0.63) 4.20 (0.61) 0.184 Help 3.98 (0.94) 4.06 (0.90) 0.083 Civic 2.13 (0.82) 2.29 (0.81) <0.000 Cohesion 3.62 (0.64) 3.82 (0.58) <0.000 Health Physical health 49.09 (10.20) 48.73 (10.12) 0.484 Mental health 52.14 (8.92) 53.08 (8.38) 0.031 Social capital Three of the social capital scores were significantly different between rural and urban participants ( Table 3 ). Rural residents reported significantly higher levels of the social capital infrastructure variable of networks and the social capital resource variables of civic activities and cohesion, compared to urban residents. There were no statistically significant differences in ‘Trust’, ‘Reciprocity’ or ‘Help’. In the urban models ( Figures 1 and 2 * ) there were a number of demographic differences in the associations with the social capital variables. Higher trust scores were significantly associated with higher income, indicating that those on higher incomes had greater levels of trust than those on lower incomes. Women had higher levels of ‘Reciprocity’ than men, and those with higher educational achievement had higher levels of ‘Reciprocity’ than those with lower educational achievement. There were no predictors of network involvement. Men and those on higher incomes had higher levels of ‘Help’ than women and those on lower incomes, respectively. Those with higher educational achievement had higher levels of ‘Civic Activities’ than those with lower levels of educational achievement. There were no demographic predictors for ‘Cohesion’. In terms of the relationship between the social capital infrastructure and social capital resource variables those with higher ‘Reciprocity’ scores had higher levels of ‘Help’, ‘Cohesion’ and ‘Civic Activities’; those with higher ‘Networks’ scores had higher levels of ‘Help’, ‘Cohesion’ and ‘Civic Acitivites’; and those with higher ‘Trust’ had higher levels of ‘Cohesion’ In the rural models ( Figures 3 and 4 ) there were also demographic differences in the associations with social capital variables, though these patterns differed to those found in the urban models. There were no demographic predictors of ‘Trust’ or ‘Reciprocity’, and ‘Networks’ decreased with age and was greater for women.‘Help’ was higher for those who had higher incomes and who were married/defacto.‘Cohesion’ increased with income and age. Those who had higher educational achievement had higher levels of ‘Civic Activities’ than those with lower educational achievement. Those who were married/defacto had higher levels of ‘Civic Activities’ than those who were not. The relationship between social capital infrastructure and social capital resource variables were identical to the urban model, with the exception that ‘Reciprocity’ did not predict ‘Civic Activities’ in the rural sample. Health There were no significant differences between the urban and rural samples in terms of ‘Physical Health’ ( p =0.484, Table 3 ). However, rural participants reported significantly better MENTAL HEALTH, compared to urban participants ( Table 3 , p <0.005). In the urban mental health model ( Figure 1 ), GENDER was significantly associated with ‘Mental health’ with men having better mental health than women.‘Marital status’ was also significantly associated with mental health, with individuals who were married or in defacto relationships better off than those of other marital status. Two social capital infrastructure variables,‘Trust’ and ‘Networks’, were associated with ‘Mental health’, with better mental health associated with higher levels of trust and greater network involvement.‘Mental health’ was also associated with the two social capital resource variables of ‘Help’ and ‘Cohesion’, with those with greater access to help and a stronger sense of cohesion having better mental health. In the urban physical health model ( Figure 2 ), physical health was better for those on higher incomes and with higher educational achievement, and was worse with age. Physical health was also positively associated with ‘Trust’ and ‘Help’. In the rural mental health model ( Figure 3 ), men again had better mental health than women, and mental health was better for older people.‘Trust’ was again associated with mental health, as were ‘Help’ and ‘Cohesion’. In the rural physical health model ( Figure 4 ) physical health again increased with income and was worse with age, though there were no differences by education. None of the social capital variables was associated with physical health. Summary of key urban/rural similarities and differences There were different predictors of the three social capital infrastructure variables in the rural and urban models with women having higher levels of reciprocity than men in the urban model, but higher levels of networks instead in the rural model. Network involvement decreased with age only in the rural model. In terms of social capital resource variables, in both the urban and rural models higher income was associated with greater access to help, and higher education was associated with greater involvement in civic activities. The relationships between the infrastructure and the resource variables was almost identical in the rural and urban models. However, there were some rural/urban differences between the models for the social capital resource variables – men had more access to help than women in the urban model but not in the rural model, and higher income was associated with higher cohesion in the rural model but not in the urban model. In terms of health, higher levels of trust, cohesion and help were associated with better mental health in both urban and rural models, and men had better mental health than women in both models, though the relationship was stronger in the rural model. However, networks and marital status were associated with mental health only in the urban model, and mental health increased with age only in the rural model. In both the rural and urban models those on higher incomes had better physical health, and those who were older had worse physical health, and this relationship was stronger in the rural models. In the urban models, trust, education, and help were also associated with physical health, but not in the rural models. Discussion In relation to the three research questions, differences were found between the populations living in rural and urban areas in terms of the predictors and levels of social capital, and the relationship of social capital with health: 1) There were rural/urban differences in the determinants of access to social capital, though in both cases it was generally those with more socio‐economic resources who had greater access; 2) Rural participants had greater access to some, but not all elements of social capital; 3) Social capital was associated with mental health in both areas, but with physical health only in urban areas. Were there rural/urban differences in the socio‐demographic determinants of access to social capital? Income and education were important predictors of levels of social capital. This supports previous research we have undertaken and reinforces the aspects of Bourdieu's theory of social capital which sees it as one of the mechanisms by which advantage is maintained. Gender was also significantly associated with social capital variables though the findings were inconsistent. These and the other demographic differences found highlight the importance of acknowledging the diversity of life in both urban and rural areas and illustrate that access to social capital is not homogenous in either setting. While there were very few differences between the urban and rural samples in terms of the relationships between social capital infrastructure and social infrastructure variables themselves, there were a number of differences in the demographic predictors of these variables, suggesting that the factors that influence access to social capital may vary for urban and rural residents. Compared to the urban sample, in the rural sample there were fewer demographic differences in social capital infrastructure variables and more demographic differences in access to social capital resource variables. In particular in the rural sample marital status appeared more important in predicting access to social capital. Marital status has previously been linked to social capital activities such as volunteering, and social support, and in rural areas can be a particularly important social and economic distinction. Were there rural/urban differences in social capital? As anticipated, respondents from rural areas reported significantly higher levels of a number of elements of social capital, with higher levels of social networks, greater civic participation and more social cohesion. This is consistent with the Australian studies outlined earlier. While the findings relate to individuals within the areas rather than a study of the effects of place (which precludes the positing of direct effects of aspects of area of residence), some aspects of living in rural or urban areas could be used to hypothesise about the findings. For example, higher civic activities in rural areas might be understood in terms of the lesser access to formalised services. Likewise the isolation of rural areas may also help explain a greater importance of social networks, and higher levels of homogeneity in rural areas may contribute to a greater sense of cohesion. Like Beaudoin and Thorson, we found no statistically significant differences in trust or reciprocity. There were also no differences in access to informal help. That there were differences in only some elements of social capital reinforces the need to both consider the elements separately, and in concert (i.e. as part of the total package of social capital), as it adds value in considering strengths and weaknesses in terms of overall access to social infrastructure and related resources. Were there rural/urban differences in the relationship between social capital and health? Despite the well documented differences in health outcomes between rural and urban Australians, there was no significant difference between the rural and urban respondents in terms of self‐reported physical health, a similar finding to Glover et al. There were significant mental health differences, with rural participants faring better than their urban counterparts. In both the rural and urban models, in line with a significant literature, men reported better mental health. Age and marital status were relevant in the rural and urban models respectively. Better physical health was associated with older age and higher income in both the rural and urban models, with education also relevant for the urban model, supporting a significant literature on the social determinants of health. Elements of social capital were more consistently associated with mental health than physical health. This supports findings from previous research The patterns of relationships between social capital and mental health were similar for rural and urban participants. In contrast to mental health, there were rural/urban differences in the relationships between social capital and physical health, with trust and help related to physical health for urban participants, but no social capital variables significant for rural participants’ physical health. This may relate to the particularly important role of trust in heterogenous urban environments where cohesion is often lower. The ‘help’ measure used related to people available to ask for potential help rather than actual help received. It is possible that in rural areas the actual exchange of favours may be more important to health given the generally lower availability of services, but that this is not reflected in terms of who could be asked. Study strengths and limitations This was a large study that drew on a good suite of theoretically reasoned social capital variables to broadly compare populations in rural and urban areas. It had a good response rate and relatively low levels of missing data. The study also had a number of limitations. First, the study aimed to assess rural/urban differences in social capital and its relationship with health and so used cross‐sectional data in order to do this. However, it is not possible to ascertain causal relationships and it is possible that health status may impact on other variables. Likewise, the variance explained by the models was small, suggesting that there are clearly other variables that are relevant to mental and physical health. Second, for parsimony the path analysis only looked at one‐way relationships and did not consider the way that, for example, access to help may lead to increased trust. Likewise, we only examined some social capital resource variables, however the range of variables we considered was more than in most studies of SC and health. Third, as data were collected at the individual level, it is not possible to make direct claims about area‐level effects. This would require a multilevel modelling approach with a much larger sample size of areas, and would be a fruitful area of further research. In addition, we categorised urban and rural very crudely. The ARIA measure used is limited in its ability to capture the heterogeneity within each category, and the categories themselves are broad. A more nuanced consideration these categories would be a useful further extension of this research. Likewise, a lifecourse approach which recognises that people may move between rural and urban areas would also be worthy of further consideration. Social capital may also vary across the lifecourse. Fourth, identical questions were used for both rural and urban participants. However, the meaning of questions may vary between these settings. For example, one of the measures of help relating to travel to the doctor is likely to have measured something slightly different between the two settings given that in sparsely populated rural South Australia the example of a lift to the doctor could be a long trip that would inconvenience the helper significantly. It is a difficult balance between being able to directly compare populations in rural and urban areas on variables measured in the exactly the same way, and a priori using different variables themselves which makes direct comparisons more problematic. Fifth, there may be unknown selection biases in the sample. For example, Aboriginal people are likely to be under‐represented due to the telephone sampling method. There has been little research on patterns of social capital among indigenous peoples and this would be worthy of further examination. Sixth, using Principal Components Analysis to create the social capital summary scores may have led to minor variations between the Confirmatory Factor Analysis undertaken in AMOS, and the subsequent analysis in MPLUS. Seventhly, the relatively low upper band of the household income measure (which was the standard measure collected in particular South Australian Department of Health surveys at the time of data collection) may have led to more conservative estimates of the association of income with other variables. Finally, the quantitative survey can describe associations between variables which are suggestive of possible relationships. Qualitative research is required to obtain a deeper understanding of the postulated relationships in rural and urban areas. Conclusion We found that the particular shape of social capital and its relationship to health differs for respondents in rural and urban areas. While rural respondents had higher levels of some aspects of social capital our findings support the view (which builds on the theories of Bourdieu ) that socio‐economic status is an important correlate of social capital regardless of the setting. While survey data can suggest associations between variables it is limited in what it can tell us about the texture of the social and power relationships that are vital to fully understanding social capital. Consequently detailed qualitative work is required to further unpack these complex relationships, especially in sub‐groups of the population. The different relationships between social capital and health for urban and rural respondents, and the demographic differences within areas, reinforces the need to consider the setting (whether because of the different demographic makeup of an area and/or characteristics of the area itself) when considering how social capital may inform health promotion approaches. Acknowledgements This study was supported by the National Health and Medical Research Council (projects 324724 and 229913). We would like to thank Katy Osborne for her assistance in preparing the paper and the three anonymous reviewers for their helpful comments. Footnotes * For both the rural and urban analyses the results for the relationships between demographic and social capital variables were identical for the mental and physical health models.

Journal

Australian and New Zealand Journal of Public HealthWiley

Published: Feb 1, 2009

There are no references for this article.