Social capital and health in Australia: An overview from the household, income and labour dynamics in Australia survey

Social Science & Medicine 70 (2010) 588–596

Contents lists available at ScienceDirect

Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed

Social capital and health in Australia: An overview from the household, income and labour dynamics in Australia surveyq Helen Louise Berry a, b, *, Jennifer A. Welsh a a b

National Centre for Epidemiology and Population Health, ANU College of Medicine, Biology and Environment, The Australian National University, ACTON ACT 0200, Canberra Centre for Research and Action in Public Health, Faculty of Health, University of Canberra, ACT 2601, Canberra

a r t i c l e i n f o

a b s t r a c t

Article history: Available online 19 November 2009

Social capital is associated with better health, but components of social capital and their associations with different types of health are rarely explored together. The aim of this study was to use nationally representative data to develop population norms of community participation and explore the relationships between structural and cognitive components of social capital with three forms of health – general health, mental health and physical functioning. Data were taken from Wave 6 (2006) of the Household, Income and Labour Dynamics in Australia Survey. Using individual-level data, the structural component of social capital (community participation) was measured using a twelve-item short-form of the Australian Community Participation Questionnaire, and the cognitive component (social cohesion) by sense of belonging, tangible support, trust and reciprocity. Three subscales of the SF-36 provided measures of health. Multiple hierarchical regression modelling was used to investigate multivariate relationships among these factors. Higher levels of participation were related to higher levels of social cohesion and to all three forms of (better) health, particularly strongly to mental health. These findings could not be accounted for by sex, age, Indigenous status, education, responsibility for dependents, paid work, living alone or poverty. Controlling for these and physical health, structural and cognitive components of social capital were each related to mental health, with support for a possible mediated relationship between the structural component and mental health. Social capital was related to three forms of health, especially to mental health. Notable gender differences in this relationship were evident, with women reporting greater community participation and social cohesion than men, yet worse mental health. Understanding the mechanisms underlying this apparent anomaly needs further exploration. Because community participation is amenable to intervention, subject to causal testing, our findings may assist in the development of programs which are effective in promoting social cohesion and, thereby, mental health. Ó 2009 Elsevier Ltd. All rights reserved.

Keywords: Australia Mental health Social capital Gender

People living in communities that are rich in social capital tend to experience more positive social (Messner, Baumer, & Rosenfeld, 2004), economic (Woodhouse, 2006) and health (Lochner, Kawachi, Brennan, & Buka, 2003) outcomes, including for mental health (De Silva, McKenzie, Harpham, & Huttly, 2005). However, the relationship between the two is not always consistent (see for example: Blakely et al., 2006; Ellaway & Macintyre, 2007; Mitchell & LaGory, 2002; Ziersch, Baum, Darmawan, Kavanagh, & Bentley, 2009). The link between social capital and health has been shown to vary according to the sample (Mansyur, Amick, Harrist, & Franzini, 2008), the type of health outcome (Ellaway & Macintyre, 2007)

q This study was funded under Australian Government Department of Families, Housing, Community Services and Indigenous Affairs Social Policy Research Grant #FACS2532, project FCH 2007/03a. * Corresponding author. þ61 2 6125 0394; fax: þ61 2 6125 0740. E-mail address: [email protected] (H.L. Berry). 0277-9536/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.socscimed.2009.10.012

and the context in which it is studied (Moore, Daniel, Gauvin, & Dube´, 2009). Some evidence suggests that social capital can be associated with worse health outcomes (Mitchell & LaGory, 2002; Ziersch & Baum, 2004). These inconsistent findings have led researchers to conclude that the relationship is ‘‘encouraging but inconclusive’’ (Falzer, 2007, p. 34). These inconsistencies may stem from poor conceptualisation and measurement of social capital. Social capital is a multidisciplinary and multifaceted concept that has suffered from conceptual confusion (Falzer, 2007) and overuse (Portes, 1998). Numerous, often fragmented definitions are used and vary according to the discipline in which it is studied, the conceptual framework employed and an unresolved debate about whether the individual or the collective is the appropriate unit of analysis (Nieminen et al., 2008). Problems with conceptualisation have translated into compromised measurement approaches. Studies often make use of ad hoc measures which were not designed to measure social capital, sometimes limited to

H.L. Berry, J.A. Welsh / Social Science & Medicine 70 (2010) 588–596

one or two questionnaire items (see for example Lindstrom, 2008; Lindstrom & Mohseni, 2009). This is problematic because social capital is a multidimensional concept which cannot be accurately measured with so few items. Studies which have aimed to develop appropriate measures of social capital often lack the conceptual foundation on which to base their measurement (Nieminen et al., 2008). The results are conceptually confusing and may contain indicators of concepts related to, but not part of, social capital, such as perceptions of safety and desirability of the neighbourhood (Onyx & Bullen, 2000). In epidemiology, social capital is most commonly defined (Farr, 2004), using Putnam’s (2000) definition, as a combination of (i) patterns of community participation and (ii) social cohesion created by participation. Participation is also known as the structural component of social capital and cohesion as the cognitive component (Almedom, 2005; Mitchell & Bossert, 2007; Whitley & McKenzie, 2005), respectively ‘‘what people do’’ and ‘‘what people feel’’ (Harpham, Grant, & Thomas, 2002). Social cohesion includes sense of belonging, social trust, generalised reciprocity and cooperation, and social harmony (Harpham et al., 2002). A key feature of this definition of social capital is the notion that structural and cognitive elements of social capital are causally related. As shown in Fig. 1: community participation is associated with greater social cohesion, which together comprise social capital (but note that this causal relationship has yet to be fully empirically tested); social capital, in turn, supports advantageous health outcomes. Social capital may be conceptualised and measured at the individual or collective level (De Silva et al., 2005). When measured at the individual level it may be termed ‘‘personal social capital’’ (Berry & Rickwood, 2000). When using this framework, the participation component is very important. This is because of the proposed causal relationship between participation and cohesion; it is participation that is responsible for creating and maintaining cohesion. It is also the component which is the most amenable to intervention. Participation can include a number of different tasks and activities. These fall into three broad categories: ‘‘informal social connectedness’’, involving contact with family, friends and neighbours (Hendry & Reid, 2000; Lee, Draper, & Lee, 2001; Putnam, 2000, p. 115); ‘‘civic engagement’’, including volunteering and joining community groups (Kuchukeeva & John O’Loughlin, 2003; Putnam, 1995; Smidt, 1999; Uslaner & Conley, 2003); and ‘‘political participation’’, including local activism and political protest (McAllister, 1998; Rich, 1999; Schudson, 1996). Fourteen discrete types of community participation have been identified, most defined by commitment, initiative and effort and by personal rather than impersonal involvement (Berry, Rodgers, & Dear, 2007). Not all of these are related to mental health. Research has also demonstrated that breadth of and perceptions about participation are also linked to mental health (Berry, 2008b). Aims of the current study Using a nationally representative dataset and controlling for a range of socio-demographic factors, the aims of this study were to

Social Capital

Community Participation

Personal Social Cohesion

Health

Fig. 1. Proposed relationship between the two components of social capital and health.

589

(i) examine patterns in frequency and enjoyment of community participation, and in social cohesion, (ii) produce population norms for community participation, (iii) examine the relationship between components of structural and cognitive social capital and three aspects of health – mental health, general health and physical functioning and (iv), controlling for physical health, assess the independent contributions of structural and cognitive components of social capital to explaining variance in mental health.

Methods The HILDA survey Data for this study were taken from Wave 6 of the Household, Income and Labour Dynamics in Australia (HILDA) Survey (Watson & Wooden, 2006, Wooden, Freidein & Watson, 2002). The HILDA Survey is a nationally representative panel survey of Australian adults aged 15 years and over in which participants report a large range of health and socioeconomic information. Data are collected annually and, in each wave, respondents complete four separate survey questionnaires. One of these is a self-complete questionnaire in which respondents report on their health and provide psychosocial information. Mostly, identical items are included in each questionnaire in every wave to facilitate longitudinal analysis. However, the self-complete questionnaire also includes a special focus supplement that varies from year to year; included in Wave 6 were items measuring personal social capital. Wave 1 included a total of 13,969 respondents. Attrition, and the addition of new entrants in waves 2–6, resulted in a 12,905 respondents in Wave 6. All waves of the HILDA Survey have included items tapping some very limited aspects of social capital (contact with family and friends, membership of clubs, volunteering, trust) and related constructs, such as social support and neighbourhood cohesion. Additional items included in the Wave 6 Self-Completion Questionnaire built on this base. Participants and procedures Data for this study were taken from the Self-Completion Questionnaire for Wave 6. Respondents were 90.7% of the total Wave 6 sample who returned their self-complete questionnaires, resulting in a final sample of N ¼ 11,709 (N ¼ 5462 men and N ¼ 6247 women). Analyses were performed using SPSS 17.0 for Windows (Statistical Package for Social Sciences, SPSS Inc.) and Stata version 9.0 (StataCorp LP). Very small amounts data were missing and these were imputed in SPSS using expectation-maximisation methods. Measures Health. Mental health was measured using the transformed mental health subscale of the SF-36, a widely used and extensively validated health screening instrument (Ware, Snow, Kosinski, & Gandek, 1993). Physical health was measured using two further subscales of the SF-36: the transformed general health and physical functioning subscales. Scores for all three scales range from 0 to 100, with higher scores indicating better health. The subscales were treated as continuous variables in all analyses. However, in order to describe the health of our sample, we examined cut-points for each of the subscales. Cut-points for diagnoses vary, but a commonly used criterion is that scores of less than 60 on the two latter subscales indicate poor physical functioning and general health and a score of less than 50 represents poor mental health (Crosier, Butterworth, & Rodgers, 2007). Approximately 28% of HILDA respondents reported poor general health and 12% reported poor physical functioning. Roughly 10% scored less than 50 on the mental health subscale, reflecting poor mental health (Table 1).

590

H.L. Berry, J.A. Welsh / Social Science & Medicine 70 (2010) 588–596

Table 1 Mean scores and standard deviations for socio-demographics and SF-36 subscales. Male

Socio-demographics Sex Age (years)a

5375 43.91

46.6 6155 53.4 11,530 100 18.14 44.31 18.45 44.12 18.31

Valid %

4772

77.5

8937

77.5

1.8

132

2.1

230

2.0

59.8

3523

57.2

6737

58.4

40.2

2632

42.8

4793

41.6

3827 1548

71.2 28.8

3651 2504

59.3 40.7

7478 4052

64.9 35.1

825 4550

15.3 84.7

996 5159

16.2 83.8

1821 9709

15.8 84.2

14.6

1274

20.7

2059

17.9

85.3

4877

79.2

9460

82.0

216

4.0

172

2.8

388

3.4

227

4.2

350

5.7

577

5.0

623 473

11.6 8.8

840 526

13.6 8.5

1463 999

12.7 8.7

1497 724 1613

27.9 13.5 30

933 937 2396

15.2 15.2 38.9

2430 1661 4009

43.1 14.4 34.8

Government income support Income support 785 recipient No income support 4583 Level of education Post grad masters or doctorate Grad diploma grad certificate Bachelor Advanced diploma/ diploma Certificate Year 12 Year 11 and below

Valid N %

77.5

Responsible for dependents Responsible for 3214 dependents Not responsible 2161 for dependents

Household type Live alone Multi-person household

Total

Valid N %

Indigenous status Aboriginal or Torres 4156 Strait Islander Non indigenous 98

Employment status In paid work Not in paid work

Female

N

SF-36 subscales M SD M SD M Physical functioning 85.01*** 22.66 82.31 23.70 Mental health 75.51*** 16.67 73.21 17.50 General health 68.94 20.73 68.81 21.61

83.56 74.28 68.87

SD 23.26 17.16 21.20

***p-Value < .001. a Age is average in years and standard deviation.

Preliminary analysis suggested that mean scores in the present wave of the HILDA study were generally consistent with the latest available Australian population norms for these subscales (Australian Bureau of Statistics, 1999). However, there were small yet statistically significant differences. A comparison of means suggests that respondents in our sample reported slightly better physical functioning, but slightly worse general health and mental health than was found in the general Australian population. This difference was minimal in the physical functioning subscale, with less than one point separating the scores for males (85.0 vs. 84.2, t(14,475) ¼ 3.03 p ¼ <.00), and for total scores (83.6 vs. 82.6, t(30,384) ¼ 2.89 p ¼ <.00), but not for females (82.3 vs. 81.5, t(15,908) ¼ 3.1.29 p ¼ >.05). All mean differences were statistically significant for mental health scores, even though scores were less than one point apart for males (75.5 vs. 77.3, t(14,475) ¼ 6.36 p ¼ <.00), females (73.2 vs. 74.6, t(15,908) ¼ 5.03 p ¼ <.00) and the total (74.3 vs. 75.9, t(30,384) ¼ 8.15 p ¼ <.00). The greatest difference was found in general health scores with the HILDA sample males (68.9 vs. 71.3, t(14,475) ¼ 6.31 p ¼ <.00), females (73.2 vs. 74.6, t(15,908) ¼ 10.29 p ¼ <.00) and total (74.3 vs. 75.9, t(30,384) ¼ 11.93 p ¼ <.00) scoring approximately three points lower than was found in the general Australian population.

Community participation. We used three approaches to measuring community participation: frequency of individual types of participation, breadth across types of participation and perceptions about participation. Frequency of participation Items measuring individual levels of community participation were derived from the Australian Community Participation Questionnaire (ACPQ), which measures fourteen distinct types of community participation (Berry et al., 2007). A twelve-item shortform of the ACPQ was developed for the HILDA Survey, to take account of items already included in every wave. Respondents indicated on a six-point scale (from 1 ‘‘never’’ to 6 ‘‘very often’’) how often they participated in each activity. Because there was a degree of overlap between the ACPQ and HILDA participation measures, items which tapped the same type of participation were combined using an unweighted mean: two items were combined to create a contact with friends and family variable, two to form a volunteering variable and two to make an item tapping involvement in community organisations and groups. Following the approach used by Berry (Berry et al., 2007), we assigned types of participation to each of the three domains of participation: informal social connectedness, civic engagement and political participation. Exploratory factor analysis using maximum likelihood factoring with oblimin rotation indicated that the items loaded on the expected domains, and we created weighted mean scores for each domain (see Box 1). Breadth of participation Following the method proposed by Berry (Berry & Shipley, 2007), who found that only seven of their fourteen types of participation were independently associated with better mental health, we computed an index to examine the relationship between breadth of participation and health. Preliminary analysis of the HILDA Survey also revealed seven types of community participation which independently predicted mental health: contact with extended family, friends and neighbours, contact with friends and family, attending organised community events, volunteering and political protest. Dichotomous scores were created for each by mean-split. A score of zero was given if the respondent scored below the mean, or one for scores equal to or greater than the mean. Scores for each type of participation were summed to form an eight-point index (possible range ¼ 0–7). Perceptions of participation One item in the HILDA Survey social support measure taps enjoyment of participation, and we used this item. Personal social cohesion. A five-item measure of social trust and two items tapping generalised reciprocity were included in the Wave 6 self-report questionnaire. The trust items were taken from two sources: the Organisational Trust Inventory (Cummins & Bromiley, 1996), as adapted for use in general populations (Berry, 2008a; Berry & Rickwood, 2000; Berry & Rodgers, 2003), and the World Values Survey (Inglehart, Miguel, Jaime, Loek & Ruud, 2004). The reciprocity items were also taken from the World Values Survey. Respondents indicated on a seven-point Likert-type response format the degree to which they agreed or disagreed with each of statements (from 1 ‘‘strongly disagree’’ to 7 ‘‘strongly agree’’). Each wave of the HILDA Survey includes a measure of social support (Henderson, Duncan-Jones, McAuley, & Ritchie, 1978; Marshall & Barnett, 1993). Consistent with the approach proposed by Cohen, Mermelstein, Kamarck, and Hoberman (1985), who made the distinction between four aspects of social support, including sense of belonging and tangible support, we created two subscales

H.L. Berry, J.A. Welsh / Social Science & Medicine 70 (2010) 588–596

Box 1. Questionnaire items. Community participation items ACPQ items 1. Have telephone, email, or mail contact with friends or relatives not living with you (ISC)a 2. Chat with your neighbours (ISC) 3. Attend events that bring people together such as fetes, shows, festivals or other community events (CE)b 4. Get involved in activities for a union, political party or group that is for or against something (PP) 5. Make time to attend services at a place of worship (CE) 6. Encourage other to get involved with a group that’s trying to make a difference in the community (PP) 7. Talk about current affairs with friends, family or neighbours (ISC) 8. Make time to keep in touch with friends (ISC) 9. Volunteer your spare time to work on boards or organising committees of clubs, community groups or other non-profit organisations (CE)c 10. See members of my extended family (or relatives no living with me) in person (ISC) 11. Get in touch with a local politician or councillor about issues that concern you (PP) 12. Give money to a charity if asked (CE) HILDA participation items 13. How often get together with friends and relatives (ISC)a 14. How many groups are you currently an active member of? (CE)b 15. Time spent volunteering each week (CE)c Personal social cohesion items Sense of belonging items 1. People don’t come to visit me as often as I would like 2. I seem to have a lot of friends 3. I don’t have anyone that I can confide in 4. I have no one to lean on in times of trouble 5. There is someone who can always cheer me up when I’m down 6. I often feel very lonely 7. When something’s on my mind, just talking with the people I know can make me feel better Tangible support items 8. I often need help from other people I can’t get 9. When I need someone to help me out, I can usually find someone Enjoyment item 10. I enjoy the time I spend with people who are important to me Trust items 1. People take advantage of others 2. People keep their word 3. People succeed by stepping on others 4. People are honest 5. People can be trusted Reciprocity items 1. People are helpful 2. People look out for themselves a

Items combined to form a contact with friends and family variable. b Items combined to form an organised community events variable. c Items combined to form a voluntary sector activity variable.

591

derived from seven sense of belonging items and two tangible support items. Exploratory factor analysis using maximum likelihood factoring with oblimin rotation indicated that each of the composite measures of personal social cohesion loaded substantially (>.5) onto a single underlying factor, providing prima facie evidence of construct validity (details available from the authors). So that direct comparisons between variables would be possible, all community participation and personal social cohesion variables were standardised to z-scores. Results of analyses using z-scores are difficult to interpret, partly because there can be both positive and negative values, and because the numbers do not relate to an intuitive scale. To assist in interpretation, all scores were returned to a common six-point scale. Socio-demographic information. Each wave of the HILDA Survey contains extensive socio-demographic information. We controlled in all analyses for: sex, age, Indigenous status, education level, responsibility for dependents, being in paid work, living alone, and poverty (operationalised as receipt of a government pension, benefit or allowance, which are strictly means-tested in Australia) (Table 1). Analytic approach We produced descriptive statistics and Pearson Product Moment and partial correlations (controlling for socio-demographic factors and all other variables) to examine, respectively, the zero-order and unique bivariate associations between community participation variables, personal social cohesion, and three subscales of health. Multiple hierarchical regression modelling was used to examine the independent relationships between components of social capital and mental health. The data were weighted to enable the production of Australian population norms (for information on weights in the HILDA Survey, see Watson, 2004). Predictor variables were added in blocks in the order suggested in Fig. 1; the model was comprehensively re-evaluated at each block so that non-significant predictors of mental health were removed, one at a time, until only significant predictors remained in the model. Changes in beta values from one step to the next were examined to assess putative mediation effects: while causality cannot be inferred from cross-sectional data, the plausibility of causal hypotheses can be examined. Because the household is the primary sampling unit of analysis in the HILDA Survey and, subsequently, there may be a degree of dependence in the observations, robust standard errors were calculated for all analyses. Results Table 2 presents Australian population norms for the community participation items adapted from the ACPQ, and for the participation items that appear in each wave of the HILDA Survey. Keeping in touch with family and friends were the most common forms of participation, while religious observance and political types of participation were the least common. One-way analyses of variance indicated that women reported higher levels of informal social connectedness and civic engagement than did men; there were no sex differences in the political participation scale. An examination of individual items revealed that, except for community activism and getting in touch with a politician, women reported engaging in all forms of community participation significantly more frequently than did men. Women reported a stronger sense of belonging and more tangible support than did men, greater trust in others, a stronger belief in generalised reciprocity and greater enjoyment of time spent with others (Table 2). Mean scores for individual items showed that women reported having more friends than did men,

592

H.L. Berry, J.A. Welsh / Social Science & Medicine 70 (2010) 588–596

Table 2 Australian population norms for community participation and means and standard deviations for social cohesion by sex. Male

Female

Total

Participation and social cohesion

M

SD

M

SD

M

SD

3.80 3.51

.80 1.30

4.12*** 4.00***

.79 1.32

3.96 3.76

.81 1.33

4.09 3.43 3.53

1.14 1.30 1.33

4.50*** 3.52*** 3.70***

1.09 1.36 1.36

4.30 3.47 3.62

1.14 1.33 1.34

4.16

.98

4.47***

.93

4.32

.97

2.53 2.04 2.44

.81 1.51 .96

2.72*** 2.30*** 2.50***

.87 1.68 .98

2.63 2.17 2.47

.85 1.60 .97

1.59 3.34

.78 1.31

1.66*** 3.72***

.82 1.27

1.62 3.53

.80 1.30

Political participation Expressing opinions publicly Community activism Political protest

1.70 1.47

.81 .84

.82 .83

1.71 1.46

.81 .84

1.97 1.64

1.21 1.00

2.07*** 1.58

1.28 .95

2.03 1.61

1.25 .97

Sense of belonging Tangible support Trust Reciprocity Perceptions about participation

4.41 4.74 4.23 3.87 5.23

.87 1.00 .91 .89 .98

4.60*** 4.92*** 4.23*** 3.88*** 5.50***

.88 1.02 .92 .91 .86

4.50 4.83 4.16 3.78 5.37

.88 1.01 .91 .88 .93

Informal social connectedness Contact with extended family Contact with friends Contact with neighbours Active interest in current affairs Contact with friends and family Civic engagement Religious observance Organised community activities Voluntary sector activity Giving money to charity

1.71 1.46

moderate, indicating a modest amount of shared variance (overlap) between types or domains of participation.

Notes: 1. ***p-value < .001. 2. All estimates are weighted to be representative of the Australian population.

Zero-order correlations between the three domains of participation, and the personal social cohesion scales were also significantly and moderately correlated, suggesting that those who reported greater informal social connectedness, civic engagement or political participation also reported somewhat greater personal social cohesion. Correlations showed that, with exception of two items, all types of community participation were modestly but significantly and positively associated with sense of belonging and tangible support, such that those who participated more frequently in their community reported greater sense of belonging and support. Community activism and getting in touch with a politician were not significantly associated with sense of belonging and tangible support, or the associations were trivial. With no exceptions, all participation items were modestly positively and significantly associated with all trust and reciprocity items: those who participated more frequently in their community reported greater trust in others and a stronger sense of reciprocity. Social capital and health Overall, men reported better health than did women (Table 1). Men reported better physical functioning and also reported better mental health, though women reported feeling more calm and peaceful. There were no sex differences in general health. Community participation and health

yet also reported greater feelings of loneliness. Men were more likely than were women to report that they had no-one to lean on or confide in. Participators and non-participators Zero-order correlations demonstrated that the three domains of participation were modestly and significantly positively associated, such that that those who engaged in one domain of community participation were more likely to report engaging in other domains (Table 3). Correlations between individual participation items showed the same trend: those who reported engaging in any type of participation were more likely to report engaging in other types (details available from the authors). Correlations were small to

The three domains of participation and the three health subscales were mostly statistically significantly related, indicating that engagement in most types of participation was related to health (Table 3). Informal social connectedness was more strongly associated with all forms of health than were civic engagement or political participation. Some individual types of participation were more strongly related to health than were others, while some types were related to better health and some to worse. However, some of these correlations were of negligible magnitude (<.10), explaining less than 1% of variance in the mental health subscale. The pattern was not quite the same with respect to the associations between the participation items and the physical health subscales. While most of the correlations were small, some of the correlations were negative,

Table 3 Bivariate associations between domains and bread of participation, perceptions about participation, personal social cohesion and health. 2

Participation domains 1. Informal social connectedness 2. Civic engagement 3. Political participation 4. Breath of participation

.41***

Perceptions of participation 5. Enjoyment Personal social cohesion 6. Sense of belonging 7. Tangible support 8. Trust 9. Reciprocity ***p-Value < .001, **p-value < .01.

3

4

5

6

7

8

9

Mental health

General health

Physical functioning

z-Order

Partial

z-Order

Partial

z-Order

Partial

.30***

.79***

.29***

.47***

.37***

.34***

.30***

.28***

.29***

.20***

.20***

.09***

.11***

.57***

.60*** .52***

.12*** .00 .20***

.21*** .09*** .38***

.16*** .04*** .28***

.24*** .11*** .34***

.22*** .12*** .27***

.18*** .08*** .25***

.16*** .07*** .26***

.09*** .02* .18***

.13*** .04*** .18***

.01 .03** .09***

.08*** .00 .11***

.40***

.39***

.23***

.18***

.20***

.21***

.17***

.16***

.09***

.09***

.68***

.42*** .41***

.34*** .34*** .65***

.46*** .42*** .36*** .28***

.47*** .42*** .35*** .27***

.34*** .32*** .26*** .20***

.31*** .32*** .29*** .22***

.18*** .17*** .05*** .03**

.14*** .17*** .13*** .09***

H.L. Berry, J.A. Welsh / Social Science & Medicine 70 (2010) 588–596

indicating that greater engagement in some types of participation were associated with slightly worse, not better, physical health. Social cohesion and health In all cases, a greater sense of belonging, higher levels of tangible support, greater trust in others and a stronger sense of reciprocity were related to better health, especially to better mental and general health. Associations between these scales and physical functioning were weak. Partial correlations between the community participation and cohesion scales indicated that, controlling for a wide range of socio-demographic characteristics and for all other types of participation and cohesion, all types of participation remained related to the three aspects of health. Social capital and mental health Preliminary one-way analysis of variance found that that (i) those who reported poor (compared to good) physical health also reported poor mental health and (ii) those with worse general health and physical functioning reported lower levels of participation. This suggested support for the hypothesis that poor physical health may be a barrier to community participation, so we controlled for physical health in all multiple hierarchical regression analyses (Table 4). Items were added in six blocks: socio-demographic characteristics (age, sex, living alone, Indigenous status, labour force status, responsibility for dependents and poverty) were added in block one and general health and physical functioning in block two. In block three, weighted means for the three domains of participation (informal social connectedness, civic engagement and political participation) were added followed by the index of breadth of participation in block four. Perceptions about participation were added in block five, and personal social cohesion measures (sense of belonging, tangible support, trust and reciprocity) in the final block. Age, sex, poverty, living alone, labour force status and level of education each contributed independently to explaining 5% variance in mental health scores. With the addition of health measures in block two, education and labour force status no longer made an independent contribution, suggesting a mediating effect whereby higher levels of participation accounted for the associations between level of education and paid employment, and mental health. Together, socio-demographic and health factors accounted for 32% of the variance in mental health. In block three, the addition of the three domains of participation contributed a further 3% to explaining variance in mental health. All predictor variables remained significant. In block four, the addition of breath of participation did not increase the total of variance explained, but all variables remained significant, including breadth of participation. There was evidence of a partial mediation effect in this model: standardised beta values for informal social connectedness and civic engagement, though still significant, fell by 17% and 25% respectively, suggesting that the relationship between these two forms of participation and mental health may be partly accounted for by participating in a broad range of community activities. When perceptions of participation (enjoyment) were added in block five, total explained variance in mental health rose by 1% to a total of 36%. There was no evidence of mediation at this step. In block six, three aspects of social cohesion made a significant additional independent contribution to explaining variance in mental health. Reciprocity did not attain significance, suggesting that its relationship with mental health was accounted for by the other factors in the model. In the final model, in order of magnitude, better mental health (higher) scores were predicted by: better general health, greater sense of belonging, increasing age, being male, greater tangible support, greater social trust, better physical

593

Table 4 Multiple hierarchical regression estimates for the prediction of variance in mental health scores by socio-demographic factors, participation, perception of participation and personal social cohesion. B

Std err B

b

Model 1: Socio-demographic factors Age Sex Poverty Live alone Not in paid work Education

.10 1.56 6.17 1.41 3.38 .25

.01 .39 .56 .51 .49 1.91

.12*** .05*** .14*** .03** .09*** .03*

Model 2: Health scores Age Sex Poverty Live alone General health Physical functioning

.20 2.10 2.20 .90 .43 .04

.01 .32 .47 .43 .01 .01

.21*** .06*** .05** .02* .53*** .05***

Model 3: Domains of participation Age Sex Poverty Live alone General health Physical functioning Informal social connectedness Civic engagement Political participation

.18 3.50 1.76 1.15 .40 .03 13.16 3.47 1.10

.01 .32 .32 .42 .01 .01 .83 .90 1.70

.19*** .10*** .04*** .02** .50*** .04** .18*** .04*** .02*

Model 4: Breadth of participation Age Sex Poverty Live alone General health Physical functioning Informal social connectedness Civic engagement Political participation Breadth of participation

.18 3.51 1.77 1.16 .40 .03 11.00 2.65 1.60 .43

.01 .32 .46 .42 .01 .01 1.22 .98 .55 .16

.19*** .10*** .03*** .02** .49*** .04** .15*** .03** .03** .04**

Model 5: Perceptions of participation Age Sex Poverty Live alone General health Physical functioning Informal social connectedness Civic engagement Political participation Breadth of participation Perceptions of participation

.18 3.75 1.75 1.05 .39 .03 9.12 2.05 1.16 .48 1.61

.01 .32 .45 .42 .01 .01 1.25 .97 .56 .16 .22

.19*** .11*** .04*** .02* .48*** .04** .13*** .03* .02* .05** .08***

Model 6: Personal social cohesion Age Sex Poverty General health Physical functioning Breadth Sense of belonging Tangible support Trust

.17 3.77 1.27 .32 .03 .47 4.61 1.45 1.45

.01 .30 .43 .01 .01 .08 .26 .23 .20

.18*** .11*** .03** .40*** .05*** .05*** .23*** .09*** .08***

R2 .05***

.32***

.35***

.35***

.36***

.43***

Notes: 1. ***p-value < .001, **p-value < .01, *p-value < .05. 2. All estimates are weighted to be representative of the Australian population. 3. All estimates were calculated using robust standard errors.

functioning, greater breadth of participation and absence of poverty. The final model accounted for 43% of the variance in mental health scores. Living alone did not retain significance in the final model, suggesting a mediating effect in which higher levels of cohesion accounted for the relationship between living alone and poorer mental health. Informal social connectedness, civic engagement

594

H.L. Berry, J.A. Welsh / Social Science & Medicine 70 (2010) 588–596

and political participation, as well as perceptions about participation, also failed to contribute independently in to explaining variance in mental health in the final model, indicating the plausibility of the hypothesis that their relationship with mental health may be mediated by personal social cohesion. Discussion The aims of this study were to describe community participation in Australia and its associations with social cohesion, to test the relationship between social capital and three forms of health, and to examine the independent contributions of structural and cognitive components of mental health controlling for physical health. Consistent with previous research (Baum et al., 2000; Berry, 2008a), there was strong evidence of sex differences, with women reporting consistently higher levels of community participation and personal social cohesion than men. Also consistent with Baum et al. (2000), social capital was more strongly related to mental health than it was to physical health.

Social capital and health in Australia As expected (see Fig. 1), social capital was related to better general physical and mental health and these relationships could not be accounted for by sex, age, Indigenous status, education level, responsibility for dependents, being in paid work, living alone, poverty, or by overlap among the participation and cohesion measures. Additionally, the relationship between social capital and mental health could not be accounted for by physical health: though respondents with poorer physical health had lower levels of participation and poorer mental health, they reported better mental health when they also reported higher levels of social capital. The relationship between each component of social capital – community participation and social cohesion – suggested that people with higher levels of informal social connectedness, civic engagement and political participation were also more likely to have greater sense of belonging and tangible support, were more likely to enjoy spending time with others, and reported greater social trust and generalised reciprocity. This convergent evidence suggests construct validity (Kaplan & Saccuzzo, 2005) in the social capital measures used in the present study. In addition, and also consistent with previous research (Berry, 2008a; Kawachi, Subramanian, & Kim, 2008), our findings demonstrated a positive relationship between the structural and cognitive components of social capital and overall health, such that higher levels of both components of social capital were associated with each other and with better health. Notably, breadth of participation remained significant in the final model, suggesting that participating broadly is associated with better mental health over and above its association with social cohesion.

Community participation and poor mental health There is evidence to suggest that some forms of community participation may be related to worse mental health. Consistent with Berry et al. (2007), we found that greater political participation was associated with worse mental health, even when controlling for socio-demographic factors and physical health. Very recent research suggests that there may be a cultural explanation for this, with political participation linked to worse mental health among Indigenous Australians in a coastal community, but not among other Australians (Berry, 2009). This is clearly an important area for further investigation.

Social capital, mental health and gender Sex differences in social capital were notable. Women reported higher levels of community participation and social cohesion: consistent with previous research (Baum et al., 2000; Berry, 2008a), overall, Australian women report higher levels of both components of social capital. Although social capital is considered a protective factor for mental health, in line with Australian norms (Begg et al., 2003), women reported poorer mental health than did men. This suggests that the relationship between social capital and mental health may be gendered. It has been proposed that analysis of women as a single group may mask ‘considerabl[e] and systematic’ variation in women’s reported levels of social capital (Berry, 2008a, p. 534), and that socioeconomic disadvantage may underpin this variability (Berry, 2008a; Caughy, O’Campo, & Muntaner, 2003; Kawachi & Berkman, 2001). Social capital confers obligations as well as privileges. For some, this may involve unmanageable demands (Ferlander, 2007), converting social capital to a liability rather than an opportunity (Almedom, 2005). This may be particularly true for women with disadvantaged socioeconomic characteristics (Kawachi & Berkman, 2001), who may not have the resources to deal with disproportionate strain (Baum et al., 2000; Berry, 2008a; Osborne, Ziersch, & Baum, 2008). It has also been hypothesised that women might not be as able as men to convert their social capital into benefits (Osborne et al., 2008). This theory is consistent with previous research that women’s networks are smaller, more homogenous and more reflective of informal bonding ties, which are limited in terms of potential resources (Harell, 2009). Men, conversely, enjoy larger and more heterogeneous social networks, characterised by bridging, weak ties, which have greater potential to provide numerous and diverse resources. It is thus possible that women invest more in the generation of social capital while deriving fewer of its benefits. This possibility implies significant questions as to how research is to incorporate gender in its conceptualisation of social capital. Limitations of our study We note four principal limitations to this study. (i) There may be a problem with the representativeness of our sample with respect to health, as respondents reported slightly worse mental health than exists in the general Australian population. This may reflect true differences between our sample and the underlying population. More likely, it reflects changes in mental health status since national data were collected in 1995, as rates of mental health problems in Australia have been rising steadily (Begg et al., 2003). (ii) A common problem in studies on social capital and health is that they employ cross-sectional designs which do not permit an analysis of causal relationships among concepts. We, too, are unable to comment on causality. However, while we did aim to consider the plausibility of causal relationships, causal inference was not the goal of the present study; instead, it was to develop population norms for community participation and to explore associations among components of social capital and three aspects of health in a nationally representative Australian sample; for this, cross-sectional analysis is appropriate. (iii) This study used selfreport data which were not validated against objective measures. Nevertheless, there is consistent evidence to suggest perceptions are highly consistent with objective measures (eg., Idler & Benyamini, 1997). (iv) Our study used individual-level data and we were unable to comment on the relationship between individual and community-level factors. While it is appropriate to relate personal social capital to individual health, such an approach does not account for the possible contribution of ambient community levels of social capital. More detailed studies using multilevel analysis are

H.L. Berry, J.A. Welsh / Social Science & Medicine 70 (2010) 588–596

needed (Kim & Kawachi, 2007; Kim, Subramanian, & Kawachi, 2006; Subramanian, Lochner, & Kawachi, 2003) to describe relationships among variables that pertain to, or across, different levels of measurement. Despite these limitations, our findings may have practical application. Interventions to increase community participation may be relatively cheaper and more acceptable than delivering increasing levels of psychiatric services. Understanding patterns of participation, as well as people’s thoughts and feelings about participation, would be useful in trying to enhance the take-up of services. Certainly, those who are active participators are more likely to have higher levels of social cohesion and, in turn, have the potential to enjoy wide-ranging benefits, especially in regard to their mental health. However, research must address the disproportionate burden placed on those who are most vulnerable. Conclusion This study examined the relationship between the structural and cognitive components of social capital and their shared relationship with three forms of health in a large, nationally representative Australian sample. Controlling for socio-demographic factors, our findings demonstrated that both structural and cognitive aspects of social capital are strongly linked to each other and to all three forms of health. The relationship between social capital and mental health was the strongest, suggesting social capital may play a particularly important role in this aspect of health. There were notable gender differences, with women reporting higher overall levels of community participation and social cohesion than men, but worse mental health. Understanding the mechanisms underlying this apparent anomaly needs further exploration. Of particular importance is the role of participation, as it is particularly amenable to policy and service intervention. With increased understanding, we may continue to develop and improve relatively inexpensive and socially acceptable interventions which are effective in improving mental health. References Almedom, A. M. (2005). Social capital and mental health: an interdisciplinary review of primary evidence. Social Science & Medicine, 61(5), 943–964. Australian Bureau of Statistics. (1999). Agriculture Australia 1997–98. Canberra. ABS. Baum, F. E., Bush, R. A., Modra, C. C., Murray, C. J., Cox, E. M., Alexander, K. M., et al. (2000). Epidemiology of participation: an Australian community study. Journal of Epidemiology and Community Health, 54(6), 414–423. Begg, S., Vos, T., Barker, B., Stevenson, C., Stanley, L., & Lopez, A. (2003). The burden of disease and injury in Australia 2003. Canberra: Australian Institute of Health and Welfare. Berry, H. L. (2008a). Social capital elite, excluded participators, busy working parents and aging, participating less: types of community participators and their mental health. Social Psychiatry and Psychiatric Epidemiology, 43(7), 527–537. Berry, H. L. (2008b). Subjective perceptions about sufficiency and enjoyment of community participation and associations with mental health. Australasian Epidemiologist, 15(3), 4–9. Berry, H. L. (2009). Social capital and mental health among Indigenous Australians, New Australians and other Australians living in a coastal region. Australian eJournal for the Advancement of Mental Health, 8(2). Berry, H. L., & Rickwood, D. J. (2000). Measuring social capital at the individual level: personal social capital, values and psychological distress. International Journal of Mental Health Promotion, 2(3), 35–44. Berry, H. L., & Rodgers, B. (2003). Trust and distress in three generations of rural Australians. Australasian Psychiatry, 11(S), S131–137. Berry, H. L., Rodgers, B., & Dear, K. B. G. (2007). Preliminary development and validation of an Australian community participation questionnaire: types of participation and associations with distress in a coastal community. Social Science & Medicine, 64(8), 1719–1737. Berry, H. L., & Shipley, M. (2007). Longing to belong: Personal social capital and psychological distress in a coastal Australian region. Canberra: The Australian National University. Blakely, T., Atkinson, J., Ivory, V., Collings, S., Wilton, J., & Howden-Chapman, P. (2006). No association of neighbourhood volunteerism with mortality in New

595

Zealand: a national multilevel cohort study. International Journal of Epidemiology, 35(4), 981–989. Caughy, M., O’Campo, P., & Muntaner, C. (2003). When being alone might be better: neighborhood poverty, social capital, and child mental health. Social Science & Medicine, 57, 227–237. Cohen, S., Mermelstein, R., Kamarck, T., & Hoberman, H. M. (1985). Interpersonal support evaluation list. In G. S. Sarason, & B. R. Sarason (Eds.), Social support: Theory, research and application. Washington DC: Martinus Nijhoff. Crosier, T., Butterworth, P., & Rodgers, B. (2007). Mental health problems among single and partnered mothers. Social Psychiatry and Psychiatric Epidemiology, 42(1), 6–13. Cummins, L. L., & Bromiley, P. (1996). The Organizational Trust Inventory (OTI): development and validation. In R. M. Kramer, & T. R. Tyler (Eds.), Trust in organizations: Frontiers of theory and research (pp. 302–330). California: Sage Publications. De Silva, M. J., McKenzie, K., Harpham, T., & Huttly, S. R. A. (2005). Social capital and mental illness: a systematic review. Journal of Epidemiology and Community Health, 59(8), 619–627. Ellaway, A., & Macintyre, S. (2007). Is social participation associated with cardiovascular disease risk factors? Social Science & Medicine, 64(7), 1384–1391. Falzer, P. (2007). Developing and using social capital in public mental health. The Mental Health Review, 12(3), 34–42. Farr, J. (2004). Social capital: a conceptual history. Political Theory, 32(1), 6–33. Ferlander, S. (2007). The importance of different forms of social capital for health. Acta Sociologica, 50(2), 115–128. Harell, A. (2009). Equal participation but separate paths?: women’s social capital and turnout. Journal of Women, Politics and Policy, 30(1), 1–22. Harpham, T., Grant, E., & Thomas, E. (2002). Measuring social capital within health surveys: key issues. Health Policy and Planning, 17(1), 106–111. Henderson, S., Duncan-Jones, P., McAuley, H., & Ritchie, K. (1978). The patient’s primary group. British Journal of Psychiatry, 132, 74–86. Hendry, L. B., & Reid, M. (2000). Social relationships and health: the meaning of social ‘‘connectedness’’ and how it relates to health concerns for rural Scottish adolescents. Journal of Adolescence, 23(6), 705–719. Idler, E. L., & Benyamini, Y. (1997). Self-rated health and mortality: a review of twenty-seven community studies. Journal of Health and Social Behavior, 38(1), 21–37. Inglehart, R. F., Miguel, B., Jaime, D.-M., Loek, H., & Ruud, L. (2004). Human Beliefs and Values: a Cross-cultural Sourcebook based on the 1999–2002 Value Surveys. Mexico City: Siglo XXI. 2004. Kaplan, R. M., & Saccuzzo, D. M. (2005). Psychological testing: Principles, applications, and issues. Belmont: Thompson Wadsworth. Kawachi, I., & Berkman, L. (2001). Social ties and mental health. Journal of Urban Health – Bulletin of the New York Academy of Medicine, 78, 458–467. Kawachi, I., Subramanian, S. V., & Kim, D. (2008). Social capital and health: a decade of progress and beyond. In I. Kawachi, S. V. Subramanian, & D. Kim (Eds.), Social capital and health. New York: Springer. Kim, D., & Kawachi, I. (2007). U.S state-level social capital and health related quality of life: multilevel evidence of main, mediating, and modifying effects. Annals of Epidemiology, 17(4), 258–269. Kim, D., Subramanian, S. V., & Kawachi, I. (2006). Bonding versus bridging social capital and their associations with self rated health: a multilevel analysis of 40 US communities. Journal of Epidemiology and Community Health, 60, 116–122. Kuchukeeva, A., & John O’Loughlin, J. (2003). Civic engagement and democratic consolidation in Kyrgyzstan. Eurasian Geography and Economics, 44(8), 557–587. Lee, R. M., Draper, M., & Lee, S. (2001). Social connectedness, dysfunctional interpersonal behaviors, and psychological distress: testing a mediator model. Journal of Counseling Psychology, 48(3), 310–318. Lindstrom, M. (2008). Social capital, anticipated ethnic discrimination and selfreported psychological health: a population-based study. Social Science & Medicine, 66(1), 1–13. Lindstrom, M., & Mohseni, M. (2009). Social capital, political trust and self-reported psychological health: a population-based study. Social Science & Medicine, 68(3), 436–443. Lochner, K. A., Kawachi, I., Brennan, R. T., & Buka, S. L. (2003). Social capital and neighborhood mortality rates in Chicago. Social Science & Medicine, 56(8), 1797– 1805. Mansyur, C., Amick, B. C., Harrist, R. B., & Franzini, L. (2008). Social capital, income inequality, and self-rated health in 45 countries. Social Science & Medicine, 66(1), 43–56. Marshall, M. L., & Barnett, R. C. (1993). Working family strains and gains among two-earner couples. Journal of Community Psychology, 35(3), 447–479. McAllister, I. (1998). Civic education and political knowledge in Australia. Australian Journal of Political Science, 33(1), 7–23. Messner, S. F., Baumer, E. P., & Rosenfeld, R. (2004). Dimensions of social capital and rates of criminal homicide. American Sociological Review, 69(6), 882–903. Mitchell, A. D., & Bossert, T. J. (2007). Measuring dimensions of social capital: evidence from surveys in poor communities in Nicaragua. Social Science & Medicine, 64(1), 50–63. Mitchell, C. U., & LaGory, M. (2002). Social capital and mental distress in an impoverished community. City and Community, 1(2), 195–215. Moore, S., Daniel, M., Gauvin, L., & Dube´, L. (2009). Not all social capital is good capital. Health and Place, 15(4), 1071–1077.

596

H.L. Berry, J.A. Welsh / Social Science & Medicine 70 (2010) 588–596

Nieminen, T., Martelin, T., Koskinen, S., Simpura, J., Alanen, E., Ha¨rka¨nen, T., et al. (2008). Measurement and socio-demographic variation of social capital in a large population-based survey. Social Indicators Research, 85(3), 405–423. Onyx, J., & Bullen, P. (2000). Measuring social capital in five communities. Journal of Applied Behavioral Science, 36(1), 23–42. Osborne, K., Ziersch, A., & Baum, F. (2008). Who participates? Socioeconomic factors associated with women’s participation in voluntary groups. Australian Journal of Social Issues, 43(1), 103–122. Portes, A. (1998). Social capital: its origins and applications in modern sociology. Annual Review of Sociology, 24, 1–24. Putnam, R. D. (1995). Bowling alone: America’s declining social capital. Journal of Democracy, 6, 65–78. Putnam, R. D. (2000). Bowling alone: The collapse and revival of American community. New York: Simon & Schuster. Rich, P. (1999). American voluntarism, social capital, and political culture. The Annals, 565(1), 15–34. Schudson, M. (1996). What if civic life didn’t die? The American Prospect, 25, 17–20. Smidt, C. (1999). Religion and civic engagement: a comparative analysis. The Annals, 565(1), 176–192. Subramanian, S. V., Lochner, K. A., & Kawachi, I. (2003). Neighborhood differences in social capital: a compositional artifact or a contextual construct? Health and Place, 9(1), 33–44.

Uslaner, E. M., & Conley, R. S. (2003). Civic engagement and particularized trust: the ties that bind people to their ethnic communities. American Politics Research, 31(4), 331–360. Ware, J. E., Snow, K. K., Kosinski, M. A., & Gandek, M. S. (1993). SF-36 health survey manual and interpretation guide. Boston: The Health Institute, New England Medical Centre. Watson, N. (2004). Wave 2 weighting. HILDA Project Discussion Paper Series. The University of Melbourne. Watson, N., & Wooden, M. (2006). Modelling longitudinal survey response: The experience of the HILDA survey. HILDA Project Discussion Paper Series. The University of Melbourne. Whitley, R., & McKenzie, K. (2005). Social capital and psychiatry: review of the literature. Harvard Review of Psychiatry, 13(2), 71–84. Wooden, M., Freidin, S., & Watson, N. (2002). The household, income and labour dynamics in Australia (HILDA) survey: wave 1. The Australian Economic Review, 35(3), 339–348. Woodhouse, A. (2006). Social capital and economic development in regional Australia: a case study. Journal of Rural Studies, 22(1), 83–94. Ziersch, A. M., & Baum, F. E. (2004). Involvement in civil society groups: is it good for your health? Journal of Epidemiology and Community Health, 58(6), 493–500. Ziersch, A. M., Baum, F. I., Darmawan, G. N., Kavanagh, A. M., & Bentley, R. J. (2009). Social capital and health in rural and urban communities in South Australia. Australian and New Zealand Journal of Public Health, 33(1), 7–16.