Social Capital and Health - Findings from the Scottish Health Survey and Scottish Social Attitudes Survey

This paper explores the Scottish evidence for a link between social capital and health outcomes to inform the ongoing development of an assets-based approach to addressing health problems and inequalities.


1. For full details of the methodology for this survey, see Bromley & Given (eds.) (2010).

2. For full details of the methodology for SSA 2009, see Annex B to Ormston (2010).

3. 'Social participation' is often used interchangeably with 'social capital', since those with higher levels of participation are assumed to have greater social capital.

4. Differences in self-assessed health by both income and SIMD were found for both sexes in Bromley et al (eds.) (2010). However, note that in the analysis conducted for this report (based on the sub-sample of participants who also answered questions on social capital), while differences by income were significant at the bivariate level, income was only marginally significant (p = 0.084) once other factors were controlled for in multivariable regression analysis.

5. Regression analysis shows that economic status is statistically significant even after controlling for age. Those who were unable to work, retired, or looking after the home were all significantly more likely than those in paid employment to say that their health was fair, bad or very bad rather than good or very good. The difference between those in paid employment and those who were unemployed was not significant in the regression model, possibly because the model also controls for area deprivation and income, although at a bivariate level this difference was significant.

6. While there appears to be a relationship between feeling able to influence decisions over the local area and general health, individual differences between comparison and reference categories were not significant in the regression model, so this is not included in the table above. The bivariate relationship, however, is very similar to that between general health and feeling involved in the community, as described above. 81% of those who strongly agreed that they could influence decisions about their area felt their health was good or very good, compared with 72% of those who strongly disagreed. General social trust was not significantly related to self-assessed health in the regression model.

7. See Annex A, Model 1. Moreover, adding these variables to the model improved its ability to predict respondents' general health, albeit only very slightly (Nagelkerke R2, a measure of how well the model 'fits' and is able to predict the outcome in question, increased from 36.1% to 38.4%).

8. Adding social capital variables to a regression model of general health based on SSA data did increase the model fit (Nagelkerke R2 increases from 30.9% to 34.8%). However, agreement/disagreement with the statement 'I feel that there are people in this area I could turn to for advice and support' was the only individual social capital variable significant at the 5% level (p <= 0.05) in the regression model. Moreover, the odds ratios for this variable were not entirely linear. However, at the bivariate level, the pattern was more linear.

9. It is worth noting, however, that although significant at the bivariate level, differences by deprivation were only marginally significant (p = 0.077) in multivariable regression analysis, when controlling for other factors (see Annex A). Marital status was also marginally significant (p = 0.084), with the odds ratios suggesting that those who are single are more likely to have low wellbeing than those who are married or in a civil partnership.

10. See Annex A, Model 3. Adding various social capital variables to the regression model improved the fit (Nagelkerke R2 increased from 17.1% to 25.1%).

11. A similar pattern was apparent at a bivariate level for feeling involved in the local community - those who were not at all involved were more likely to have low mental wellbeing and less likely to have average or above wellbeing compared with those who felt any level of involvement in their local community. However, as individual differences between reference and comparison categories were not significant in the regression model (see Annex A), these figures are not included in Table 5.


Email: Linzie Liddell

Back to top