IJE Advance Access originally published online on May 9, 2006
International Journal of Epidemiology 2006 35(3):614-615; doi:10.1093/ije/dyl084
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Commentary |
Commentary: Bonding, bridging, and linking—but still not much going on
1 Department of Public Health, Wellington School of Medicine and Health Sciences, University of Otago, Wellington, New Zealand
2 Department of Public Health, Wellington School of Medicine and Health Sciences, University of Otago, Wellington, New Zealand
* Corresponding author. PO Box 7343, Wellington, New Zealand.E-mail: tony.blakely{at}otago.ac.nz
Much has been said about social capital and its association (if any) with health. And a common plea is to return to the theoretical underpinning of social capital, and why and how it might be associated with various outcomes including health. Szretzer and Woolcock have proposed a possible step forward in the conceptualization of social capital, arguing that there may be ‘linking’ forms of social capital in addition to previously recognized ‘bonding’ and ‘bridging’ forms of social capital.1 Bonding social capital refers to trusting and co-operative relations between members of a network who are similar in a socio-demographic sense, and bridging social capital comprises relations of respect and mutuality between people who are dissimilar. Linking social capital is ‘norms of respect and networks of trusting relationships between people who are interacting across explicit, formal, or institutionalized power or authority gradients in society’.1 For example, citizens' interactions with local government and health planning authorities are representative of linking social capital.
Why might such linking social capital be associated with health? Vertical ties between citizens and institutions of power might not only make us feel we can do something about arising issues and be part of civil society but also might increase our chances of securing health-promotion resources (e.g. parks, clinics) in our neighbourhoods.
In this issue Kavanagh et al.2 (2006) test the possible association of linking, and other, social capital with self-rated health in Tasmania. Their headline finding was that of no prominent associations of social capital measures with health, with possible exceptions of social trust (bonding) and political participation (a measure they conceptualize as a resource). They report their findings as beta coefficients. Below, we have translated this output into odds ratios that capture an effect size, so we can assess the magnitude of effect as well as its statistical significance. To do this, we calculated the odds ratio for a 2 SD change in the social capital measure using the means and standard deviations given in their Table 1. Another way of thinking of this effect size is the odds ratio comparing the 16th with the 84th percentile of the (assumed) normal distribution of areas by each area-level measure.
|
The first point to note is that the strength of the association of area-level socioeconomic status (SES) with health is less than that often observed. One reason for this is that model 3 adjusts for personal SES; the beta coefficient in Table 3 of Kavanagh et al. for model 2 (unadjusted for personal SES) is over twice as high. Another possible reason is that the areas being compared are often large (average population 11 200); a comparison of smaller neighbourhoods, with consequently greater socioeconomic differences between neighbourhoods, would probably increase the strength of this association. The second point to note is that, generally speaking, none of the social capital associations with health are particularly strong. Nevertheless, the direction of the association of four out of the five non-statistically significant measures is as one would expect, the exception being increasing linking social capital that is associated with a 4% increase [95% confidence interval (95% CI) –8 to 19%] in the odds of fair/poor self-rated health.
Third, the only social capital measure with a 95% CI excluding the null is that for increasing social trust, which (as expected) is protective against fair/poor self-rated health. What is peculiar, though, is that social trust as an exclusive measure of bonding social capital would be expected to be associated with health at smaller levels of aggregation or neighbourhoods; it is difficult to see a lot of bonding going on between an average of 11 200 people in each area. But it is easier to conceive of this scale of area being appropriate for the level at which bridging and linking social capital occur, yet they are non-statistically significant.
Finally, when the authors actually adjusted the social capital measures for area-SES, it ‘attenuated the effect of social trust and political participation so that they became statistically non-significant while the effects of area-SES remained statistically significant’.
Where to next? Well, if like us, you thought the idea of linking social capital proposed by Szretzer and Woolcock was an interesting one to explore, then more studies are required. Kavanagh et al. make a great start to the testing of this idea. However, whilst being a multi-level study, only 41 areas are compared in this study and they are large areas. One cannot help but wonder if the study is underpowered and examining an area-scale that is too large.
 |
References |
|---|
 Top References |
|---|
1 Szreter S, Woolcock M. Health by association? Social capital, social theory, and the political economy of public health. Int J Epidemiol 2004;33:650–67.
2 Kavanagh A, Turrell G, Subramanian SV. Does area-based social capital matter for the health of Australians? A multilevel analysis of self-rated health. Int J Epidemiol 2006;35:607–13.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||