Kawachi 1997 - "Social capital, income inequality, and mortality"

Kawachi, Kennedy, Lochner, Prothrow-Stith
"Social capital, income inequality, and mortality"
American Journal of Public Health
September 1997; v87, n9, pp 1491
On the Web
Relevance: High

This is perhaps the single most compelling study that social capital--in the broad sense of civic engagement--affects health. The authors use the General Social Survey's result to perform "ecologic" analyses of 39 states' levels of social capital and income inequality compared to mortality. Still, there are some important short-comings.

The authors claim that "income inequality leads to increased mortality via disinvestment in social capital" though it is not clear to me that they actually demonstrate this. Instead, they demonstrate that income inequality is correlated to social capital and also show a correlation between social capital and mortality. (It's relatively well-documented that income inequality is associated with higher mortality).

They use three principles measures of social capital. The results are:

Social Trust (measured by responses to "Do you think most people would try to take advantage of you if they got the chance, or would they try to be fair?") There's a strong correlation between states with high levels of mistrust and high levels of all-cause mortality (r=0.77). In fact, "variations in level of social trust explained 58% of the variance in total mortality."

Other Social Capital Measures (measured by agreement to "You can't be too careful in dealing with people" and "People mostly look out for themselves.") The results here were nearly identical to the results from social trust (r=0.79).

Group membership (measured by the per capita number of groups and associations to which residents in each state belonged). In some ways, this is the most relevant question to our study as it is these kinds of civic associations that are putatively affected by sprawl. There is a strong correlation between group membership and social capital (r=-0.49), but it is weaker than the correlations of the other two measures. This finding is not adjusted for poverty, but the authors claim that the correlation remains statistically significant.