Measures of personal social capital over time: A path analysis assessing longitudinal associations among cognitive, structural, and network elements of social capital in women and men separately.

Studies on personal social capital and health have relied on several key measures of social capital - trust, participation, network capital - all with the aim of capturing the resources to which individuals or groups might have access through their social networks. As this work has evolved, researchers have sought to differentiate among key measures, often arguing that each represents a different type of social capital. Despite the importance of this work, few studies have examined (a) whether these measures are in fact distinct constructs, particularly over time, (b) if these relationships are causal, and (c) whether gender patterns the ways these measures are related. Using a probability-based sample of adults with 1-3 observations per respondent, we apply generalized structural equation modeling to assess in women and men separately whether generalized trust, trust in neighbors, network diversity, social isolation, and social participation are associated with each other, hypertension, and self-reported health over a five-year period. The initial response rate was 38.7%, with cooperation rates of 60.4% and 56.3% at waves two and three. Findings highlight stability in the longitudinal relationship of the same measure across waves. They also suggest that social capital measures operate differently for men and women, with key measures of one type of social capital more often associated with another type in women than men. Nevertheless, the strengths of the associations remain weak in women and men, particularly over time, suggesting that these measures (especially generalized trust) may be inadequate proxies for each other. Lastly, social capital seemed more salient for women's than men's health. Future research on social capital might consider more deeply the role and meaning of gender in interpreting the results of studies linking social capital to health. Further consideration of trust, participation, and network capital as distinct constructs is also warranted.