Is generalized trust useful as the indicator of social capital for small area analysis in Japan?

Several studies have used generalized trust as an indicator of social capital. However, its validity in Japan due to cultural differences is questioned. Here, we emphasize the need for more sophisticated discussions underlying trust measurement for social capital, specifically generalized and particularized trust. The aim of this study is to investigate which questions are appropriate as a indicator of social capital of small area analysis in Japan.We conducted a nationally representative survey based on the geodemographic segmentation system, which classifies households by allocating them to one of 212 segments. Each neighborhood was randomly selected from within each segment. A postal questionnaire was sent out in 2008 to household heads and their spouses in these neighborhoods. A total of 8,221 individuals responded to the survey. Generalized and particularized trust was measured by a single item. These questions were rated on a 10-point scale. We finally conducted a multilevel analysis on 6,863 individuals nested within 202 neighborhoods.The null model with no predictors revealed a significant variation in the generalized trust between neighborhoods. However, this variation was insignificant after adjusting for potential confounders. The second multilevel analysis showed variance in the neighborhood trust between neighborhoods. The null model revealed a significant variation in this trust. This variation remained after adjusting for potential confounders. Neighborhood trust could be seen as a truly contextual factor; generalized trust might be confounded by compositional factors. This indicates that neighborhood differences in generalized trust may arise from differences in personal characteristics. We need more nuanced attentions toward measuring social capital in different cultural contexts.