Constructing the Social Activity Factors
To construct the two sociability indices, I employ the individual-level data on how frequently within the last 12 months an individual has participated in one of 20 different activities. These activities include: bowling, camping, playing cards, attending church or other place of worship, going to a classical concert, going to a pop or rock concert, going clothes shopping, going to a club meeting, working on a community project, giving or attending a dinner party, going out to dinner at a restaurant, entertaining people in ones own home, sending a greeting card, attending a lecture, going to the movies, going on a picnic, attending a sporting event, going swimming, playing tennis, and visiting an art gallery or museum. Following Putnam’s (2000) findings and theoretical notion of social capital, one would expect that they all loaded onto a single component in the principal components analysis used here to construct the indices. Yet, as Table 6 shows, the data form two distinct dimensions, which clearly adds important nuance to Putnam’s concept. Five principal components have eigenvalues above one, but a standard scree plot shows a clear elbow at two components, which is the solution chosen in this paper.
With respect to the interpretation of the components, the picture may at first seem confusing, as is often the case when using principal components analysis. However, the variables loading onto the first component seem to share the characteristic that the activities are to some extent formal and organized. The remaining variables loading onto the second component, on the other hand, capture social activities that are less formally organized and that are not necessarily planned in advance, e.g. going swimming or shopping. This interpretation is further sustained by noticing that the only variable loading onto both components—giving or attending dinner parties—can reflect both formal and informal social contact.
Happiness Determinants at the Individual Level
While the DDB Needham (2007) data do not form a panel at the individual level, it provides a large set of observations on individual happiness and other variables from 1983 to 1998. Table 7 provides an ordered probit analysis of the determinants of happiness at the individual level, taking into account income, age, education, religiosity (measured by the frequency of attendance), employment status and civil status, following the set of robust determinants in Bjørnskov et al. (2008). This analysis also includes the three social capital variables: social trust, informal sociability and formal sociability.
The analysis of the Needham data exhibits the standard findings. Income and religiosity are positively associated with happiness while higher education in general is negatively associated as is being without a spouse (for several reasons) and age shows a curvilinear relation with the lowest point at approximately 42 years. As is also standard in developed countries, gender is not associated with happiness. Finally, the three social capital variables all exhibit strongly positive associations with individual happiness, a set of findings that is consistent with the broader social capital literature (Putnam 2000, 2001; Uslaner 2002; Bjørnskov 2003). As such, there is evidence of effects of both social trust as argued in Uslaner (2002) and of sociability as in Winkelman (2006).
The main purpose of the individual level analysis, however, is to take out most of the individual-level effects of social capital in a rather simple way. As such, I calculate the residuals from the individual-level regressions and aggregate them annually at the cross-regional level or across the full period 1983–1998 at the state level. Across time, the raw happiness data and these residuals correlate at 0.90, at the regional level, the correlation is 0.87, ranging from 0.83 to 0.96, while the correlation at the cross-state level is 0.71. In principle, the control of individual-level effects may therefore make a difference, which is explored in Tables 3 and 4 in the main text.