The relationship between specific and global measures of life satisfaction or happiness is usually seen as an asymmetric bottom-up relationship. Domain satisfaction and general satisfaction are independently measured and the typical question is whether the former has an impact on the latter or to what extent the former contributes to the latter. This paper, in contrast, investigates whether general life satisfaction can be statisti-cally derived from domain-specific satisfaction. It is suggested to conceptualize life satisfaction as a second-order factor, which is only indirectly related to observable variables. In the first part of the empirical analysis, it is shown that such a model fits the data of the first AsiaBarometer quite well. The factor models for the Asian countries under investigation are not equivalent but strikingly similar. The four first-order factors refer to satisfaction with (1) the family, (2) social status and social health, (3) economic status, and (4) collective goods. These factors load highly on a second-order factor, which is called general life satisfaction.

The results suggest the calculation of a specifically weighted factor score for each country and to base the further analyses on these scores. In the second part of the empirical investigation, it is demonstrated, however, that simple unweighted indices, which are calculated under the assumption of meaning equivalence, are very highly correlated with the country-specific factor scores. The further analysis is therefore based on an unweighted index.

In the last part of the empirical study, the reliability and validity of the life satisfaction index is examined and compared with the validity of a single indicator of overall happiness. Similar variables are used in many international surveys. Index and indicator are moderately correlated. It is shown that the life satisfaction index on average correlates higher with well-established determinants of happiness than the single item. This is particularly true for the pooled international data set, which includes the data from all participating countries. Accordingly, indices seem to be more reliable and valid measures of general life satisfaction, and may be better suited for comparative analyses at least for the countries under investigation.