Some comments on ‘the structure of subjective well-being in nine western societies’ by andrews and inglehart

This paper makes a number of critical comments on a recent article by Andrews and Inglehart (1979), where the similarity of subjective well-being data from nine different countries was investigated. It is first shown that different MDS representations can be constructed for the data which differ considerably among each other and with respect to the similarity hypothesis. The goodness of the MDS solutions is discussed by using finer indices than overall Stress. Then, the measures of configurational similarity are evaluated statistically. Finally, it is shown that other, equally valid measures lead to completely different conclusions. It is argued that this poses an insurmountable problem. A facet theoretical approach which looks for prototypical structures rather than a single similarity index avoids this dilemma.