> There can be
> a significant self-selection bias in the sample, as the better off people
> are more likely to participate in life satisfaction surveys .
>
Better off, as in higher income, etc.? I doubt that. The first rule of survey design is that the demographics of the samples are checked against the demographics of the reference population. These surveys have been done for decades. Lots of problems can creep into them, but not such an obvious error on such a well understood issue. Besides, why would it happen more in one country than another?
> people in the "middle" will report higher
> levels of subjective satisfaction than people in the "middle" of countries
> with lower level of social stratification, because they reference groups are
> different. The former are much better off than the underclass (their
> reference group) and say so in the survey.
>
The underclass is the reference group for people in the middle? Why not equally the overclass?
> Finally, I think Joanna is absolutely correct - there is a cultural
> difference between the US and Europe in how whining is perceived. That may
> explain quite a bit about the differences in subjective perceptions.
>
Yes, I think that's Kahneman's whole point. That's why he calls SHS "reality-free."
> In general, cross-national comparisons of absolute scores of subjective
> scales are pretty meaningless because of these cultural differences.
In general, they consistently show strong and predictable correlations with objective measures, like income.
> A
> better way is comparison of relative scores,
Woj, there's already a huge literature in which relative scores are compared and analyzed. But ideally you'd also like to find meaningful ways to investigate cross-cultural SWB.
SA