that should be true. but the reference group effect is about group averages, so when you compare widely different groups, those should-be differences would be suppressed. does that make sense?

Another way of putting this is that subjective SES measures have some nomological validity with objective economic standards. So you should take cross-cultural differences in self-report literally as long as your measure has validity with the thing it intends to measure.

We found the opposite: rich respondents in rich countries rated themselves higher than poor respondents in poor countries. It appears that "middle class" means the same thing across economic conditions, and this consistent interpretation produces valid cultural differences.

We examined the association between subjective SES and GDP. If subjective ratings are distorted by economic standards, you should not find a positive correlation, as rich respondents would underestimate their ratings in wealthy countries but overestimate them in poor countries.

Cross-cultural researchers often worry about using Likert scales across countries. Why? Because of the infamous reference group effect — the idea that people in each culture rate themselves relative to local norms.