External validation of the use of vignettes in cross-country health studies

Cross-country comparisons of subjective assessments are rendered difficult if not impossible because of sub-population specific response style. To correct for this, the use of vignettes has become increasingly popular, notably within cross-country health studies. However, the validity of vignettes as a means to re-scale across sample sub-populations critically rests on the assumption of "response consistency" (RC): that vignettes and self-assessments are evaluated on the same scale. In this paper, we seek to test this assumption by applying objective measures of health along with subjective measures and vignettes. Our results indicate that the assumption of RC is not innocuous and that our extended model relaxing this assumption improves the fit and significantly changes the cross-country rankings of health vis-à-vis the standard Chopit model.