Does reporting heterogeneity bias the measurement of health disparities?

Heterogeneity in reporting of health by socio-economic and demographic characteristics potentially biases the measurement of health disparities. Responses to anchoring vignettes have been proposed as a means of identifying reporting heterogeneity. We apply the vignette methodology to data from Indonesia, India and China in order to test for systematic differences in reporting of health by sex, age, urban/rural location, education and income and to establish the sensitivity of estimated disparities in health to the purging of reporting differences. The hypothesis of homogeneous reporting across all socio-demographics is rejected. Homogeneity tends to be most consistently and decisively rejected across urban/rural, income and age differences and less consistently across sex and education groups. In general, younger, male (not Indonesia), better educated (not China) low income and urban respondents display lower health expectations. Correcting for reporting heterogeneity tends to reduce disparities in health by age, sex (not Indonesia), urban/rural and education (not China) and to increase income disparities in health. Overall, while homogeneous reporting is significantly rejected, the results suggest that the size of the reporting bias in measures of health disparities is not large.