Researchers often rely on household survey data to investigate health disparities and the incidence and prevalence of illness. These self-reported health measures are often biased due to information asymmetry or differences in reference groups. Using the World Health Organization study on global ageing and adult health, I find that the poor use a different reporting scale from the more affluent, leading to overestimation of their health status.

This is tested by using the relatively novel anchoring vignettes approach and applying the hierarchical ordered probit model. Underestimation by the poor of their ill health could mean that South Africa’s high levels of socioeconomic health inequalities are greater than realized.