Adjusting Self-Assessed Health for Potential Bias Using a Random-Effects Generalized Ordered Probit model

We study how socioeconomic conditions, especially relative household income, affect self-assessed health. We use a random effects generalized ordered probit model with data from China Health and Nutrition Survey (CHNS) to test for heterogeneity in how people assess their health. The results show that individuals with high relative income are less likely to report poor health, but they are also less likely to report extremely good healthy. Although SAH capture many aspects of health elements, it might be biased on some socioeconomic features.