Ordinal health disparities between population subgroups: measurement and multivariate analysis with an application to the North-South divide in England

Health disparities between population subgroups classified on the basis of nominal characteristics such as sex, caste, race or region are of major academic and policy concern. The paper develops a novel analytical framework to not only measure differences in ordinal health outcomes between population subgroups but also account for such disparities in terms of the individual-level socioeconomic and demographic characteristics of their members. The measurement approach is directly applicable to the ordinal health and well-being data commonly available from general social surveys, building on the concept of statistical preference to motivate the definition of summary indices of comparative subgroup health and between-group variation in health. The analysis employs indirect standardisation techniques based on the estimation of a health distribution regression model for the population to identify the effects of compositional and conditional health differences on subgroup health outcomes. An illustrative empirical study finds that about half of the regional variation in self-reported health within England in 2016/17 can be accounted for by sociodemographic factors, with age and educational qualifications both more important predictors than income.