A unified framework to account for selective mortality in lifecycle analyses of the social gradient in health

This paper establishes a unified framework to fully account for the changing social gradient over the lifecycle in terms of a sufficient set of mobility indices characterizing the coevolution of socioeconomic status and health within each of a series of overlapping cohorts. We proceed to demonstrate the impact of selective mortality on health inequality changes, making use of a counterfactual health distribution for the start of the study period that leaves out those who are known to die before the end. Specifically, initial differences between the average health and educational attainment rank of survivors and nonsurvivors are found in our empirical study to be an increasingly important factor explaining changes in the education‐health gradient in older cohorts in Great Britain. Our identification strategy has the advantage that it does not require the imputation of the “would be” health of nonsurvivors, which if carried out using inverse probability weighting procedures—as in several previous studies—is shown to bias estimates of this direct effect of selective mortality toward zero. Parallel results for the income‐health gradient exhibit stronger confounding influences due to a number of other factors given that income is a less stable indicator of socioeconomic status in adulthood.