A Counterfactual Simulation Method to Evaluate Mortality Selection Effect in the Cohort Trend of Health Disparities

• In Event: 4150 - Topics in Sociological Methodology, Session 1

Abstract

Objectives: Unobserved individual frailty is prevalent and consequential in the population pattern of health and mortality. Changing frailty variance across cohorts can alter the “survivor” compositions of the population over life course and the slope of mortality acceleration across cohorts. If frailty variance changes within groups (e.g., lower educated vs. higher educated) across cohorts, mortality selection mechanism may further alter the slope of mortality acceleration for each group, and complicate the cohort trend in health disparities and life expectancy gap between groups. Besides mortality selection, cross-cohort changes in health disparities can be affected by various confounding mechanisms. Methods: We propose a counterfactual simulation procedure to remove these confounding factors and evaluate the extent to which the widening mortality disparities by two educational groups (any college vs. high school or less) in the United States are due to mortality selection. The basic idea is to create two scenarios: one with mortality selection effect present, the other counterfactual one with mortality selection effect absent. Then the residual differences in cross-cohort changes in health disparities between these two scenarios are due to mortality selection effect. Results: Using the number of diseases before age 17 from Panel Studies of Income Dynamics as a proxy for frailty and cohort changes in age-dependent mortality pattern from National Health Interview Survey data, we find mortality selection contributes to -14% of the widening life expectancy gap between these two educational groups from the 1950s to 1960s cohort among men while 41% of the widening life expectancy gap among women. Conclusions: We can either overestimate or underestimate life expectancy gap across cohorts without purging mortality selection effect. Counterfactual simulation method is flexible in its assumptions with regard to hazard function and distribution form of frailty.

Author

• Hui Zheng, The Ohio State University