Temporal Changes in Self-Rated Health: APC Models of Racial Disparities

Audrey N. Beck, San Diego State University
Brian K. Finch, RAND Corporation
Shih-Fan Lin, San Diego State University

This paper uses the National Health Interview Survey (1972-2009) to describe and explain temporal patterns in racial health disparities. We employ cross-classified random effects models of age-period-cohort (APC) to understand the patterns of self-rated health disparities over time. Second, we use decomposition models to understand compositional shifts across cohorts. Finally, we examine the extent to which cohort-level measures of relative cohort size and economic and health conditions at birth explain the disparity. Cohort disparities increase through the 1935 cohort for women, falling thereafter; disparities for men begin their overall decline earlier. Differences in socioeconomic composition consistently contribute to disparities; notably, differences in marital status are increasingly important across cohorts for women whereas employment differences emerge as increasingly salient for men. Our cohort characteristic models suggest that cohort economic conditions (percent large family, farm or Southern birth) dramatically reduce disparities in health.