Data Sources for the Estimation of Hierarchical Models of Individual Mortality Outcomes

Troy Blanchard, Mississippi State University
Jeralynn S. Cossman, Mississippi State University
Martin L. Levin, Mississippi State University

A central challenge facing mortality research is the lack of available datasets for performing hierarchical analyses that: a) assess the influence of residential context on individual life chances, and b) examine how contextual factors condition the relationship between individual characteristics and mortality. The sample design of the most popular data sources largely preclude a hierarchical analysis. We develop a new method for the hierarchical analysis of individual mortality outcomes by pooling data from the 1990 Census of Population and Housing Public Use Microdata Samples and 1985-1990 Multiple Cause of Death Data. This methodology provides a data source for estimating Hierarchical Discrete Time Hazard Models that incorporates spatial context into the modeling of mortality outcomes. The implementation of this method provides a means to better understand how contextual factors, such as health infrastructure, community organizations, and the physical environment, condition health disparities across rural and urban portions of the U.S.