Causal Models of Segregation and Health

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Residential segregation has been argued by many as a fundamental cause of racial disparities in health (Acevedo-Garcia 2003; Schulz et al. 2002; Williams & Collins 2001). Segregation disproportionately affects minority groups by concentrating poverty, diminishing the quality of neighborhood social and physical environments, and attenuating individual socio-economic attainment and upward mobility (Collins and Williams 1999). However, we still do not fully understand the extent to which neighborhood segregation contributes to health disparities-not simply because relatively few studies have explicitly addressed this question, but because many of the past research attempts to measure the effects of racial segregation on health have used inappropriate methods and untenable assumptions (Duncan et al. 1997) and have relied largely on aggregate level data (Williams & Collins 2001). We are able to remedy many of the shortcomings in conventional analysis in the following ways. One, while prior studies have relied almost exclusively on aggregate data that is subject to ecological fallacy, we are able to utilize individual-level data to examine relationships between segregation and health (self-rated health and mortality) as well as neighborhood-level (census-tract) data to examine the possible pathways through which segregation affects health. Second, utilizing panel data, we can estimate the effects of segregation at various points in time during the life course; prior studies have almost all relied on a single (contemporary) point-in-time estimate of neighborhood effects on health. Third, we are able to calculate a broad set of segregation measures in addition to the traditionally used dissimilarity index-measures that have a much stronger theoretical and analytical correlation with concentrated poverty and other aspects of social space that may have a much stronger connection to health. Finally, and most importantly, by using panel data we are able to at least partially control for unobserved heterogeneity and recover causal estimates of the effects of segregation on health outcomes by controlling for unobserved individual-level characteristics through the application of various fixed-effect analyses. This project investigates the effect of residential segregation over the life-course on the health and mortality of Black and White adults using a sophisticated panel data set, merged with census data at the census-tract and county levels. This project will estimate unbiased effects of segregation and neighborhood poverty to determine how much any potential observed effects contribute to known racial disparities in health and mortality.