Do Neighborhoods Matter: Self-Assessed Health and Residential Selection Among Older Americans

Objective: Studies of neighborhood influences on self-assessed health have consistently shown correlations between self-assessed health and the economic status of the surrounding neighborhood environment. For older adults, self-assessed health is not only an important measure of current health but is also a predictor of future health trajectories. It has been shown to predict mortality and the onset of chronic conditions, disability and other health outcomes. Policies aimed at successful aging in place will require an understanding of neighborhood effects on the health and health trajectories of older adults. However, for these policies to be effective, it is essential to determine whether the relationship between self-assessed health and the neighborhood economic environment is causal or due to residential selection. Methods: The study uses data from two waves of the Health and Retirement Study (HRS), 1992/1993 and 2000. Data on neighborhood characteristics in 1990 and 2000 were obtained from RAND's Center for Population Health and Health Disparities and were drawn primarily from the 1990 and 2000 Census and the Environmental Protection Agency’s Air Quality System. All measures were created using uniform boundaries across time, in terms of 1990 census tracts. Because many neighborhood measures are highly correlated, we formed scales for each year by applying factor analysis to neighborhood environment characteristics. Two scales reflecting the economic circumstances of the neighborhood were formed for each year: one reflecting economic disadvantage and the other economic advantage. Multilevel models of self-assessed health as a function of individual, family and neighborhood characteristics were estimated, stratified by gender. Neighborhood characteristics included economic advantage, economic disadvantage and three additional scales representing immigration concentration, residential stability and air pollution, in order to control for non-economic features of the neighborhood. The potential for residential selection was examined using data from 1992/1993 and 2000. Neighborhood economic features (advantage and disadvantage) were modeled in a difference-in-difference framework. Models included a dummy variable indicating year, a dummy variable indicating baseline self-assessed health status, an interaction between the two, and all individual and family-level controls. The models were stratified by move status between 1992/1993 and 2000. All models were run separately for men and women. Results: A higher level of neighborhood economic disadvantage was found to be associated with a lower probability of being in excellent or very good health for men. A higher level of economic advantage was associated with a higher probability of being in excellent or very good health among women. Some evidence of selective residential mobility with respect to neighborhood disadvantage was found for both men and women. The direction of the bias for cross-sectional models of self-assessed health, however, was downward. Similar findings held with respect to neighborhood economic advantage, but only for women. Thus, while there is evidence of selective residential mobility, it tends to bias the findings of neighborhood effects on health status towards the null.