Investigating the Relationship between Neighborhood Context and Health: A Marginal Structural Modeling Approach
D. Phuong Do, University of South Carolina
Lu Wang, University of Michigan
Michael Elliott, University of Michigan
Despite well-documented associations between between neighborhood context and health, causal inference has been difficult to establish. Common concerns leading to bias include measurement error, omitted individual-level variables, and off-support inferences. However, one important limitation which is perhaps less frequently noted is that variables such as income and employment status may themselves be affected by prior neighborhood conditions and may consequently be simultaneously confounders and mediators. Using data from the 1996-2007 Panel Study of Income Dynamics, this study applies a marginal structural modeling strategy to appropriately adjust for factors that are simultaneously confounders and mediators to recover estimates of neighborhood poverty on 2007 health status: good health, poor health, and death. Estimates from the marginal structural modeling strategy reveal statically significant larger effects of neighborhood poverty on poor health and mortality than conventional "naïve" regression stratification adjustments. Results suggest that conventional strategies have been underestimating neighborhood effects on health.