You are here:
Structural confounding of area-level deprivation and segreation: an empirical example
Citation:
MESSER, L. C., S. Mason, AND J. M. Oakes. Structural confounding of area-level deprivation and segreation: an empirical example. Presented at Population Association of American Annual Meeting, New Orleans, LA, April 17 - 19, 2008.
Impact/Purpose:
research results
Description:
The neighborhood effects literature has grown, but its utility is limited by the lack of attention paid to non-random selection into neighborhoods. Confounding occurs when an exposure and an outcome share an underlying common cause. Confounding resulting from differential allocation of people to place, by virtue of selection processes, is called structural confounding. Structural confounding ensures that people residing in different residential environments are not exchangeable, because even when matched on typical covariates, they still differ on the unmeasured confounders that contributed to the differential neighborhood selection. Non-exchangeability makes them non-comparable, and their non-comparability renders causal contrasts across neighborhoods meaningless. Given that most statistical modeling approaches depend on exchangeability across contexts, structural confounding limits investigators' ability to make causal inference from observational data. Here, we explore structural confounding in Durham and Wake County NC. The analysis presented is an empirical demonstration of the limitations to multilevel modeling posed by structural confounding.