Does Race have Confounders?

Studying of the effect of racial differences on health outcomes is a difficult task for reasons including varying definitions of race - many of which may not adhere to current best practices - its philosophical acceptance as a well-defined exposure, and appropriate utilization of factors with important roles in the relationship between race and outcome. Inattentive approaches to any of these issues can lead to biased, irrelevant, or unreplicable study findings. We highlight these concerns paying special focus to the issue of covariates in analyses of racial causal effects. We show that mis-identifying many common covariates as confounding variables can bias estimates of racial causal effects. Rather than defaulting to using covariates as confounding variables, researchers should carefully consider the role each variable plays in the relationship between race and outcome and how accounting for each will affect effect estimates and their interpretability.