On the Causal Interpretation of Race in Regressions Adjusting for Confounding and Mediating Variables

Abstract

We consider several possible interpretations of the “effect of race” when regressions are run with race as an exposure variable, controlling also for various confounding and mediating variables. When adjustment is made for socioeconomic status early in a person’s life, we discuss under what contexts the regression coefficients for race can be interpreted as corresponding to the extent to which a racial inequality would remain if various socioeconomic distributions early in life across racial groups could be equalized. When adjustment is also made for adult socioeconomic status, we note how the overall racial inequality can be decomposed into the portion that would be eliminated by equalizing adult socioeconomic status across racial groups and the portion of the inequality that would remain even if adult socioeconomic status across racial groups were equalized. We also discuss a stronger interpretation of the “effect of race” (stronger in terms of assumptions) involving the joint effects of race-associated physical phenotype (e.g. skin color), parental physical phenotype, genetic background and cultural context when such variables are thought to be hypothetically manipulable and if adequate control for confounding were possible. We discuss some of the challenges with such an interpretation. Further discussion is given as to how the use of selected populations in examining racial disparities can additionally complicate the interpretation of the effects.

Sex and Racial Disparities in Cardiac Rehabilitation Referral at Hospital Discharge and Gaps in Long‐Term Mortality

“…To examine the proportion of sex, racial, and regional gap in mortality that could potentially be reduced through optimal CR referral, we performed a causal mediation analysis 33, 34. We estimated the association between sex, race, and region with 1‐ and 3‐year mortality using the generalized estimation equation model.…”

Explaining intersectionality through description, counterfactual thinking, and mediation analysis

“…Proof The non-parametric results under no covariate feedback follow directly from those of VanderWeele and Robinson [11] and also VanderWeele and Tchetgen Tchetgen [16]. It follows from the results of VanderWeele and Robinson [11] that these formulae hold under an assumption that the effects of test scores M are unconfounded given race, SES, and covariates ethnicity, gender, and age (along with consistency and positivity assumptions for test scores).…”

“…Many of Schwartz's questions revolve around the causal status of race, a topic which has been considerably debated in the statistics, social science, and more recently the epidemiology literature [5][6][7][8][9][10][11]. Contrary to Schwarz's assumption, our study did not consider race as a cause, nor did we consider potential outcomes for race.…”

“…Specifically, we noted that causal mediation analysis [11] could be used to examine how much the joint disparity in say, log-wages, comparing blacks with low SES to whites with high SES, would decrease after removing disparities in some target, say some test of pre-market skills obtained through education. Similar analyses examining racial disparities have garnered considerable attention in the field of labor economics.…”

“…But simply observing that a disparity exists does not tell us how to reduce it or what certain interventions might achieve. To make progress, we move from counterfactual thinking to counterfactual methods that conceptualize a finer model where a disparity in some outcome (e.g., wages) is perpetuated by disparities in, or the differential effects of, manipulable determinants of that outcome (e.g., educational attainment) [11,[23][24][25]. With counterfactual methods, we examine the extent to which an intervention on that determinant might reduce disparities (as in observational data) [26] or actually does so (as in a trial) [27,28].…”