Intersectional decomposition analysis with differential exposure, effects, and construct

In recent years a wide array of proposals for bringing intersectional perspectives into quantitative studies of health disparities have appeared, from studies of interaction, predictive discrimination, to mediation. Bauer and Scheim, in a companion set of articles, extend these proposals by developing new attribution-blind measures of perceived discrimination and using VanderWeele's 3-way decomposition to quantify its contribution to disparities through differential exposure and differential effects (sometimes called differential vulnerability or susceptibility). In this commentary, after providing an overview of causal inference interpretations with social characteristics, we provide a broad overview of old and new decomposition methods in the social sciences literature and contrast their strengths and weaknesses for studying intersectional inequalities. We then examine how different forms of differential effects can be expressed within these decompositions and discuss their utility for the purpose of informing interventions for reducing disparities. Last, we discuss the tension in social sciences research when prominent explanatory variables represent constructs that are only defined or exist for certain marginalized populations and may not neatly fit within the decomposition methods framework. Through these discussions, we aim to provide greater conceptual clarity for applied researchers who are interested in using decomposition methods and other approaches to advance intersectional equity.