In addition to reconsidering how we operationalize racialized categories in our statistical models, to move towards a more anti-racist research agenda we must also consider intersectionality (ethnicity, gender, age, sexual orientation, social economic class, disability). I am grateful for this terrifically written article by Greta Bauer from 2014 that I only wish I would have found sooner…Incorporating intersectionality theory into population health research methodology: Challenges and the potential to advance health equity that clearly describes both theory and methods. https://www.sciencedirect.com/science/article/pii/S0277953614001919
Of high relevance to data scientists doing the modeling work see section “4.5 Understanding differences between types of regression models for intersectional applications” Dr. Bauer reviews why we should be estimating interaction on the “additive-scale” rather than the “multiplicative-scale” and encourages researchers to use related measures like RERI, or the synergy index. Data scientists who are participating in meaningful health research should become aware and stop just testing cross-product terms in logistic regression when trying to test if “risks interact”! Link to a presentation on testing interactions