Prediction with Differential Covariate Classification: Illustrated by Racial/Ethnic Classification in Medical Risk Assessment

Classifying people into groups—like race and ethnicity—is common in public policy and clinical medicine, but inconsistent definitions can make these classifications unreliable for predicting outcomes and making decisions. A recent working paper by Charles Manski, John Mullahy, and Atheendar Venkataramani explores the challenges of using data from different classification schemes, particularly when data are limited.