Event



Penn Sociology Colloquium Series: Issa Kohler-Hausmann, Professor of Law and Associate Professor of Sociology, Yale University and Lily Hu, Ph.D. Candidate in Applied Mathematics and Philosophy, Harvard University

Colloquium
"What Is Perceived When Race Is Perceived and Why It Matters for Causal Inference"
Nov 3, 2021 at - | McNeil 150

Many pressing social scientific inquiries concern how race influences decision making in a domain where selection into that domain may have also been affected by race. Much debate about how to carry out these causal analyses has centered on highly technical methodological questions about which race-causal estimands are identifiable, under what assumptions, and with what type of data. In this paper, we instead direct attention to a set of conceptual questions that must be prior to these methodological ones. We focus on a particular estimand that has been of substantial interest to causal inference practitioners, one which looks to identify the causal effect of (only) race on (only) a second-stage outcome. First, we ask what this estimand at the center of these longstanding debates substantively means. Second, given what the estimand claims to capture, we ask whether the statistical moves proffered in fact achieve these aims. We will show that race-causal estimands that are defined to capture the isolated effect of race only do so if one assumes that a decisionmaker presented with a racial cue perceives nothing more than the membership conditions for racial groups, something like skin color or ancestry. This assumption is plainly at odds with a rich social psychology literature that has developed in recent decades on racial cognition. We argue, therefore, that foremost methods do not realize the double isolation aim of their causal inference exercise.