Analyzing the external validity of causal findings against subpopulation shifts

• Activity Number: 396 - Distributional Robustness, Validity, Causality, and Generalizability
• Type: Invited
• Date/Time: Thursday, August 12, 2021 : 2:00 PM to 3:50 PM
• Sponsor: IMS
• Author(s): Hongseok Namkoong*
• Companies: Columbia University
• Keywords: causal inference; semiparametric inference; distributional robustness; subpopulation shifts; fairness

Abstract

Typical data collection systems represent only a particular set of points in space and time, and are biased against marginalized demographic groups. We propose the worst-case treatment effect (WTE) across all subpopulations of a given size, which allows analyzing the sensitivity of a study's findings to unanticipated covariate shift. Compared to the average treatment effect (ATE) whose validity relies on the covariate distribution of collected data, WTE is robust to unanticipated covariate shifts, and positive findings guarantee uniformly valid treatment effects over subpopulations. We develop a semiparametrically efficient estimator for the WTE, leveraging machine learning-based estimates of the heterogeneous treatment effect and propensity score. By virtue of satisfying a key (Neyman) orthogonality property, our estimator enjoys central limit behavior---oracle rates with true nuisance parameters---even when estimates of nuisance parameters converge at slower rates.

This talk is based on a joint work with Sookyo Jeong.