Propensity Score Weighting for Causal Inference with Multiple Treatments

Seminar Series

Thursday, January 31, 2019 - 11:00
CRTP Classroom
Fan Li

Abstract:  Unconfounded comparisons of multiple groups are common in observational studies. Motivated from an observational study comparing three medications (causal comparison) and a racial disparity study in health services research (unconfounded descriptive comparison), we propose a unified framework, the balancing weights, for estimating causal effects with multiple treatments using propensity score weighting. These weights incorporate the generalized propensity score to balance the weighted covariate distribution of each treatment group, all weighted toward a common pre-specified target population. The class of balancing weights include several existing approaches such as inverse probability weights and trimming weights as special cases. Within this framework, we focus on a class of target estimands based on linear contrasts and their corresponding nonparametric weighting estimators. We further develop the generalized overlap weights, constructed as the product of the inverse probability weights and the harmonic mean of the generalized propensity scores. The generalized overlap weights correspond to the target population with the most overlap in covariates between treatments, similar to the population in equipoise in clinical trials. These weights are bounded and thus bypass the problem of extreme propensities. We show that the generalized overlap weights minimize the total asymptotic variance of the nonparametric estimators for the pairwise contrasts within the class of balancing weights. We consider two balance check criteria and propose a new sandwich variance estimator for estimating the causal effects with generalized overlap weights. We apply these methods to study the causal effect of three anti-coagulants on patient’s mortality and to estimate the racial disparities in medical expenditure. The operating characteristics of the new weighing method are further illustrated by simulations.

Speaker: Fan Li
Duke Biostatistics PhD Candidate