B1583
Title: Automatically Calibrated Sensitivity Models for Nonparametric Causal Inference with Unmeasured Confounding
Authors: Alexander McClean - Carnegie Mellon University (United States) [presenting]
Zach Branson - Carnegie Mellon University (United States)
Edward Kennedy - Carnegie Mellon University (United States)
Abstract: Accounting for unmeasured confounding is crucial when estimating causal effects with observational data. For this purpose, we propose automatically calibrated sensitivity (ACS) models, which nonparametrically bound the error due to unmeasured confounding by an analogous notion of error due to measured confounding, multiplied by a sensitivity parameter --- thereby combining sensitivity and calibration models. We illustrate how to construct ACS models via several examples and demonstrate their advantages over standard sensitivity and post hoc calibration analyses. We then outline efficient estimators for bounds on the Average Treatment Effect and a one-number summary of effect sensitivity. We observe that either a margin condition or smooth approximation is required for efficient estimation and develop methods for estimation and inference with both. Moreover, we establish that our estimators are doubly robust, and attain parametric efficiency and asymptotic normality under nonparametric conditions on estimators for the probability of receiving treatment and outcome regression function. Finally, we illustrate our methods with two data analyses.
