CMStatistics 2021: Start Registration
View Submission - CMStatistics
Title: Enhanced doubly robust procedure for causal inference Authors:  Ao Yuan - Georgetown University (United States) [presenting]
Anqi Yin - Georgetown University (United States)
Ming Tan - Georgetown University (United States)
Abstract: The doubly robust estimator is a popular tool in causal inference, which provides double protection of unbiasdness. However, most existing methods for such an estimator use parametric models, and are not robust enough. We propose a semi-parametric model for this estimator, in which both the propensity score and outcome models are semi-parametric, with a non-parametric link function to enhance the robustness and parametric regression effects for easy interpretation. Simulation studies are conducted to evaluate the performance of the proposed method and compared with existing parametric and naive methods, showing a clear advantage of the proposed method. The method is then applied to analyze real clinical trial data.