|
Activity Number:
|
151
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 30, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #308541 |
|
Title:
|
Semiparametric Efficient Causal Inference with Missing Data
|
|
Author(s):
|
Yue Shentu*+ and Minge Xie
|
|
Companies:
|
Merck & Co., Inc. and Rutgers University
|
|
Address:
|
10 Landing Lane, New Brunswick, NJ, 08901,
|
|
Keywords:
|
Propensity Score ; Observational Study ; Missing at Random ; Semiparametric efficiency bound
|
|
Abstract:
|
A missing data problem is investigated in the context of causal inference. In real-life observational studies, the response of interest may often be missing for a subgroup of subjects, and the missing mechanism may depend on post-baseline outcomes as well as baseline information. Motivated by a semiparametric efficiency bound of consistent treatment effect estimation, we proposed an estimator that incorporates the propensity-weighting and the regression imputations. We show that under mild assumptions, the proposed estimator is n-1/2-consistent and semiparametrically efficient. In addition, we show that the proposed estimator is robust against some of the model mis-specifications. Simulation studies were carried out to compare the proposed estimator with other existing estimators and to demonstrate its desirable properties.
|
