Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 56 - Causal Inference
Type: Contributed
Date/Time: Sunday, August 8, 2021 : 3:30 PM to 5:20 PM
Sponsor: Section on Statistics in Epidemiology
Abstract #318604
Title: Estimation of ATO Achieving Covariate Balancing with Generalized Empirical Likelihood
Author(s): Ryo Otani* and Hiroshi Yadohisa
Companies: Graduate of Culture and Information Science, Doshisha University and Doshisha University
Keywords: propensity score; average treatment effect; overlap weighting; covariate balancing; generalized empirical likelihood

In the analysis using the propensity score, it is very important that estimated propensity score are balancing covariates and the method is not susceptible to bias caused by propensity scores with extreme values. At recent years, covariate balancing propensity score(CBPS) and average treatment effect for the overlap population(ATO) using overlap weighting are proposed to deal with covariate balancing and extreme-value propensity scores respectively, but neither can deal with both covariate balancing and extreme-value propensity scores. In this study, we propose the method that can deal with both covariate balancing and extreme-value propensity scores using generalized empirical likelihood(GEL), which is augmentation of generalized method of moment(GMM) used in covariate balancing propensity score for overlap weighting(CBPSOW). This proposed method is also augmented method of CBPSOW. In numerical simulation, the proposed method is relatively more stable in estimating causal effects in situations with small sample size data than the existing methods described above. We further show real data application and compare other methods.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2021 program