报告题目： Efficient and robust estimation of average treatment effect with an error-prone treatment

报告时间：2023年11月15日（周三）15:00—16:00

报告地点：腾讯会议（会议 ID: 457-254-161）

报告人：魏少杰 4556银河国际在线

报告摘要:  

Many estimation procedures used for causal inference rely on accurately measured data. However, in experimental and observational studies, measurement errors are widespread. When a binary treatment indicator subjects to measurement error (misclassification occurs), methods that involve error-free treatment indicators may instead lead to biased estimates. In this paper, we present three classes of estimators that are respectively consistent and asymptotically normal in three different models and, meanwhile, propose a multiply robust estimation of the average treatment effect based on a semiparametric theory framework, for settings with misclassified treatment. The proposed multiply robust estimation is consistent when either one of the three models holds and achieves the semiparametric efficiency bound as long as all the models are correct. We demonstrate the satisfactory performance of the proposed methods through simulation experiments and real data analysis.

报告人简介:

魏少杰，现为4556银河国际在线师资博士后。魏少杰于2022年获得北京工业大学统计学博士学位，同年加入4556银河国际在线做博士后研究。研究兴趣包括因果推断、测量误差、生物统计。