统计与数据科学学院
---------------------------------------------
学术讲座
讲座题目：Theory and implementation of doubly robust estimators with nonignorable missing data
讲 座 人：苗旺 （北京大学）
时 间：2019年6月3日(周一)上午11:00-12:00
地 点：统计与数据科学学院126教室
摘 要：We consider identification and estimation with an outcome missing not at random (MNAR). We study an identification strategy based on a so-called shadow variable. A shadow variable is assumed to be correlated with the outcome, but independent of the missingness conditional on the outcome and fully observed covariates. We describe a general condition for nonparametric identification of the full data law given a valid shadow variable under MNAR. Our condition is satisfied by many commonly-used models, and thus essentially state that lack of identification is not an issue in many situations. Moreover, our condition is imposed on the complete cases, and therefore it has testable implications with only observed data. As a result, identification or lack of it can in principle be assessed. We describe semiparametric estimation methods and evaluate their performance on both simulation data and a China Home Pricing example. We derive the set of all influence functions that have a double robustness property, and we characterize the semiparametric efficiency bound for the class of doubly robust regular and asymptotically linear estimators.
邀 请 人：王磊博士
欢迎广大师生参加！
统计与数据科学学院
2019年5月23日