时 间：2026年9月23日（周三）15：30-17：00
地 点：普陀校区理科大楼A1514室
报告人：Changbao Wu University of Waterloo教授
主持人：刘玉坤 华东师范大学统计学院
摘 要：
The statistical analysis of nonprobability survey samples has become a major area of research in survey sampling and official statistics. We consider a two-sample setting in which a nonprobability sample contains labeled data, while a probability reference sample provides auxiliary information about the target population. In this setting, standard transfer-learning methods cannot be applied directly to achieve valid and efficient statistical inference. We review recent advances in analysis of nonprobability survey samples and examine the issues and challenges associated with three inferential frameworks under the two-sample setting. We also present theoretical and empirical results on estimating participation probabilities (i.e., propensity scores) under both ignorable and nonignorable participation mechanisms.
报告人简介：
Changbao Wu is Professor of Statistics in the Department of Statistics and Actuarial Science at the University of Waterloo. His research interests include the design and analysis of complex surveys, resampling methods, missing data analysis, causal inference, and the integration of data from multiple sources. He is a Fellow of the American Statistical Association and the Institute of Mathematical Statistics, an Elected Member of the International Statistical Institute, and the 2026 recipient of the Statistical Society of Canada Gold Medal. He has served on the editorial boards of several journals, including Biometrika, The Canadian Journal of Statistics, the Journal of the American Statistical Association(Theory & Methods) and Survey Methodology. He is the lead author, with Mary Thompson, of the book “Sampling Theory and Practice”, published by Springer in 2020. Since 2015, he has also served on Statistics Canada’s Advisory Committee on Statistical Methods.