摘要
在抽样调查等实际问题中,经常出现数据缺失.针对这类问题,通常的处理方法之一是对数据进行插补。本文综述了抽样调查中处理缺失数据常用的插补方法。重点讨论了单一插补的方差估计与多重插补的简化计算以及使用回答概率的单一插补等。最后讨论目前插补所面临的问题与其发展方向.
During survey sampling and many other practical problems, there are often missing data. Imputation is one common tool to solve this problem. This paper is devoted to make a review of imputation methods for missing data. Then, variance estimates of single imputation, simplified calculation of multiple imputation and imputation using response probability are studied. The present problem and development trends of imputation are pointed out in the end.
出处
《数理统计与管理》
CSSCI
北大核心
2008年第5期821-832,共12页
Journal of Applied Statistics and Management
基金
总装预研重点基金9140A19030106HK0108资助
关键词
缺失数据
插补
单一插补
多重插补
回答概率
missing data, imputation, single imputation, multiple imputation, response probability
