多重填补的方法及其统计推断原理.PDFVIP

See discussions, stats, and author profiles for this publication at: /publication Article · January 2003 CITATIONS READS 0 1 3 authors, including: Yang Cao Örebro University, School of Medical Sciences 146 PUBLICATIONS 90 CITATIONS SEE PROFILE Some of the authors of this publication are also working on these related projects: Technique development in CT View project Prognostic predictors of ECT for depression View project All content following this page was uploaded by Yang Cao on 03 May 2017. The user has requested enhancement of the downloaded file. 　2003 6 10 2 ·77 · 1 2 1 曹　阳 　谢万军 　张罗漫 　　【】　　描述数据缺失的特征和数据缺失模式, 对 Rubin 最早提出的多重填补(multiple imputation, M I)的基本概念、填补和分析缺失数据的方法、综合 计推断进行了探讨, 分析了M I 的特点、局限性以及应用M I 方法处理不完整数据集时需要注意的地方。　通过计算机模拟, 用M I 方法将每 一个缺失值用 一系列可能的 值填补, 然后使用常规的、针对完全数据集的 计方法对多重填补后得到的若干数据集进行分析, 并把所得的结果 进行综合。　多重填补值显示出了缺失数据的不确定性, 使得已有数据得到了充分利用, 从而对总体参数做出 了更为准确的估计。　M I 方法为处理存在缺失值的数据集提供了有用的策略, 并且适用于多种数据缺失的场 合。 【】　多重填补　缺失数据　广义线性模型　logistic 回归　马尔科夫链蒙特卡罗 :R195.1　:A 　:1006-5253(2002 )04-0077-05 Methods of Multiple Imputation and Related Inference Theory 　Cao Yang, Z hang Luoman.Dep artm en t of Health S tatis tics, The S econd M ilitary M ed ical Un iversity , Shanghai, 200433.X ie Wan un, Classif ication, A ssess men t and S urvey Un it, WH O, S witzerland, Geneva. 【Abstract】　Objective　To describe the basic concepts, methods or imputing and analy zing missing data, combining in erence o multiple imputation (M I)technique w hich w as originally proposed by Rubin, and discu ss limitations and cau-