Alternative Paradigms for the Analysis of Imputed Survey Data
作者:
RobertE. Fay,
期刊:
Journal of the American Statistical Association
(Taylor Available online 1996)
卷期:
Volume 91,
issue 434
页码: 490-498
ISSN:0162-1459
年代: 1996
DOI:10.1080/01621459.1996.10476909
出版商: Taylor & Francis Group
关键词: Fractionally weighted imputation;Missing data;Multiple imputation;Rao—Shao variance estimator
数据来源: Taylor
摘要:
Rubin has offered multiple imputation as a general approach to inference from survey data sets with missing values filled in through imputation. In many situations the multiple imputation variance estimator is consistent. In turn, this observation has lent support to a number of complex applications. In fact, however, the multiple imputation variance estimator is inconsistent under some simple conditions. This article extends previous work of Rao and Shao and of Fay directed toward consistent variance estimation under wider conditions. Extensions of Rao and Shao's results tofractionally weighted imputationcombines the estimation efficiency of multiple imputation and the consistency of the Rao—Shao variance estimator.
点击下载:
PDF (1384KB)
返 回