Extending and Implementing Multiple Imputation Technology

点击翻译按钮获取中文摘要

Multiple imputation technology refers to a family of techniques for dealing with incomplete data. This innovative technology, to be perfected under the recommended accomplishment- based renewal, will be especially useful in the context of data bases that have been constructed at great public expense and are to be made available to many data analysts with their arsenals of standard statistical software. Therefore, and in addition to its importance for the academic research community, this innovative technology has implications for statistical endeavors undertaken by government, such as census undercount adjustment and similar forms of compensation for deficient data (so-called "statistical repair work"). Such work is needed more than ever before as data collection resources shrink and missing data problems multiply owing to financial and other restrictions. Good quality analysis is also at stake when, as is becoming increasingly true, incompleteness pervades intensively used scientific data from sources as far-ranging as social surveys and clinical medical protocols.