作者: Donald B. Rubin
DOI: 10.1080/07350015.1986.10509497
关键词:
摘要: Statistically matched files are created in an attempt to solve the practical problem that exists when no single file has full set of variables needed for drawing important inferences. Previous methods matching reviewed, and method concatenation with adjusted weights multiple imputations is described illustrated on artificial example. A major benefit this approach ability display sensitivity inference untestable assumptions being made creating file.