Abstract
Lacking a widely accepted general survey theory, which incorporates sampling errors as well as nonsampling errors, survey statisticians in charge of sample surveys have been forced to work out definitions and methods ad hoc to tackle problems more or less specific to a particular survey. As a result, we are now facing an embarrassing variety of terms and concepts, which makes an efficient communication of important survey findings impossible.
To highlight the existing semantic problems in one of the areas of nonsampling errors, viz. errors due to nonresponse, quotations of some definitions from dictionaries, well-known textbooks, and the 1979 Washington Symposium on Incomplete Data are given. Also, a short account of the basic framework for discussing and studying incomplete data used by Statistics Sweden is presented.
Get full access to this article
View all access options for this article.