|
Abstract:
|
Statisticians have developed many ways to assess census (or other registration) coverage. The standard, of course, is dual-system estimation. Related techniques, such as reverse record check and the "megalist" method, also have been proposed. These techniques generally use microdata, rather than aggregates; use actual records, rather than estimates of aggregates; and have a well-established statistical theory behind them that allows statistical inferences. In both Census 2000 in the United States and Census 2001 in the UK, however, it was demographic methods of assessing coverage that ultimately provided important clues that the census enumeration process had over- and under-coverage in particular areas. This paper discusses whether information integration theory can help link demographic and statistical methods in an overarching framework.
|
