518 – Inference from Combined Data Sets
Estimates of Correct and Erroneous Enumeration with Duplicates in the 2010 U.S. Census
Rachel Bray
U.S. Census Bureau
Although historically there has been low net undercount in the census, using a single measure to describe coverage may mask additional errors. The 2010 Census Coverage Measurement (CCM) program produced estimates of the components of census coverage in addition to estimates of net error. The components of census coverage for persons in the 2010 Census are correct enumerations, erroneous enumerations, whole-person imputations, and omissions. The data collected in 2010 CCM support the breakdown of correct and erroneous enumerations into various subgroups. The estimates include correct enumerations at the national level, for all states, and large counties and places. A key subgroup of erroneous enumerations is duplicates. This paper explores the results of 2010 CCM estimation, with particular emphasis on duplicates. We report estimates by the distance separating the linked records, including duplicates at the same address, or in the same block, county, or state. We also estimate the count of duplicates according to their characteristics, such as whole households of duplicates, and whether the duplication is a result of mailback or non-mailback census enumerations.
