Analyzing the Imputations for Nonresponse in the 2020 Census Post-Enumeration Survey

• Activity Number: 19 - 2020 Census Post-Enumeration Survey Estimation Methods: Research and Challenges
• Type: Topic Contributed
• Date/Time: Sunday, August 7, 2022 : 2:00 PM to 3:50 PM
• Sponsor: Survey Research Methods Section
• Author(s): Michael Beaghen* and Richard Neil Turner
• Companies: U.S. Census Bureau and U.S. Census Bureau
• Keywords: census-coverage error; logistic regression

Abstract

The Census Bureau conducted a Post-Enumeration Survey (PES) to assess the 2020 Census' coverage of population and housing units (HU). Census person coverage errors included omissions, duplication, people enumerated in the wrong place, and people who should not have been enumerated. The 2020 PES was a probability sample of about 170,000 HUs nationwide. PES methodology required an Independent Listing of HUs and people in HUs in a sample of geographies. This independence of the Census was necessary to satisfy the requirements for the dual-system estimator. The PES listings were matched to the census listings of HUs and people. Accurate matching included automated probability-based matching, followed by clerical matching, and by field interviewing to resolve differences in the listings. When clerical matching and field interviewing did not provide enough information to determine the status of P-sample or E-sample people, these statuses were imputed with logistic regression models. This report documents the methodology and the results of these imputations.