578 – Statistical Explorations for the Post-Enumeration Survey of the U.S. 2020 Census
Creating a Hard-To-Enumerate Score to Stratify the 2020 Post-Enumeration Survey Sample
Krista Heim
U.S. Census Bureau
Courtney Hill
U.S. Census Bureau
T. Trang Nguyen
U.S. Census Bureau
Timothy Kennel
U.S. Census Bureau
This research identifies and explores characteristics highly correlated with coverage issues for possible use in the sample design of the 2020 Post-Enumeration Survey. If the relationship is sufficiently strong, grouping primary sampling units into strata based on these variables could produce smaller variances than if the stratification had not used these variables. We look at several data sources, including the American Community Survey and extracts from the Master Address File. We create various versions of what we refer to as the Hard-to-Enumerate Score, including straightforward summarizations and model-based approaches. Further research compares the stratification of primary sampling units based on these scores to the stratification based on the prior post-enumeration survey, the 2010 Census Coverage Measurement Survey. We found that the stratification for the 2010 Census Coverage Measurement Survey resulted in similar variances to the alternative stratifications.