A Spatial Change of Support Model for Differentially Private Measurements, with Application to Estimation of Counts of Persons in AIAN Areas by Detailed Race Groups

• Activity Number: 50 - Advances in Spatial Statistics for Survey Methodology and Official Statistics
• Type: Topic-Contributed
• Date/Time: Sunday, August 8, 2021 : 3:30 PM to 5:20 PM
• Sponsor: Survey Research Methods Section
• Author(s): Ryan Janicki* and Andrew Raim and Kyle Irimata and James A Livsey and Scott Holan
• Companies: US Census Bureau and U. S. Census Bureau and U.S. Census Bureau and U. S. Census Bureau and University of Missouri

Abstract

Over the past decade, advances in data science and the availability of vast amounts of public data has increased the risk of identifying individuals from published statistics. For this reason, the U. S. Census Bureau is developing a formal differentially private disclosure avoidance system for protecting 2020 Census data. A consequence of this is that it may not be feasible to publish the more granular Census data, such as counts for some detailed race, ethnicity, and tribal groups at low levels of geography, that has been produced in the past. In this paper, we introduce a spatial change of support model for the differentially private measurements, which utilizes auxiliary publicly available data sources, such as American Community Survey data, and past Census data. We show that accurate, model-based estimates of the number in a detailed race group in a geography which may be misaligned from the source data, such as an American Indian or Alaska Native area, can be made.