Application of Differential Privacy to the 2020 U.S. Census and American Community Survey: Researchers’ Understanding and Reactions

Abstract

The U.S. Census and the related American Community Survey (ACS) are used for studies of many types by researchers in a variety of domains. The U.S. Census Bureau discovered that the methods intended to preserve the privacy of individuals used for the 2010 survey were not adequate. Thus, a decision was made to apply differential privacy (DP) to the data from the 2020 Census. DP methods introduce noise into the data set with the intention of allowing the group statistics still to be useful, while protecting the individual data by way of random perturbation. Researchers expressed concern as to how much this application of DP will impact research. We report results of a survey conducted of researchers who published studies based on the 2010 U.S. Census data or the ACS data to assess their understanding of and concerns about DP.