The Effect of the Differential Privacy Disclosure Avoidance System Proposed by the Census Bureau on 2020 Census Products: Four Case Studies of Census Blocks in Alaska
David Swanson
University of California Riverside
Tom Bryan
Tom Bryan Demographics
Richard Sewell
Alaska Department of Transportation and Public Facilities
The Census Bureau plans to introduce a new Disclosure Avoidance System known as Differential Privacy (DP) for its 2020 census data products. Using two DP demonstration product files provided by the Census Bureau, we assess the errors introduced by DP on census block data in Alaska in the form of four case studies and find them to be substantial by type and level. We use both the May 27th 2020 DP demonstration product and the most recent, the April 28th 2021 DP demonstration product relative to our four cases studies and compare the changes. This comparison is important because the Census Bureau reports that accuracy should improve because the privacy budget was increased in response to user complaints about poor accuracy. We find that the April 28th, 2021 release does produce more accurate data but that the level of accuracy remains unsuitable for use by those who work with small area data. Because it is likely that the results we found in Alaska will be found in other states, our examination leads us to conclude that it is likely that the errors introduced by DP of the type and at the level found in the most recent demonstration product files we examined will render the nation's b