Impact of the Census Disclosure Avoidance System on Redistricting

In attempting to protect the privacy of 2020 Census respondents, the Census Bureau has made its data unsuitable for redistricting purposes.

Christopher T. Kenny (Department of Government, Harvard University) , Shiro Kuriwaki (Department of Government, Harvard University) , Cory McCartan (Department of Statistics, Harvard University) , Evan Rosenman (Harvard Data Science Initiative) , Tyler Simko (Department of Government, Harvard University) , Kosuke Imai (Departments of Government and Statistics, Harvard University)
May 28, 2021

The U.S. Census Bureau plans to protect the privacy of 2020 Census respondents through its Disclosure Avoidance System (DAS), which attempts to achieve differential privacy guarantees by adding noise to the Census data. The Bureau has asked for feedback on the adequacy of DAS-protected data for real-world purposes. The ALARM project has conducted an extensive analysis into how the DAS-protected data affects redistricting and voting rights analyses, and has submitted these findings to the Census Bureau.

Read the report: The Impact of the U.S. Census Disclosure Avoidance System on Redistricting and Voting Rights Analysis

Also see: Answers to Frequently Asked Questions Added on June 2, 2021.

A figure from the report, indicating partisan biases in the DAS-protected data.

By applying redistricting simulation and analysis methods to DAS-protected 2010 Census data, we find that the protected data are not of sufficient quality for redistricting purposes. Compared to the original Census 2010 data, we find that the DAS-protected data:

Our primary recommendation is to release Census P.L. 94-171 data without using the Disclosure Avoidance System, and instead rely on a swapping method similar to that applied to the 2010 Census data in order to protect respondent privacy.

If the Census Bureau decides to apply the current DAS to Census PL. 94-171 Data, we recommend increasing the privacy loss budget and allocating the increase to improving redistricting outcomes. In particular, preserving the accuracy of populations at the voting tabulation district level would be critical. The Bureau must avoid injecting noise that systematically undercounts certain racial and partisan groups in the privacy-protected data. Additional recommendations are given in the paper.