Balancing 2020 Census Cost and Accuracy: Consequences for Congressional Apportionment and Fund Allocations (WP-18-10)


Zachary Seeskin and Bruce Spencer


The researchers examine how accurate the 2020 census needs to be, given that accuracy is expensive but inaccuracy distorts distributions of congressional seats and federal funds. Although the 2010 census had small measured errors for states, 0.6 percent on average (as measured by root-mean-square error, RMS), they project that Texas loses and Minnesota gains a seat if the 2020 census has the same errors. Projections further show that if 2020 census error for state populations increases to 0.7 percent RMS, an additional seat is lost by Florida and gained by Ohio, and if error increases to 1.7 percent RMS, Texas loses a second seat, to the benefit of Rhode Island. The researchers find expected distortions in fund allocations increase about $9–$13 billion for each 0.5 percent increase in average error.

Zachary Seeskin, Statistician, NORC at the University of Chicago

Bruce Spencer, Professor of Statistics and IPR Fellow, Northwestern University

The software and data used to generate the analyses are available here.

PDF icon Download working paper PDF