How Valid are Voter Registration Statistics? A Demographic Approach for Assessing the Validity of EAVS Voter Registration Data
52 Pages Posted: 27 Jan 2019 Last revised: 2 May 2019
Date Written: January 11, 2019
This paper explores the test validity of items in the Election Administration and Voting Survey (EAVS) that pertain to voter registration and voter-list maintenance. The method that is developed relies on testing the correspondence between EAVS items and related demographic statistics that should be the main drivers of the reported administrative statistics. (For instance, the number of removals from the voter rolls due to death that a county reports should track closely with the number of adult deaths in the county for the same period.) This approach finds a close correspondence with underlying demographics for two items: total new registrations and removals because of death. The approach finds much lower correspondence between list-maintenance statistics and corresponding demographic measures, especially items related to the removal of voters from the rolls because they moved. The approach also finds lack of correspondence between the number of pre-registered young people and the population of young people in a county. Lack of high correlations between certain EAVS items and their associated demographic factors can be due to a number of factors, including differences in recordkeeping and policy implementation, but also raise the question of whether EAVS items reflect a uniformity of data reporting across the country.
Suggested Citation: Suggested Citation