BEHAVIORAL SCIENCES WORKSHOP
Abstract: We develop an unbiased estimator through which we can quantify the number of times that someone voted in two states in this election, building on a probabilistic model first developed in McDonald and Levitt (2008). We apply this estimator to a national voter file to estimate the number of double votes in the 2012 presidential election. If we assume that electronic records are a completely accurate account of who voted, we estimate that about 30,000 people voted twice in the 2012 presidential election. But this assumption of no measurement error in electronic voting records is unlikely to be correct. Our audit of poll books in Philadelphia suggests that there is sufficient measurement error linking poll books to voter files - about 1% - to explain possibly all of the excess double votes we initially found. We conclude by evaluating the implicit tradeoff made by the Interstate Crosscheck Program, which compares voter registration data across participating states to aid in the purging of duplicate records and prosecution of double voters. We find that not only would few double votes be eliminated by purging the multitude of voter registration records flagged by Crosscheck, but many more records used to cast legitimate votes would be put in jeopardy.
Marc Meredith is an Associate Professor of Political Science at the University of Pennsylvania. Marc’s research examines the political economy of American elections, with a particular focus on the application of causal inference methods. His substantive research interests include election administration, election law, political campaigns, and voter decision-making.