Vote Overreporting: The Statistical and Policy Implications

Abstract

Voting is a critical way for citizens to gain representation for their interests, as well as the most direct way that most citizens participate in the political process. Yet, many citizens fail to cast their vote on election day. America's comparatively low turnout rate has motivated reams of research on voting behaviors, but researchers are limited by data quality problems. In particular, vote overreporting, when systematic, can bias analysis of who votes and why, which hampers efforts to understand and increase turnout. Previous research using individual-level National Election Survey (NES) validated data found demographic and regional patterns of overreporting. Since the NES no longer validates voting, individual-level data is no longer available and region is the only remaining determinant of systematic vote overreporting that can be studied. This paper uses census data to examine regional variations in vote overreporting in the 2000 and 2004 Presidential elections. After reviewing the actual turnout rate and how it increased from 2000 to 2004, I use both conventional ways to measure overreporting and examine variation by region. My results using both methods fit with past research indicating that overreporting is slightly higher in the South, but only in 2000.