In the United States, people are asked to vote on a myriad of candidates, offices, and ballot questions. The result is lengthy ballots that are time intensive and complicated to fill out. In this paper, we utilize a new analytical technique harnessing ballot scanner data from a statewide midterm election to estimate the effects of ballot complexity on voting errors. We find that increases in ballot length, increases in the number of local ballot questions, and increases in the number of candidates listed for single offices significantly increase the odds of encountering ballot marking and scanning errors. Our findings indicate that ballots’ characteristics can help election administrators make Election Day planning and resource allocation decisions that decrease ballot errors and associated wait times to vote while increasing the reliability of election results and voter confidence in the electoral process.
