Post-Election Auditing: Effects of Procedure and Ballot Type on Manual Counting Accuracy, Efficiency, and Auditor Satisfaction and Confidence

A task common to nearly all types of election auditing is that of human auditors physically counting ballots by hand. This task, fundamental to the goal of accuracy in an audit, can be a source of error. While somewhat basic in its nature, the process of counting can be strongly influenced by many procedural and legal factors. In the current study, we examine how specific group counting procedures and ballot types affect the accuracy, efficiency, and subjective judgments of usability of a post-election audit. These two procedures, quite different in their implementation and employed in real elections in two U.S. states, have built-in redundant checks and multiple tallies to help bolster accuracy; we found that even with this redundancy, errors are surprisingly frequent. Additionally, certain counting procedures are more efficient, as well as less variable in the amount of error they introduce into the audit process. We found that well-specified procedures, as well as division of labor amongst group counting members, help ensure more accurate and efficient ballot audits.