Election Forensics

Abstract Determining the veracity of an election is a crucial task for democracies. Numerous safeguards, such as election monitoring and dispute resolution institutions, exist to try to protect electoral integrity and ensure that election results match the will of voters. Election forensics—a collection of statistical techniques designed to help determine whether the outcome of an election represents the voters’ will—can be a valuable supplement to these efforts. Forensic techniques rely on the difficulty of perpetrating election fraud without leaving any trace: credibly falsifying electoral results surreptitiously is not straightforward. Whether election fraud is attempted after the fact or during the election itself (e.g., by stuffing the ballot box or preventing opposition voters from casting their ballots), these actions distort results in ways that can potentially be measured. Election forensics add value to other forms of election monitoring, as they are comparatively low-cost and do not require the presence of election observers or additional institutional processes. Furthermore, they have the benefit of providing geographically disaggregated estimates of fraud. Finally, these techniques can also provide an estimate of the magnitude of fraudulent votes in a particular election, which can be crucial in determining whether or not any electoral problems actually influenced the overall outcome. These factors contribute to their practicality for widespread use across a wide range of countries. Despite this potential, these techniques have significant drawbacks and the output of these techniques must be interpreted cautiously and within context to avoid the risk of falsely labeling a clean election as fraudulent.