Do Shifts in Late-Counted Votes Signal Fraud? Evidence from Bolivia

Abstract

Shifts in late-counted votes often spark unfounded claims of electoral fraud. These claims exploit the early-count mirage: the expedient illusion that, absent fraud, an early advantage will persist. We characterize the early-count mirage and evaluate associated fraud claims in four disputed elections, focusing on the case of Bolivia in 2019. When late-counted votes delivered a narrow victory for the incumbent, fraud accusations followed—with dramatic political consequences. But we find that the vote-share trend can be explained without invoking fraud and that the allegedly suspicious shift in late-counted votes was actually an artifact of methodological and coding errors on the part of electoral observers. We document similar patterns in the other three cases. The details are context specific, but the core insights are general: time trends from legitimate vote-counting processes are far more varied—and errors in influential analysis far more frequent—than election skeptics allege.