The Evolution of Monitoring: Evidence from Text Analysis of Election Monitoring Reports

Abstract

After the Cold War, election monitoring activities increased significantly, and research on the topic has risen sharply in the last ten years. These are valuable contributions, but we believe one point requires further consideration: empirically clarifying how monitoring has changed over time. This is because fraudsters have begun to shift election manipulation from the day of the election to other times in order to adapt to election monitoring. If fraudsters have evolved over time, have monitors kept pace with this evolution? Employing an original dataset of election monitoring reports published by major monitoring organizations, this study performs large-scale text analysis of these reports to show that the criteria used by monitoring teams have indeed changed over time. Specifically, although election monitoring groups have been criticized for their bias in emphasizing manipulation on election day more than other factors, we demonstrate that such bias has declined over time.