Detection of Election Fraud

Abstract

While election fraud is mostly detected and reported by real political actors like election observers, academic research can contribute through systematic analysis of previous information in determining when fraud has occurred and estimating how many votes were affected. This chapter presents social science research, based on the idea that election fraud is conceived as an intervention in random processes that generate fraud-free “normal votes.” The authors discuss two points: first, which fraudulent acts exist and how they intervene in fraud-free election processes; second, which assumptions enable random fraud-free election processes. They distinguish among several approaches in existing research: those that set up a fraud-free random process in a theory-guided way, which requires some predictors of normal votes; natural- or field-experimental approaches that exploit existing random processes; and models that capture the random process in a single election result.