Election Observers and Electoral Fraud
Abstract: We report results of a project that leverages the random assignment of over $1,000$ domestic elections observers to polling places during Ghana's 2012 general elections to study three measures of likely election day fraud in a developing democracy. Our research strategy is designed to measure the impact of observers on fraud at the stations where they are deployed and to detect whether election observers induce the strategic movement of fraud from the polling places where they are stationed to nearby but unobserved polling places. We find evidence that observers reduce the probability that more voters will vote in a polling place than are officially registered by about 60 percent, and they similarly reduce unnaturally high levels of election turnout. In party strongholds observers also reduce instances of ballot stuffing. We additionally find evidence of spatial spillover effects. Unobserved polling places spatially proximate to observed polling places are 75 percent more likely to experience fraud than unobserved polling places in areas without observers nearby. These spillover effects are concentrated in the strongholds of Ghana's two major political parties, empirical evidence that political parties are most effective at coordinating fraud by moving it away from polling stations under observation where the party's local organizational capacity is greatest. Our results suggest that fraud on election day in Ghana is the product of deliberate behavior by well-organized political parties seeking to inflate their national vote totals to win the presidency, and have implications for our general understanding of the conditions in which parties are able to coordinate electoral fraud.
