This study leverages a unique opportunity to randomly assign election observers to a sample of over 1,000 of Ghana's 21,000 polling places during the country's December 2012 general elections. With the direct collaboration of the Coalition of Domestic Election Observers (CODEO), Ghana's most prominent domestic observation group, and CODEO's secretariat, the Centre for Democratic Development (CDD), the project's random assignment of a large subset of election observers permits causal estimates of the effects of observers on electoral violence and voter intimidation. The project compares outcomes at polling places located in constituencies with a high concentration of observers to outcomes at polling places in constituencies with a low concentration of observers. These comparisons will yield direct policy implications with respect to the optimal and most cost effective deployments of observers in future elections.

The intellectual merit of this project lies with how it advances understanding of political violence and intimidation. If election observers are effective, as donors and civil society groups believe, their presence at a polling place will generate a noticeable reduction in violence and intimidation. The data collected will allow estimations of the size of this effect. In addition, reductions in violence can also be used to estimate the impact of violence and intimidation on a range of other relevant outcomes. These include voter turnout, the political participation of women and other vulnerable groups, the extent of electoral support for incumbent legislators, and citizen perceptions of the election's legitimacy. The project will also produce a publicly-available micro-level dataset with which to test theories about where violence is most likely to occur.

The project will have broad impact on the placement of election observers in countries whose elections have histories of violence and intimidation. It will be the first to experimentally study the impact of election observers on violence and intimidation. The findings will yield direct policy implications regarding the optimal geo-spatial and numerical deployment of observers for future observer missions.

Project Report

International organizations deploy observers to monitor polling places in most countries around the globe on election day. These activities are expensive and demanding of the personnel involved. The funding comes from countries such as the United States and the United Kingdom. There is good evidence that international election observation reduces electoral malfeasance and improves election integrity on election day itself. Research under the direction of UCLA political scientist Professor Miriam Golden studies the effectiveness of domestic (rather than international) election observation in the 2012 combined presidential and parliamentary elections in Ghana in order to assess whether domestic observers have similar effects. One implication of finding that they do is that election observation may gradually be shifted from the international community directly to well-functioning domestic organizations without compromising observation effectiveness. The project was designed to investigate the impact of domestic election observers on electoral fraud, violence and other irregularities prior to and on election day using a randomized control trial. Approximately 1,000 of Ghana’s 4,000 domestic election observers were randomly assigned to polling places in four of the country’s ten regions, home to about half of the country’s population. Enumerators were sent a matched set of 1,000 control polling stations to assess the impact of observers. Enumerators also collected information about the conduct of the election campaign from approximately 6,000 randomly selected voters. The results of the research show that domestic election observers are effective in reducing election fraud. However, when political parties encounter an election observer, they respond strategically by shifting election fraud tactics to a neighboring polling station. They are especially successful in doing this in their stronghold areas where they exercise electoral dominance. This result is important because it suggests that domestic election observation is less effective in single-party dominant areas and that additional observation is warranted in such settings. The research also finds that where election observers were not present, biometric identification machines were more than twice as likely to experience failure and these failures were associated with greater election fraud. This result is important because it suggests that technical solutions alone are inadequate to improve election integrity. The research involved the participation of UCLA graduate students Joseph Asunka, Sarah Brierley, Eric Kramon and George Ofosu in every stage of the work. It also strengthened ties with Ghana’s main democracy-promoting NGO (the Center for Democratic Development), which was involved in implementing the research.

National Science Foundation (NSF)
Division of Social and Economic Sciences (SES)
Standard Grant (Standard)
Application #
Program Officer
Brian D. Humes
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
University of California Los Angeles
Los Angeles
United States
Zip Code