Brief 16: Spillover Effects of Observers in Ghana

The authors of this study attempt to analyze the extent of the inflation of pools as politicians' reactions to the number of election observers at a particular polling site.

Link to Full Study

Category: Corruption, Elections

Date of Publication: Friday, March 20, 2015

EGAP Researcher: Nahomi Ichino

Other Authors: Matthias Schündeln

PDF: EGAP-Policy-Brief-Ichino-Ghana-Final1.pdf

Click to Download the Data

Geographical Region: Africa

Research Question:

How does the presence of domestic election observers affect electoral irregularities in a new democracy? Does monitoring displace irregularities to un-monitored areas?

Preparer: Sarah Zimmerman, Lauren Young



Domestic and international observers are a common policy tool to prevent fraud during elections. Fraud can occur ahead of elections during voter registration and during the actual poll. One common and under-studied type of fraud is inflation of the voter roll, which can enable a party’s supporters to vote multiple times. Higher than expected voter registration is indicative of an inflated voter roll, and it can be deterred by monitoring during the registration process. However, because electoral monitoring typically covers only a subset of registration or polling places, it is important to understand how parties strategically react to the presence of monitors. If parties react to observers by moving irregularities to nearby areas, it may affect the distribution but not the overall level of fraud. Understanding displacement is key to accurately evaluating the impact of observers on the quality of the election.

Research Design:

The authors designed and implemented a randomized field experiment to test two hypotheses: first, areas with observers should have smaller increases in registrations than areas without observers. Second, areas without observers located near observed areas should have larger increases in registrations than those that are far away from observed areas. Four of the ten regions in Ghana were selected, including Ashanti, Brong Ahafo, Greater Accra, and Northern Regions (see map below), to cover a wide range of constituencies and incumbent and opposition strongholds.

The researchers randomize observers in two stages. First, they randomly assign some constituencies to treatment and some to control. This ensures that there are some areas without any direct observations or spillovers. Second, within treatment constituencies, electoral areas were assigned to treatment or control. Approximately 25% of electoral areas in treatment constituencies were assigned to actually receive observers.

The observers were recruited from the Coalition of Domestic Election Observers (CODEO) a Ghanaian NGO, and all received training and accreditation. Observers stayed at each registration center about 1-2 hours on each first visit, and up to a full day on return visits; afterwards they returned one-page checklists with questions about the registration center to the CODEO secretariat.

The researchers estimate the impact of observers by considering that observers affect the registration center where they are directly located, as well as the centers within 5km and 5-10km radii. They also allow for a center to be affected by both direct observation and spillover effects. They use the percent change in registrations by registration center as an indicator of inflation of the voter roll.


The authors found strong evidence that voter roll inflation is lower at directly observed registration centers. However, the reduction in registration is negated by displacement of fraud to electoral areas that are close to areas with domestic observers.

The results show that electoral areas in treatment constituencies that received observers (but were not exposed to any spillovers) had reductions in registrations of 7.6 percentage points compared to control constituencies. If that observed electoral area was also within 5 km of a second observed electoral area, the drop in registrations is reduced to 5.9 percentage points. For electoral areas that are not directly observed but are within 5 km of an observed area, there is a reduction of only 1.4 percentage points.

These results support both of the authors’ two hypotheses. First, there was less fraud in constituencies with observers than those without (4.1 percentage points). Within treatment constituencies, there was less fraud in observed than unobserved electoral areas (3.5 percentage points). This shows that observers do reduce fraud. Second, there is evidence of spillover effects that suggest that parties are reacting strategically to the presence of observers. Specifically, electoral areas within 5 km of observed areas had increases of 2.7 percentage points in fraud.

Policy Implications:
  • Policymakers should focus on how best to prevent misconduct during the often overlooked voter registration stage.
  • Even a small number of domestic observers can have a significant effect on reducing fraud; therefore, election observation organizations should consider even a small allocation of observers to be a worthwhile investment.
  • Domestic observers should be densely concentrated near the areas where high levels of fraud are expected.
  • Policymakers should carefully consider whether their interventions could have spillover effects, and if so, how to evaluate the impact of their programs taking these into account.
  • More scholarship is necessary in using randomized field experiments to study how democracy works in practice