PDF version

The Perils of Building Democracy in Africa: Evidence from Kenya

Location: Kenya
Sample: 49 million registered voters from 12,160 polling stations
Timeline:
2013
Research Initiative: Governance Initiative
Target Group: 
General
Outcome of Interest: 
Electoral participation
Intervention Type: 
Information
Partners:

Citizen trust and participation in the political system are necessary for stable democratic regimes. During the 2013 National Elections in Kenya, the Independent Electoral and Boundaries Commission (IEBC) sent text messages to approximately two million registered voters to promote public interest and knowledge and to raise voter turnout. Researchers found that the text message campaign increased voter turnout but decreased trust in the electoral commission. Trust decreased most among voters linked to the party that lost the election for president and those who experienced violence because of the election. Overall, the results indicate that raising expectations about the quality of elections may reinforce distrust and dissatisfaction with democracy if institutions responsible for organizing and supervising elections do not fully deliver on their promises. 

Policy Issue 

A key challenge faced by young democracies is how to organize fair and transparent elections that build citizen trust in electoral institutions. Citizen trust and participation in the political system are necessary for the stability of democratic regimes, as dissatisfaction (especially among losers of an election) can lead to violent forms of protest. One potential strategy to increase citizen satisfaction with democracy is to improve the administration of elections, which could increase citizens’ confidence that their vote was counted and increase their perception of the government’s performance. While there is a growing body of research on the impact of providing information on short-term electoral outcomes, there is a need for more evidence on the effects of information campaigns on citizens’ trust in democracy and the electoral system.

Context of the Evaluation 

In the March 2013 national elections in Kenya, citizens were asked to vote for six different positions: President, Member of Parliament, Ward Representative, Governor, Senator, and Women’s Representative. The 2013 elections were the first organized by a new electoral commission, the Independent Electoral and Boundaries Commission (IEBC). The IEBC replaced the Electoral Commission of Kenya (ECK), which had been held partially responsible for the failure of the previous national election that led to an outburst of interethnic violence in 2007-2008. Kenyan politics are organized along ethnic lines; individuals typically vote for candidates representing their ethnic tribe.  

The IEBC took several measures to reduce voter fraud and ensure fairness of the elections, including using biometric voter registration kits to uniquely identify voters with their fingerprints and photographs. In addition, the IEBC invested in electronic systems that would improve transparency by publicly reporting online real-time results from polling stations.

However, on the day of the election, the IEBC faced major difficulties organizing the ballot. The biometric identification either did not work or was not used in about half of the polling stations, and the real-time results display was delayed for various technical reasons. Officials at many polling stations had to identify voters and count ballots manually, and the online system displayed results later than anticipated. These technical difficulties caused public doubt about the IEBC’s management of the elections.

Details of the Intervention 

Working in partnership with the IEBC, researchers used a randomized evaluation to measure the impact of information dissemination on voter participation and trust in the new electoral institutions. The study included 12,160 polling stations covering 4.9 million registered voters with a Safaricom cell phone number, the dominant telecom operator in Kenya.

During the six days leading up to the March 2013 elections, the IEBC sent text messages to approximately two million registered voters across the entire country. The messages aimed to promote public interest and knowledge about the election and to raise voter turnout. Registered voters were randomly assigned, by polling station, to receive either: 

  (i.)         Basic encouragements to vote

  (ii.)         Information on the responsibilities of each position up for election, excluding the president, plus an encouragement to vote for each of the six positions

 (iii.)         Information on the transparency and neutrality of the IEBC

 (iv.)         No messages

Researchers measured impacts on voter participation using official electoral data from the IEBC. Researchers also measured political attitudes via a phone survey around eight months after the election and the intensity of election-related violence, using geocoded violence data, right before and eight months after the election.

Results and Policy Lessons 

The text message campaign led to a small increase in voter turnout, but it decreased trust in the electoral commission. Trust decreased most among voters who experienced election-related violence in their constituency and among voters linked to the party that lost the election for president. Overall, the results indicate that raising expectations about the quality of elections may reinforce distrust and dissatisfaction with democracy if institutions responsible for organizing and supervising elections do not fully deliver on their promises.

Voter turnout: Voter turnout increased at polling stations where voters received text messages. Receiving a text message treatment increased turnout by 0.3 percentage points, from a base of 87.7 percent in polling stations where voters received no messages. This effect was driven by messages that provided a basic encouragement to vote.

Trust in the electoral commission: The information campaign decreased trust in the new electoral commission. Survey respondents who received any of the messages were 3.7 percentage points less likely to trust in the IEBC, compared to 80 percent of respondents who did not receive any messages. Respondents who received any of the messages were also 2.6 percentage points less likely to report being very satisfied with Kenyan democracy, relative to 32 percent of respondents who did not receive messages. These results suggest that the messages interacted with external information on the actual quality of the elections, including the failure of the biometric and electronic system. The text messages may have increased expectations about the ability of the IEBC, which then led to greater doubt about the election’s fairness when the election system failed.

Role of election violence and ethnicity: The messages had a larger negative effect on political attitudes for voters who experienced election-related violence in their area and for voters from ethnic tribes whose coalition lost the presidential race. Of those receiving messages, 90.5 percent of winning coalition members expressed trust in the IEBC, compared to 64.8 percent of losing coalition members

Support for democratic principles: Decreased trust in the IEBC did not debase support for democratic principles in general. Citizens who received any messages remained equally likely to find democracy preferable to any other kind of government; to agree that leaders should be chosen through regular, open, and honest elections and be actively questioned; to agree that all individuals should be permitted to vote; and to disagree with the use of violence in politics.

Marx, Benjamin, Vincent Pons and Tavneet Suri. “The Perils of Voter Mobilization,” NBER Working Paper #23946, October 2017.