Political Violence and Social Networks: Experimental Evidence from a Nigerian Election

Pedro C. Vicente
1500 Nigerians living in 24 enumeration areas within six states of Nigeria
2007 - 2007
Target group:
  • Voters
Outcome of interest:
  • Electoral participation
  • Violence
  • Voter Behavior
Intervention type:
  • Information
  • Social networks

Violence affects the legitimacy of elections and electoral participation worldwide. Researchers conducted a randomized evaluation in Nigeria to determine whether a campaign against electoral violence could spread through social networks to affect the perceptions and behaviors not only of the individuals targeted by the campaign but also others to whom they have social ties. The clearest effects of the campaign were on decreasing the perceptions of violence among those with social ties to the targeted individuals, most often through family links.

Policy issue

Threats of violence may negatively affect electoral participation, especially in young democracies. Awareness campaigns can sometimes decrease the negative effects of violent intimidation on voter turnout. It is usually infeasible for campaigns discouraging electoral violence to reach every voter in a country. Still, campaigns may have the potential to indirectly affect more individuals than they can reach directly. For example, family ties, social interactions, and geographical proximity could help spread the content of information campaigns, allowing them to reaching more individuals. Can the effects of a campaign against electoral violence be indirectly transmitted through social and geographical proximity? 

Context of the evaluation

Since its independence in 1960, Nigeria has experienced corruption and political instability, with a long period of military rule. In 1999, Nigeria passed a new constitution and adopted civilian rule. However, early elections (in 1999, 2003, and 2007) were affected by many instances of electoral misbehavior, including ballot fraud, violence, and voter intimidation. This evaluation coincided with the 2007 election, which included all federal and state political bodies. Again, international electoral observers reported that the election was compromised by corruption including ballot fraud and violence. Human Rights Watch reported that at least 300 individuals died due to violence related to the elections. Umaru Yar'Adua of the People’s Democratic Party (PDP) won the presidential election with 70 percent of the votes, and the PDP party won 28 of 36 races for governor. 

This evaluation examines the indirect effects of an information campaign carried out by ActionAid International Nigeria (AAIN) with the goal of reducing voter intimidation due to violence. The program took place in Lagos, Oyo, Delta, Rivers, Kaduna, and Plateau states, which were chosen due to their past history with political violence. 

A group of people in Nigeria enters a polling station
Voters gather at a polling place in Nigeria.
Photo: Gregade | Shutterstock.com

Details of the intervention

Researchers partnered with AAIN to conduct a randomized evaluation to test the indirect effects (through social networks) of AAIN’s voter awareness campaign on perceptions of violence and voter behavior. AAIN’s goal was to lower the perceived threat to individual voters through delegitimizing and therefore reducing the effectiveness of the use of violence and voter intimidation. The two-week campaign took place two months before the 2007 elections. The campaign included anti-violence slogans (No to political violence! Vote against violent politicians!) printed on posters, pamphlets, and clothing. AAIN also facilitated town meetings, which provided an opportunity for voters to meet with local representatives to discuss ways of counteracting politically motivated violence. Finally, public theater in Yoruba, Hausa, and English featured one good and one bad politician, with the bad politician relying on violent intimidation.

Researchers randomly selected two enumeration areas (EAs) from each of the six states in the sample to receive the campaign (the treatment, or targeted, group) and two to not receive the campaign (the comparison group). Researchers conducted pre and post-election surveys with 50 individuals in each EA. An additional 300 individuals (25 per EA) who lived in treatment EAs but did not personally interact with the campaign were surveyed in the post-election survey. This defined a third group—the untargeted group, as opposed to the treatment or comparison groups. 

The goal of the evaluation was to learn how the effects of the campaign could be spread through these social networks. In some cases, targeted individuals were socially or geographically close to other targeted individuals, in which case a reinforcement effect might have caused the effects of the campaign to be stronger for these individuals. In other cases, untargeted individuals were socially or geographically close to targeted individuals, in which case a diffusion effect might have caused the effects of the campaign to reach individuals who otherwise would not have been at all exposed to the program. 

Researchers used surveys to measure the perceptions and experiences of violence and the voter behavior of targeted, untargeted, and comparison individuals. Researchers also measured empowerment by giving participants pre-stamped postcards with a message opposing voter intimidation and measuring how many postcards were sent. In order to understand how the campaign’s effects spread across social networks, researchers collected information on family ties, frequency of social interactions, and physical distance between individuals. 

Results and policy lessons

Overall, the campaign led to a 47 percent reduction in reports of violence during the election. Researchers also found both reinforcement and diffusion effects—in other words, the campaign had both a stronger direct and a wider indirect impact due to its message being spread through social networks. In general, perceptions of violence were affected by both reinforcement and diffusion, while empowerment and voting behaviors were more affected by diffusion than reinforcement. 

Perceptions of violence: The program had a stronger effect on the perceptions of violence of targeted individuals with family ties to other targeted individuals, as well as those who lived close to other targeted individuals. Additionally, the program affected the perceptions of violence of untargeted individuals with family ties or who lived close to targeted individuals, meaning that the program was able to lower the perception of violence of those who were not directly affected by any part of the campaign. 

Voter empowerment: Both targeted and untargeted individuals who reported chatting to targeted individuals were more likely to mail the postcard with an anti-violence message as a result of this social interaction. Although family ties and geographical proximity to targeted individuals did not reinforce the empowerment results of the program on other targeted individuals, family ties did make it more likely that untargeted individuals would mail the postcards.   

Other voter behaviors: Voter turnout was not affected by social ties to individuals targeted by the campaign, but family ties to targeted individuals did make other targeted individuals more likely to vote for incumbents (who tended to be less associated with electoral violence). Both chatting to and having family ties to targeted individuals made it more likely that untargeted individuals would vote, and that they would vote for the incumbent. 

Taken together, these results suggest that voter awareness campaigns focused on reducing electoral violence may have indirect effects on both perception of violence and actual voter behavior, and that these effects may be passed on through social networks. Family ties seem to be the strongest pathway for these effects. Specifically, perception changes may be increased by geographical proximity to campaign materials, but voting behavior seems to require actual social interactions with those targeted by the campaign. These findings suggest that it may be possible to increase the effect of campaigns against electoral violence by taking advantage of social networks. This could be achieved, for example, by mobilizing civil society through local organizations or volunteers.