Common knowledge and voter coordination: Experimental evidence from mali

Abstract

Information is well-known to impact voter behavior through its influence on an individual’s ability to evaluate politician performance. However, less is known about whether and how public information, or information shared by members the relevant public, additionally impacts voter decisionmaking relative to private information, or information available only to the relevant individual. When disseminated through public rather than private channels, information can lead voters to update priors about how other voters will behave in an election. If voters are acting strategically, then these revised expectations about others’ behavior can, in turn, influence an individual’s calculus at the ballot box. Public information is most likely to affect voter behavior through this mechanism when voters face a coordination problem. I argue this is the case in clientelist democracies where voters faces a trade-off between the certainty of a clientelistic transfer and the greater, but uncertain benefit afforded by a promising candidate campaigning on programmatic performance. I test this possibility using data from a public information experiment among 95 communes in Mali, some of which were provided with a civics course in six of their constituent villages. First, I show that respondents in treated villages are more likely to predict how others in their community vote in a hypothetical election. Second, I take advantage of variation in commune size and thus treatment dosage to show that effects arise only when a sufficient proportion, or majority, of the commune receives treatment. These data provide evidence in favor of the idea that public information can help voters coordinate when there are strategic complementarities to doing so.