Are Voters Rational? Information, Identity, and Voting Amongst the Baoulé and Dioula

Are voters rational? Since Kahneman and Tversky's original analysis indicating that people are vulnerable to framing effects, claims have been made that people are not rational. This includes voters. In many ways, voters have been shown to be vulnerable to framing and other effects which many candidates have used to their political advantage. If voters are not rational then this poses numerous problems for elections and democracy more generally. Moreover, if voters are not rational then relying on public opinion for policymakers is problematic because opinions could be easily swayed depending on the way questions are posed. Are voters rational? This paper studies this question developing a simple test. In a simple two candidate set-up with no information voters should prefer the candidate they most closely identify with but with additional information about the candidate’s policy positions then voters should choose the candidate they most closely identify with ideologically. In this study, we examine these expectations in an African context focusing on the role of ethnicity and voting. We examine the effect of information on voting decisions amongst west African respondents. We focus on two major ethnic groups in Côte d’Ivoire, specifically the Baoulé, an Akan group that has ruled the country for much of its time since independence, and the Dioula, a northern Mande group that has recently secured national power. The two groups have had contentious and conflictual political relations over the last twenty or so years. In a simple two candidate race, amongst Baoulé respondents we would expect that if shown just a picture and name that they will choose the Baoulé candidate over the Dioula candidate. However, if we add ideological information and make the Dioula candidate relatively preferable to the average voter (socially conservative, economically liberal) then if Baoulé voters are rational they should prefer the Dioula candidate over their co-ethnic. The same is true for Dioula respondents who should prefer the Dioula candidate in the low-information environment but the relatively conservative Baoulé candidate in the high information environment. If voters are not rational then they will vote for their co-ethnic in both low and high information environments. We test these claims in a series of field experiments in Yamoussoukro, a Baoulé stronghold, and in Abobo a Dioula stronghold.