Interpersonal influence and network effects on voting behavior: experimental evidence from Mozambique

During the 2009 Mozambican elections a randomized controlled trial implemented a voter education campaign. My thesis analyses the peer effects triggered by that campaign. The first essay looks at peer effects within the household. The second essay (co-authored with Marcel Fafchamps and Pedro Vicente) focuses on peer effects at the village level. The campaign targeted heads of household or their spouse (i.e., the primary target). In the first essay (Chapter 2) I test the effect of the campaign on other household members. I interpret this effect as evidence of interpersonal influence exerted by the primary target over other household members. The main finding of this analysis is that the transmission of voter education campaign’s effects tends to occur through sharing of opinions and social pressure, instead of information sharing. In this essay I also explore the determinants of interpersonal influence within the household. I test whether the effect of the campaign on other household members (i.e., secondary targets) varies with age, gender and education. I find a stronger effect on younger secondary targets, consistent with the idea that they are more susceptible to social pressure. The second essay (Chapter 3) examines whether the campaign’s effect is transmitted within the village through social networks (kinship and chatting) and geographical proximity. We test whether the impact of the campaign on targeted and untargeted individuals depends on proximity to other targeted individuals in the village. Our main finding is that the campaign increases voter participation on average, but much less so among individuals who are socially or geographically closer to other targeted individuals. This result is interpreted as evidence of free-riding on voting as a civic duty.