Personal Connections to the Political World: Social Influences on Democratic Competence in Brazil and in Comparative Context

Conversation is at democracy’s core. In this dissertation, I examine citizens’ political discussion networks and their effects on democratic competence, meaning what citizens know about and how they interact with their political systems. I investigate how patterns of discussion and discussion’s impacts vary across the world, paying particular attention to Brazil. Data come from panel studies spanning Brazil’s 2002, 2006, and 2010 presidential elections, as well as a case study of the 2008 local elections; and from an eleven country study in the 1990s. I address three broad research questions. First, does political discussion affect democratic competence, and for whom? While the claim that political discussion has democratic benefits is common, selection effects make demonstrating causal claims difficult, since politically knowledgeable and engaged citizens are likely to choose to discuss politics. Using fixed effects and instrumental variables models, I find strong evidence that conversation promotes knowledge and participation, and that it has a “leveling effect,” helping citizens with lowest initial knowledge catch up with their neighbors. Moreover, spouses are particularly influential, and women give higher priority to spouses as their closest political discussants. Second, how does knowing people with different political opinions affect democratic competence? The key to solving longstanding debates requires recognizing that divergent preferences take two forms—the total preferences in the network (diversity) and the extent of disagreement with the reference person (conflict). Using multilevel models, I find that in systems with low numbers of candidates, conflict is demobilizing, but only when the network homogeneously disagrees with the reference person. Moreover, conflict combined with diversity promotes learning. In systems with more candidates, however, the effects of conflict disappear. Third, how do the electoral and party systems shape networks? And what are the downstream consequences for democratic competence? The number of candidates in a political system strongly affects exposure to diverse and conflicting preferences as well as the probability of knowing candidates and activists. I estimate that three-quarters of respondents in the local election I study in Brazil knew personally a candidate; using matching, I find that such connections promoted political engagement, but also fostered clientelism