Learning to Vote: the Effect of Local Context on Individual Turnout

Abstract: How does context influence individual voting behavior “in the wild”? Experimental field studies have shown that social factors such as peer pressure or the prospect of shame powerfully influence voting behavior. But the degree to which typical citizens are exposed to such pressures in practice remains an open question, as people select the company they keep and the news they read. We use panel data for twelve million Mexican voters to study the effect of local context on individual voting behavior. We exploit observational variation in context induced by citizens who move from one locality to another between the 2012 and 2015 federal elections. We find that differences in average turnout between the origin and destination localities substantially influence the movers’ probability of turning out in the election subsequent to moving. We deploy a variety of empirical strategies to isolate causation, including stratification by locality or by city bloc of origin or destination. The effects persist in subgroup analysis by age, sex, region, voting history, and distance from home to voting booth. We also find spillover effects on household members who did not move.