Rogers et al., 2017: Encouraging voting with social pressure increased voter turnout among low-to-medium propensity voters in Wisconsin
Extending Gerber et al., 2008, 664,020 registered voters in Wisconsin identified as likely supporters of the Democratic candidate Tony Barrett in the June 2012 recall election of Governor Scott Walker in Wisconsin (a high-profile election) received a mailer in the final week of the campaign (1) urging them to vote, (2) emphasizing that voting records are public (e.g., “After the June 5th election, public records will tell everyone who voted and who didn’t”), (3) presenting voters their own turnout record in the previous two federal elections, (4) presenting the turnout records for a dozen neighbors assigned to the treatment group, and (5) including a blank entry for the upcoming election. A small random sample received no mailer. Among those receiving mailers, there was a 2x2 between-subjects design: Mailers were either nonpartisan or partisan (opposing Scott Walker) and either did or did not encourage people to ask their neighbors to vote. As compared to the no-mailer control, all four treatments increased turnout (66.42% versus 65.38%), with no difference among the mailer conditions. The greatest treatment effect was observed among voters with low to medium propensity to vote (0-20% propensity: 0.91% effect; 20-40% propensity: 3.28% effect and 40-60% propensity: 3.63% effect versus 60-80% and 80-100% propensity: <0.5% effect). The mailer treatment also raised turnout in the November general election by 0.41%.