Persuasion,surveillance, and voting behavior

More than 4000 students received one of three letters urging them to vote in a campus primary election. Two letters suggesting that voting behavior would be monitored produced 37% turnout compared with 32% for a nonsurveillance letter and 26% for noletter controls. Differences among these three groups are significant at p < .01. The letters also increased turnout in a runoff election held one week after the primary; however there were no differences among the three letters. These results are discussed in terms of Thibaut and Kelley's three-step process for enforcing group norms.