Pollsters survey public opinion in an attempt to predict what the voting public will do when it comes to filling out actual ballots. Only after the polls close and votes are tallied is it possible to tell how accurate predictions were. But what if, as complaints about inaccurate polling implicitly suggest, voter decisions to vote at all are affected not only by the usual suspects—information, education, preferences, political awareness, etc.—but also by how poll results affect their beliefs about outcomes? In this paper we examine how election-as-horse-race information from opinion surveys should affect who chooses to turn out and who chooses to stay home rather than vote for different views of why people vote in the first place. We test four literature-based hypotheses about the the turnout decision using polling and election data from the United States.
