Nonvoting and the decisiveness of electoral outcomes

The authors reexamine the relationship between closeness and turnout by looking at aggregate-level variables and studying fifty-seven elections. They also attempt to estimate how much voter turnout is likely to change as a function of change in the closeness of elections. Specifically, they hypothesize that in first-past-the-post elections, turnout will be affected by the voters' expectation of how close the elections are going to be, by the turnout in the previous election, and by the closeness of the previous election. The authors test their hypotheses using turnout and closeness data for all fourteen general elections held in the United Kingdom since 1955 and all forty-three congressional elections held in the United States between 1920 and 2004. Their findings strongly support their hypotheses.