Analyzing the turnout-competition link with aggregate cross-sectional data

Numerous papers by Public Choice oriented scholars and others have sought to test the hypothesis inspired by Downs (1957) that, ceteris paribus, turnout should be higher when elections are close. Most look in cross-sectional terms at variations in turnout at the constituency level for elections of a given type. By and large the results have been disappointing (see, e.g., Foster, 1984). We are skeptical of these weak findings and argue that a complete portrait of the turnout-competition link requires us to examine that link for at least three different types of turnout (turnout among potential eligibles, turnout among registrants, and turnout for a given office relative to other offices such as top of the ticket), and to take into account longitudinal changes in turnout. For recent off-year elections to the U.S. Senate and also for off-year elections to the U.S. House of Representatives in states where there was no gubernatorial or senatorial contest on the ballot, we find strong evidence for higher turnout among eligibles in close contests. For these elections to the U.S. Senate we also find evidence for an ecological effect that leads to turnout being maximized at values of competition other than 50–50. Moreover, when we look at turnout for office relative to top of the ticket voting in the California Assembly and the U.S. Senate in presidential years, we again find some evidence for an ecological effect in which turnout is maximized at a value of Republican vote share above 50 percent and find further strong evidence for a clear link between competition and turnout for office among those at the polls.