State-dependent effect on voter turnout: The case of US House elections

Abstract

In models of voter participation, the effects of election margin and campaign expenditure can be shown to be state-dependent – varying with low/high turnout. We empirically assess these implications for observed turnout, employing data from US House elections from 2000 to 2008 by means of quantile regression analysis. We document that the effects of expected election margin and campaign spending on turnout are state-dependent: the later is positive and decreasing, whereas the former is negative and U-shaped. Other determinants’ influence on turnout (e.g. education, population density) is also shown to vary across the conditional distribution of turnout rate. Our findings are robust to a number of extensions.