Reevaluating Competition and Turnout in U.S. House Elections

Does electoral competitiveness boost turnout in U.S. House elections? Using an individual panel of turnout records compiled from the voter files of all 50 states, we exploit variation in district competitiveness induced by the 2012 redistricting cycle to provide credible estimates of the effect of competitiveness on turnout. When tracking the same voters across time under differing levels of competitiveness, we precisely estimate the effect on turnout to be near zero. Although past cross-sectional research reports a link between competitiveness and turnout in House elections, we demonstrate that residents in competitive districts differ markedly from those in uncompetitive districts along a number of observable characteristics correlated with turnout, and we argue that this induces bias in most cross-sectional estimates. Secondary evidence tracking voter perceptions of competitiveness and campaign behavior provides support for our finding. Voters have scant awareness of competitiveness in House elections, and, while campaign spending is strongly related to competitiveness, it is directed into avenues that do not appreciably increase turnout. Our findings have important implications for the competitivenessturnout relationship in other electoral settings with geographically compact, single-member districts. ⇤Ph.D. Candidate. danielmoskowitz@fas.harvard.edu. †Assistant Professor. benjamin_schneer@hks.harvard.edu. Participation in elections is considered a primary indicator of democratic performance (Powell 1982). In a given election, the level of voter turnout has implications for whose views are represented (Fowler 2013), which party wins and retains office (Nagel and McNulty 1996), and even future levels of participation (Meredith 2009). Given the importance of voter participation, political scientists have devoted considerable attention to what factors lead to higher turnout, and an extensive body of research points to electoral competitiveness as a primary determinant (Blais 2006; Geys 2006). Enos and Fowler (2014) note that of 70 papers examining turnout published in top political science journals since 1980, 41 mention the importance of closeness or competitiveness.1 While scholars disagree on the primary mechanism, the implication in this line of research is that competitive elections cause more citizens to vote. In this paper, we explore the link between competitiveness and turnout in one of the most studied electoral settings: the U.S. House of Representatives. Because the redistricting process offers policy makers a mechanism to alter the competitiveness of House elections, this electoral context is unique and of particular interest. Indeed, citizens’ groups and lawmakers have argued that redistricting should be used to draw more competitive districts with the express goal of increasing turnout. A recent report by Common Cause noted, for example, that “competition pushes candidates to work harder to connect with voters, boosting turnout and strengthening democracy” (Vicuna, Morris, and Eisman 2017). In a Washington Post op-ed entitled, “Voting’s Neglected Scandal,” David Broder, drawing on a study by the Democratic Leadership Council, noted that “if district lines were redrawn to emphasize competitiveness,” then “11 million more Americans might show up at the polls, decreasing our chronically low voting participation rates” (Broder 2008). We demonstrate that, in the context of recent U.S. House elections, the causal relationship between electoral competitiveness and turnout is very tenuous — contrary to the received wisdom among political scientists, policy makers, and journalists. Only in a small subset of races — for example, when the House race is on the top of the ticket — do we find that competitiveness increases turnout, and even in this case the substantive size of the effect is minimal. 1We use the term “competitiveness” to indicate the underlying propensity for a close election outcome (ex ante), while the term “closeness” indicates a close election outcome (ex post).