Incumbency status and candidate responsiveness to voters in two-stage elections beginning with a primary

ABSTRACT

Theories of representation suggest that candidates should respond ideologically to their constituency. Two-stage elections like those in the U.S. force candidates to decide which parts of their constituency they should respond to: citizens who are active enough to participate in primaries or those who only participate in general elections. We posit that non-incumbent candidates should mostly focus on the preferences of primary voters while incumbents should be largely unmoved by the preferences of either set of voters. We test these expectations using data from U.S. House and Senate contests and find support for our theory. Our results suggest that scholars should pay closer attention to the two-stage nature of U.S. elections when evaluating electoral responsiveness.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 There are of course exceptions, including Louisiana's open primary where if one candidate receives more than 50% of the vote in the primary there is no general election, the newer top-two primaries in California and Washington where an open primary might produce two candidates for the general election of the same party, and Virginia's frequent use of nominating conventions for statewide offices rather than primary elections.

2 In another study, Banda and Carsey (2015) observe how candidates engaged in two-stage elections respond to each other's advertising strategies.

3 We are unfortunately unable to expand our analyses beyond 2010 because there does not appear to exist sufficient public opinion data that includes an indicator for whether or not respondents cast a ballot in non-presidential primary elections.

4 Numerous scholars argue that individual candidates perform a similar calculation when they decide whether or not to run (e.g. Maestas et al. 2006; Carsey and Berry 2014).

5 There remain important debates revolving around proximity (e.g. Downs 1957) and directional (see Rabinowitz and Macdonald 1989; Kropko and Banda 2018) models of voting that differ sharply on the decision-making tendencies of the median voter, but those debates do not change the expectation that if one constituency is more liberal compared to another, that the distribution of candidates from the first constituency would also be expected to be more liberal.

6 Major party candidates will generally ignore the opposing party's primary voting constituency because citizens who participate in these contests are unlikely to support their candidacies during general elections.

7 Using contributions to measure candidate ideology assumes that interest groups and Political Action Committees seek to support candidates who share their ideological positions – an assumption that is fairly non-controversial in the field (Bonica 2014). Some groups donate merely to gain access to elected officials Kollman (1997). However, recent evidence suggests that there is next to no direct significant effect of donations as a form of investment (McCarty and Rothenberg 1996; Ensley 2009; Bonica 2014) and that donations are a means to express ideological sincerity (Ansolabehere, de Figueiredo, and Snyder 2003; Bonica 2014). The CF scores we use do not control explicitly for committee positions as the item response theory (IRT) version of the scores do (Bonica 2013), but the results between IRT and common space measures do not meaningfully differ (Bonica 2014). Therefore, the scores either capture the sincere ideologies of interest groups or constitute costly signals (Austen-Smith 1995), both of which capture stable ideologies of office holders and interest groups alike.

8 In these data, about 81% of Democratic primary voters also voted in the general election while nearly 89% of Republican primary voters also participated in the general election.

9 We describe these data and the ideology measures in greater detail below.

10 We used the probability weights included in the CCES when we aggregated these data.

11 This decision does not affect the substance of our findings. Interested readers can see these results in the supplemental appendix.

12 The DIME data utilized an automated coding scheme based on candidates' names to determine their sex. We coded by hand the few candidates whose sexes were not included in the data. We also double checked candidates with ambiguous names.

13 To simplify producing Figure 3, we set all of the remaining covariates in the model to their median or modal – as appropriate – values.

14 There is one exception to this across these additional tests: non-incumbents running in U.S. Senate election appear to shift their ideological positions in the direction preferred by general election-only voters.

15 Though we do not attempt to observe the effects of the preferences of non-voters on candidate ideology, it seems unlikely that candidates have much of an incentive to respond to them.

Additional information

Notes on contributors

Kevin K. Banda

Kevin K. Banda is an assistant professor of political science at Texas Tech University. He studies campaigns, public opinion, political behavior, political communication, and state politics in the U.S.

Thomas M. Carsey

Thomas M. Carsey was the Thomas J. Pearsall Distinguished Professor in the Department of Political Science at the University of North Carolina at Chapel Hill. He studied electoral behavior, campaigns, political parties, public opinion, state politics, and legislative politics in the U.S.

John Curiel

John Curiel is a Ph.D. candidate at the University of North Carolina at Chapel Hill. He studies state politics, representation, electoral systems, and political institutions in the U.S.