How Do General Election Incentives Affect the Visible and Invisible Primary?

• Published date: 01 November 2023
• Author: Mackenzie Lockhart,Seth J. Hill
• Date: 01 November 2023
• DOI: http://doi.org/10.1111/lsq.12415

833

LEGISLATIVE STUDIES QUARTERLY, 48, 4, November 2023

DOI: 10.1111/lsq.12415

MACKENZIE LOCKHART

University of California San Diego

SETH J. HILL

University of California San Diego

How Do General Election Incentives

Affect the Visible and Invisible

Primary?

Previous research finds that nominating more centrist candidates in-

creases vote share and win probability in congressional general elections. Yet

party primary elections often nominate non- centrist candidates, increasing

polarization between the American parties. We develop a model of choice in

nomination politics that shows when and how actors respond to incentives of

the general election. We then combine 200 million contribution records with

data on 22,400 candidates in 7100 House primary elections from 1980 through

2016. We find that potential candidates and primary voters respond to general

election incentives but do not find clear evidence for contributors or the win-

nowing process. Connecting these results back to our model, this implies that

actors in the invisible primary either place higher value on in- party candidate

ideology or have different beliefs about the general election than do primary

voters. Our evidence adds to a body of research that suggests primary voters

are a larger moderating force than elites in American party politics.

Conventional wisdom holds that primary elections push

the Republican party to the ideological right and the Democratic

party to the ideological left. But political science evidence shows

that extreme candidates are less successful in general elections (e.g.,

Ansolabehere, Snyder, and Stewart 2001; Canes- Wrone, Brady,

and Cogan2002; Hall 2015).1 A belief in returns- to- moderation

at the general election means that actors in nomination contests

face a tradeoff. They must decide whether to support a less cen-

trist candidate and incur an increased risk that the other party’s

candidate wins the general election or support a more centrist can-

didate to increase their chances of winning the general election

but at the cost of less preferred policy views (e.g., Aranson and

Ordeshook1972; Coleman1971, 1972).

© 2023 Washington University in St. Louis.

834 Mackenzie Lockhart and Seth J. Hill

Some political elites, however, have expressed skepticism that

there are returns- to- moderation at the general election. In the 2020

Democratic presidential primary, candidate Pete Buttigieg said,

“It’s true that if we embrace a far left agenda, they’re going to

say we’re a bunch of crazy socialists. If we embrace a conserva-

tive agenda, you know what they’re going to do? They’re going to

say we’re a bunch of crazy socialists. Let’s stand up for the right

policy, go up there and defend it.” (Basu 2019). Evidence from

Broockman et al.(2021) suggests local party officials agree with

Buttigieg. Local political elites they surveyed reported there is no

trade- off between ideological purity and chances at the general

election.

The choices made by political actors across thousands of

primary elections determine the candidates nominated for office

and, subsequently, the extent of political polarization between the

parties in Congress and state legislatures. How much more these

actors value extreme candidates over moderates and what they be-

lieve about the relative chances of each at the general election de-

termines which candidate they support. While political scientists

have offered evidence that actors in primary elections prefer more

extreme candidates (e.g., Brady, Han, and Pope2007; Hill2015;

Hill and Tausanovitch2018; Rogowski and Langella,2015), we

have to date much less evidence on the beliefs of primary actors

about relative chances at the general election. It is the combination

of the two that determines political choices.

In this article we study how beliefs about returns- to-

moderation at the general election influence the behavior of po-

litical actors in nomination politics. We start with a model of the

choice faced by potential candidates, contributors, activists, and

primary voters when deciding whom to nominate to face a known

opponent of the other party. While most models of nomination

politics consider two parties making nomination choices concur-

rently in anticipation of the general election, we model a setting

where one party knows their general election opponent during the

nomination contest. In practice, the vast majority of US House

districts have one candidate (the incumbent) known with near

certainty. In the 2020 House elections, for example, only eight of

399 incumbents who ran for reelection were defeated in primaries,

meaning that in 98% of cases with a sitting incumbent, the incum-

bent made it to the general election ballot.2 Therefore actors in

non- open- seat out- party primaries know with near certainty who

they will face in the general election. Out- party actors must decide

835How General Elections Affect Primary Elections

which candidate to support in anticipation of facing that incum-

bent at the general.

Our model shows that the general election has a moderating

effect on primary elections only under certain combinations of be-

liefs and preferences. A moderate candidate in a primary election

gains support only if the primary voter, contributor, or activist

believes the return- to- moderation is large, if their preference for

an extreme nominee over a moderate nominee is not too great, or

some combination of the two.

In other words, the response of actors in primary elections to

a more moderate or extreme opponent is ambiguous. While some

political actors might always support an extremist or a moder-

ate regardless of the opponent, the support of others might be

contingent.

Political actors whose support in the primary is contingent,

however, do not all follow the same strategy. Our model shows that

while some choose to match the extremity of their opponent—

supporting a moderate to face a moderate opponent and an ex-

tremist to face an extremist opponent— others choose to zag the

extremity of their opponent— supporting an extremist against a

moderate and a moderate against an extremist.

Actors who match the extremity of their opponent do so be-

cause, even though they would prefer that an extremist of their

party represented them, an extremist’s lower chances of winning

the general election against a moderate opponent lead them to

nominate a moderate to improve their chances at the general.

When their opponent is an extremist, however, they believe they

have a higher chance of winning the general election with their

extremist and so are willing to take the gamble of nominating an

extremist from their party.

Actors who zag the extremity of their opponent do so be-

cause, even though their extreme nominee is less likely to win the

general election against a moderate opponent, they are willing to

take the gamble on the extremist. If they win the general election,

they get their more preferred representative. If they lose the gen-

eral, the winner is a moderate from the other party. But if the in-

cumbent is an extremist, the disutility of the incumbent winning

reelection is large enough to induce actors who zag to support a

moderate from their party in the primary.

Thus, our model shows that there can be a paradox of

moderation. Even with a general electorate that prefers centrist

candidates and actors in primary politics who know the general