Divisive primaries have the ability to reduce a party’s probability of winning by an average of 21 percentage points in the general election.

The political climate in the United States has become vastly polarized over the last decade, with candidate ideology stretching all across the spectrum. Information technology and intra-party polarization have become much more prevalent, yielding highly contested and divisive primary races. But what effect does this have, if any, on political parties and election turnout? 

A paper by Assistant Professor Alexander Fouirnaies of the University of Chicago Harris School of Public Policy, along with Professor Andrew B. Hall of Stanford University, concludes that divisive and prolonged primaries do in fact have a negative effect on political parties in the general election in most cases.

“As the Democratic candidates attack one another on their way to the convention, they may want to ask themselves if their rhetoric is helping re-elect President Trump, who has no real opposition in the Republican primary,” said Fouirnaies. “Our paper suggests that a bitter, divisive primary could spell serious trouble for the party that claims to be focused on beating Trump.”

Former Vice President Joe Biden, one of the two frontrunners for the Democratic nomination in 2020.

Comparing data from runoff elections over the past several decades, findings suggest that divisive primaries may affect vote share in the general election by 6-9 percentage points, thus decreasing the party’s probability of winning by 21 percentage points.

The paper finds that those negative effects are most evident in campaigns that receive the most attention from the population. In other words, the negative effect on U.S. Senate elections is higher than the effect on U.S. House elections, for example.   

The study finds, however, that in state legislative races, prolonged and divisive primaries have a null effect – if not a slightly positive one. A positive effect of divisive primaries may exist because a more divisive campaign, for a race that does not typically capture the public’s attention, may make voters pay attention to an election that they were not previously attuned to. 

These contrasting trends are attributed to the varying flow of information, due to how much voters, the media, donors, and others pay attention to different races, something the authors call the “salience” of the campaign. Campaigns that people pay more attention to are of higher salience, and naturally expose the voter to a higher influx of information on candidates than campaigns of lower salience. When a race is prolonged or especially competitive, candidates in races of higher salience may resort to shedding negative light on one another in order to draw a contrast between themselves.

Senator Bernie Sanders, one of the two frontrunners for the Democratic nomination in 2020.

In races of lower salience, however, candidates strive for opportunities to provide more information about themselves since it is more difficult to bring attention to their candidacy and their race. A divisive race for the state legislature, for instance, may be the only reason voters, donors, and the media focus on that election at all, increasing interest and turnout.

The decreased vote share in the general election is a substantial price that parties pay in order to maintain a democratic and transparent process of nominee selection. For this reason, the authors infer that “parties in high salience contexts have a strong incentive to avoid publicly visible conflict among potential nominees.”

“This is the message that Vice President Biden, Senator Sanders, and Senator Warren must take to heart if they are serious about beating Trump: drawing distinctions among themselves without creating undue conflict that benefits the Republicans,” added Fouirnaies.

At a time where intra-party division seems to be at a high, in both political parties, further research could be beneficial in assisting parties to better strategize in future elections.

For this study, the authors examined data from states that hold legislative runoff primaries – including Alabama, Georgia, Oklahoma, North Carolina, Texas, Florida, Arkansas, Mississippi, and South Carolina – which typically occur when one candidate does not receive a certain threshold in the first election.

The research was featured in the first episode of Not Another Politics Podcast, hosted by three Harris Public Policy professors.