Do Divisive Primaries Actually Affect General Elections?

Do divisive primaries actually affect how candidates will perform in general elections? It's a question political scientist have been trying and failing to untangle, but we found someone who may have an answer.

With the 2020 Democratic primary getting into full swing, we're kicking off our inaugural episode with Assistant Professor Alexander Fouirnaies, whose research gives us new insights into the effects of divisive primaries and what we can expect from the 2020 Presidential election.

Listen on Apple Podcasts or wherever you enjoy your podcasts.

Transcript:

William Howell:

I'm Will Howell.

Wioletta Dziuda:

I'm Wioletta Dziuda.

Anthony Fowler:

I'm Anthony Fowler.

William Howell:

And we're three professors from the Harris School of Public Policy at the University of Chicago.

Anthony Fowler:

And this is Not Another Politics Podcast.

Tape:

Everyone agrees American politics has become more chaotic in recent years, that it changed and not always for the better.

William Howell:

Why is it not another political podcast?

Anthony Fowler:

I would say we're trying to do a serious political science podcast. We are trying to delve into the science of the political process.

Wioletta Dziuda:

So each week we are going to read a piece of research, interview the author, and try to understand as a group what this research tells us about the headlines.

Anthony Fowler:

And when they say something that we think is BS, we're going to say so and we're going to talk about it.

William Howell:

And I think it's fair to say that what animates the three of us is a dissatisfaction with the way in which politics is talked about more broadly. Standards of evidence aren't taken seriously. Theory decidedly is not taken seriously, and there's a lot of shooting from the hip.

Tape:

Well, clearly the mainstream media has an agenda here, and their agenda is to hurt the president of the United States.

William Howell:

And headline chasing, and overemphasis of the latest phenomenon.

Tape:

Bears repeating. This is certainly not normal, except abnormal is kind of now normal.

William Howell:

And we want to try to cut through that thicket in order to speak to kind of the more foundational forces that define and propel our politics both here in the United States and abroad, no?

Anthony Fowler:

Plus my mother can learn what I do for a living. She usually wonders why I'm not a pundit on MSNBC or Fox News or something.

William Howell:

This is your big chance.

Anthony Fowler:

This is my chance.

We are in the middle of a fairly interesting, potentially contentious presidential primary right now and so for our very first episode, this seems like an interesting topic to take on. What happens when there are these divisive primaries? Is that a good thing for the party? Is that a bad thing for the party?

William Howell:

It's a question that doesn't have an obvious answer, because there are lots of different stories that one could tell. On the one side, one might say that amidst a really contentious primary, what you see are people marring one another, taking down one another and compromising one another in a ways that whoever is the eventual kind of victor ends up limping into the general election.

On the other hand, what you might say is that, well, what we might call a divisive primary is one that excites lots of people, generates lots of interest, forces candidates to sharpen their argument and raise their game on a stage that very few of them have actually participated in politics.

Wioletta Dziuda:

I think it's even more complicated than what you are saying.

William Howell:

Exactly, yeah.

Wioletta Dziuda:

Because it's also important to understand why we have a divisive primary. Is there something happening in the political system that makes politicians being very divisive? I agree with you that it's a really complicated issue.

William Howell:

But Anthony, you found a terrific paper by a couple of terrific scholars that grapples with this.

Anthony Fowler:

I did, yeah. I talked to our colleague Alex Fouirnaies about his paper with Andy Hall at Stanford. They have written a paper that I think is perhaps the best paper available to us on this topic.

William Howell:

So, we're going to go listen to that interview that you conducted with Alex, Anthony, and then we're going to get back together, the three of us, and reflect upon the paper and what we can learn from it.

Anthony Fowler:

Thank you Alex, for talking to me today.

Alex Fouirnaies:

Thank you.

Anthony Fowler:

Before you guys wrote your paper, what did we know about divisive primaries, and what made you interested in the topic to begin with?

Alex Fouirnaies:

It was clear that there was this hypothesis out there that if you have a hard in-fight within a party to win the party's nomination, then that could potentially harm you and the party, in the general election. We decided to look at the empirical literature on this, and to our surprise, we found that the results in the empirical literature is very mixed.

So typically the way that people have studied it is to look at the association between the number of candidates that enter in the primary election, and then the performance of the party in the general election.

Anthony Fowler:

And so of course, you can imagine all kinds of biases associated with those designs. In fact, I mean, theoretically you would expect that, for example, when the party has a better chance of winning the general election, more people should want to run in the primary to begin with because, you know...

Alex Fouirnaies:

Yes, exactly. So that's the key. That's the key problem. Like the places and the times where you observe a very divisive primary, that's not going to be at random places and random times. So what we decided to do was to exploit the institution of run-off elections. So in some states you have this institution that if the front runner in the primary election doesn't get more than some threshold, typically 50%, there's a second primary election, essentially.

So what we thought was, well, in a way, that's kind of like you're intensifying the primary campaign, and you're prolonging the primary campaign. So, that at least captures some aspects of what we think about when we think about what a divisive primary is. If you have these run-off elections, you can essentially compare places where you marginally avoid a run-off to places where you are… marginally are forced into a run-off.

And so the idea is that if we compare the performance of the parties that are just above and just below this vote share threshold, then we can use that to estimate the causal effect. It's like doing a randomized experiment, at least a local randomized experiment here.

Anthony Fowler:

Okay. So let's talk about the results. So, what do you actually find? Are divisive primaries good or bad for the party?

Alex Fouirnaies:

Yeah. So, if we start with the results at the federal level, we find that it on average harms the party in the general election. We observe a drop in the vote share of six to nine percentage points. So it's quite a substantial harm to the party potentially.

Anthony Fowler:

And then what do you find in the state legislative races?

Alex Fouirnaies:

Yeah, so interestingly, at the state level, we find basically nothing. Either nothing or if anything, the effect seems to go in the other direction. Which we find is like, that's potentially an interesting finding that can help us shed some light on what's going on. So we have the federal level where we have campaigns that in general tend to be high salience, and there's a lot of information. But if you think about the news coverage of their state representative election, that's not clear that they get nearly as much coverage.

And maybe that can help us understand a little bit of what, what's going on here. Maybe there's something about the information that's going on. Like, maybe people didn't even want to show up to this primary election, but now that you hear that there's actually something at stake, you get more information about the candidates, maybe that induces some people to show up and support the party.

Anthony Fowler:

Sure. Yeah, that's kind of fascinating. So, yeah. So if you were going to extrapolate a little bit, where of course presidential primaries are an incredibly high salience setting, you might argue that this is perhaps really detrimental, even more detrimental for the party, than you observe in House and Senate races.

Alex Fouirnaies:

Yes. If we believe that the pattern would be the same, and this is about salience or information, and then we have the low salience, low information elections, that's the state-level elections. Then we have the federal elections, congressional elections, that's the more high salience, and then we have the presidential elections. That's super high salience, and a lot of information. Then we might think that it actually could potentially hurt the party even more at presidential level.

Anthony Fowler:

So, let's talk about some other potential explanations. So you've given one explanation as to what might be going on, that there's something just about the sheer amount of information. But let's set that aside. What else could potentially be going on here? What other stories did you consider that might explain why divisive primaries seem to, on average, be bad for the party? And maybe why they're especially bad in high salience settings.

Alex Fouirnaies:

So, one story could be about collecting money. One story could be campaign finance, and resources more generally. If you reach out to donors and ask for money to win the party nomination, maybe there's a limit to whether you can go back and ask the same donors once again to support you in the general election. So if you spend all the money in the primary campaign, then you might not have enough money to actually spend in the general election.

And the same thing, you could think about that with other resources as well. So imagine you're organizing get out the vote campaigns. You want people to knock on doors on your behalf. And maybe you can only ask people to do that once, or... It's more difficult for sure, and more costly the second time around, to convince people again to go out and do it one more time for you. So if that's the case, maybe that could explain the results.

William Howell:

So Anthony, before we went into the interview, you said that you thought that this was the best piece of research currently available to us about this question about the causal effect of divisive primaries on the electoral fortunes of the winner in the general election. The thing that they spend most of their time focusing on is this general salience, but this is a general... I mean, this is more pointing to these two things. We've got salience. We've got money. There's also the adaptive response of the media, the adaptive response of the party, right? Patterns of endorsements.

Wioletta Dziuda:

The selection effect.

William Howell:

The selection effect. What we're getting from the paper is sort of an average of all these things. That's the best that they can do, which is no small thing.

Wioletta Dziuda:

This is where I think theory's really useful, given that you're not constrained by the data. I like the paper a lot, but I was very surprised by the effect that they find, because if we take this information, our story seriously, that somehow there's more information revealed during the run-off, and on average leading to negative outcomes for the party. You would expect that yes, sometimes, we are going to learn negative things about those candidates. They are going to bring up some dirt on each other.

But sometimes we should see positive things coming up. We learn that actually those candidates are very good at debating, or they have a lot of policy ideas. So, yes, sometimes we should see that the party loses. Sometimes we should see that the party gains. But on average we should see that the party remains on average.

And if anything, the voters will be more knowledgeable about the candidates who went through the run-off, so they should face less uncertainty about them. They should be more willing to put them in office as opposed to the candidates from the other party that hasn't been vetted so much. So I was very surprised by the story. That's why I like the story about financing a little bit more, because that really seems to be resonating. But I was puzzled by this difference between states and congressional elections. So that's why I ask you why? You know, can we actually reconcile this differential effect?

Anthony Fowler:

So Bayes' rule tells us that on average, our average beliefs shouldn't change in response to more information. Sometimes we get good news, sometimes we get bad news. So we're going to get more information about these candidates. Let's say there's a front runner, and that front runner isn't going to change, which is actually true in almost all of their cases. So you don't actually change the winner of the primary very often, but you change the length of the campaign. Maybe you're getting more information about this leader.

So Wioletta's point is, on average, our beliefs about this person shouldn't be changing. Sometimes they're going up, sometimes they're going down. Should that mean that on average, there's no effect of the prolonged primary? That's one theoretical prediction. But Bayes' rule doesn't say that there should be no average effect on votes. It just says average beliefs shouldn't change. Suppose more often than not, you get one of these close competitive primaries in places where that's the dominant party, and so all else equal, our beliefs are already above the threshold for this candidate. We're probably already going to vote for him. Now, it could be the case that even though on average our beliefs aren't changing, our beliefs are more likely to be dropped down below this threshold and we're voting the guy out.

William Howell:

But what we haven't said much about, which I think we should try to get our heads around, is the magnitude of these effects. They aren't just negative. They're massive. It's a 20 to 26 percentage point decrease in the probability of winning, and a six to nine percentage point decrease in the vote share. Right?

Wioletta Dziuda:

Right.

William Howell:

And the size of those effects are comparable to things like major scandals happening in the context of a campaign. Which strikes me as, I don't want to say implausible, but it to my mind raises questions about what are we actually tapping into? Because the effect is real. So what is the effect speaking to? Is it really just about divisiveness? Could we say that this, a few more weeks of campaign, can it have that kind of an effect? That's profound. I mean, you think about the amount of money that campaigns spend on advertising and the returns from it. This dwarfs the kinds of effects that we see there.

Anthony Fowler:

It's worth pointing out, these effects are comparable to effects that Andy Hall, in other work, identifies as the effects of nominating a relative extremist. So, when a party nominates a relative extremist versus a relative moderate, the party does much worse in the general election. And there you could ask them, why haven't they gotten together and just nominated relatively moderate candidates to make sure they win the general election?

And I think you have some of the same questions, which is: one, the party is not a unitary actor. Two, the kinds of people who get involved in politics are ideologues themselves. They got into politics because they feel really passionately about free college and a $20 minimum wage, or no immigrants coming into the country, or... You know, think of the kinds of things that those people are very passionate about. So to me it's not surprising that they haven't been able to get together and figure out, "Okay, let's do this to maximize our chances of winning." Because they, in fact, are ideologues, and they in fact can't coordinate with each other.

Wioletta Dziuda:

I know, but you would think that it works both ways. If run-offs, if the prolonged fights, actually decrease your chances of winning, then parties could scale down the amount of presence that they have in the media. They would actually try to minimize the number of interviews and debates, and this is not what we seem to be observing. We seem to be observing that actually, the candidates are seeking attention. They are trying to be as much as possible in the media. The parties are pushing for that.

So, I think in light of all that, and in light also of the 2016 election, where the candidate that actually got mostly negative coverage, but a lot of coverage, won. I do find those effects puzzling.

William Howell:

One might think about the follow-on study that one would like to conduct, building off of what's been done here, what that might look like. So, some obvious things are, collect data on other races that vary along the dimensions that we've been talking about. That would give us additional leverage to speak to what this average effect is actually constituting.

A piece that I would love to see some evidence on but I don't know quite how to do it, would be the effect of primaries on candidate skills, quality, energy. This is a story that was told about Barack Obama, and why he benefited in 2008 from the prolonged race against Hillary.

Tape (Hillary Clinton):

It is sometimes difficult to understand what Senator Obama has said, because as soon as he is confronted on it, he says that's not what he meant.

William Howell:

He sharpened his game.

Tape (Barack Obama):

Given the magnitude of our challenges when it comes to energy and healthcare and jobs and our foreign policy, you'd think that we'd be having a serious debate.

William Howell:

He emerged a more formidable candidate. That isn't in-keeping with the findings that are on offer here, because you see again, on net it appears to be negative, if we want to extrapolate from these congressional elections to a presidential election. And they offer, in the paper, some reasons why we might want to do that. But it doesn't negate the possibility that some of these local effects are actually positive and that this is a space in which... No, no, no. Like, going after this for a longer period of time, you become more formidable. I don't know how we could pin that down or back out that particular quality, but I'd like to see it.

Anthony Fowler:

Sure, sure, sure. I mean, there are lots of other countries where they'll have experts rate the debates and say who won the debate and who performed the best, and so forth. You could imagine doing that in these races. If they're, you know, general election debates, at the end. And say, "Did the party that went through the run-off have a better debate performance, or something like that?"

William Howell:

Well, you look at the final debates on each, and you find some way of measuring the difference.

Anthony Fowler:

So, there's still the open question. It could be that you're better by the end because you've gone through more. You're a more practiced candidate. But you've also had more opportunities to embarrass yourself. So, even that gaff during the run-off. Even if you end up winning the run-off, it could still hurt you in the... While you were learning and you were still an inexperienced candidate then, your opponent can use that against you.

William Howell:

So this is the thing that's been carrying through our conversation here, which is again, the difference between where you end up and the accumulation of discord or gaps. And what is relevant when you move into a general campaign? Do we look back at... In the general campaign, is what really matters is all the bad things that were said in January and February? Or is what really matters the candidate who emerges at the very end? And I don't think we have clear answer to this.

William Howell:

But this paper that's before us is certainly in-keeping with the story that says it's about the accumulation of discord and division that–

Anthony Fowler:

If you had, to guess there's got to be some combination of both. Right? If you in fact make a big error, even in early on in the primary–

William Howell:

For sure.

Anthony Fowler:

Your opponent can always use that against you.

William Howell:

Can always access it. That's right. But the other story is to say, well. Better it be aired then. It's going to come out eventually, and then that's a net positive you get. This is another one of these just so stories, is that we want to find out about all the scandalous behavior months and months before the general election so that we can take the sting out of it, and so that the candidates get smarter and better and more adept at speaking about it.

Wioletta Dziuda:

And I think those two stories, even though they seem to be pointing in the opposite directions, actually they point in the same direction. So it might be the case that certain candidates hone their skills and they become better candidates, and then later become better representatives, presidents. It might be that some other candidates don't perform so well. They stumble and they end up losing the election. Perhaps for the voter, both those effects are good. We may want to figure out which candidates actually stumble. We may want to weed them out, because they might not be the best representatives for us, and we may want to also have those candidates that perform super well and end up being better candidates as a result of long election cycles.

So, both effects, even though they lead to different outcomes for those individuals. So when we just run the regression and we see a zero average effect, they point in the same direction, that they might be good for the voters.

Anthony Fowler:

It seems hard to construct a compelling argument as to why more information and more campaigning is in fact bad for the voters, or bad for the electoral process as a whole. It may be bad for the party, but when one would think, one would think that more campaigning, more media coverage, all of those things should, on net, allow voters to make better decisions.

Should we finish with some speculation about the 2020 presidential race?

Wioletta Dziuda:

Should we have, like–

Anthony Fowler:

Are there any interesting lessons to be drawn about, say, the Democratic primary?

William Howell:

I wouldn't be comfortable. I guess there's too much uncertainty about those. Both the locality of these estimates and, by extension, what the story is that we want to tell about the effects, to make claims about the presidential election. But what I would say is, I quite agree with your original characterization, is this is the best evidence we have. And in a world where all we have are just-so stories, we should update. My inclination was to say it ought to be a good thing, but I'm not so sure about that anymore.

Anthony Fowler:

Would you feel comfortable saying something like, "Kamala Harris attacking Joe Biden is probably, on net, probably a bad thing for the Democrats' chances in the general election?"

William Howell:

There's so much distance between a paper that shows that a prolonged, prolonged primary speaks to the costs, the electoral costs, associated with specific accusations in the context of a primary. It's just so much distance. I wouldn't be inclined to me.

Anthony Fowler:

Okay, well, I'll give you another one. Would you be comfortable speculating that perhaps... Say, a less experienced candidate. So there's a bunch of these candidates. Someone who's a little less viable, but could be a problem, and maybe Mayor Pete's a good example. Somebody who is a strong Democrat who could be a viable candidate in the future, but is low on experience now and is young. Would the party be better off if they encouraged him to wait? Say if they said, if they don't run now, wait another four, eight years. Would the party be better off if they did things like that.

William Howell:

I guess I'm more comfortable there. I am. But I want to hear what you think, Wioletta. I'm more comfortable there.

Wioletta Dziuda:

I'm not, I'm not. I think there's nothing in the paper that talks about the number of candidates, and also of the effect of exposure for an unknown candidate. If I were pressed by you multiple times to actually make a prediction, I think the only thing I would say is, maybe we can look at the length of the Democratic primary. I know that there's some sort of calendar that's fixed, but candidates decide when they start talking about running for elections and so on at their own time, so we can look at, did we have an unusually long... Did we start unusually early this year? And if we did, then perhaps the paper tells us something about how negative this might be for Democrats, but I think that's as far as I would be willing to go.

Anthony Fowler:

Okay. All right. I'll give you another one. Just for fun. Hypothetically...

Wioletta Dziuda:

We'll be here for five hours.

Anthony Fowler:

We'll just... Okay, don't get mad at me about any candidates that I pick right now. Hypothetically, it's down to the end. we're deep in primary season and everyone is sick of this, and it is down to Bernie Sanders and pick your favorite Democrat. Joe Biden. It's Biden and Sanders, okay? And Sanders is probably not going to win, but he's the last one there. And you know, Biden's basically got it locked up, but Bernie's not given up, and he wants to stick it out till the bitter end.

Would the party be better off going to Bernie and saying, "Look, Bernie, you've got to stop it. You've got to drop it. Stop doing this. We're dragging this out way too long, where you're burning up resources." Would that be a good idea for the Democratic party? No, not that they could in fact convince Bernie to do that. But would they or should they do that?

William Howell:

We're closer. I mean, again, this is where–

Wioletta Dziuda:

I am not closer.

William Howell:

You're not closer.

Wioletta Dziuda:

No.

William Howell:

But your examples are getting closer and closer to the evidence in the paper. That is, then take it out of the presidential context and put it in the South and have it be a run-off for...

Anthony Fowler:

Oh, come on. You're not–

William Howell:

You want to move out. That's right.

Anthony Fowler:

No generalization whatsoever.

William Howell:

So, look, based on the evidence in the paper, I would be inclined to pause at that moment and say, "Look, the best evidence that we have, Bernie, is that this is likely to hurt. Come on, for the good of the party."

Wioletta Dziuda:

I think this boils down to the discussion we had earlier that, again, going back to 2016, actually one could say that actually what helped Donald Trump be elected is the fact that he was all the time in the media. That people got excited. That it became like a game, in a sense. It became like an entertainment, and this excitement somehow translated into the vote. So perhaps a fight between two very colorful candidates might be something that makes voters engaged and remember that, you know, they are excited about politics.

So thank God I'm a theorist. I don't have to make those decisions. But I would be reluctant to–

William Howell:

But this is, again–

Wioletta Dziuda:

To give this advice based on the paper. Perhaps, you know, you have your own experiences and so on that make you make this decision.

Anthony Fowler:

I think if we are... I mean, I've given you a few hypotheticals, obviously. If we think that empirical social science is ever going to help inform things like party strategy, it's the case that I just gave you. And so if you're not willing to say, "Okay, let's let this piece of evidence inform our decision in this case," then you are never going to find a social science paper that helps inform your decision in–

William Howell:

No, but what we'd say is, we want a dozen of these papers. That's what we want. We want to see another context.

Anthony Fowler:

Sure, sure.

William Howell:

We want to know that the effect is real. We want to get some greater clarity about what these underlying dynamics are before we feel comfortable coming out and saying, "Here's what the reform ought to be."

Anthony Fowler:

Of course, these party leaders, they are going to make whatever decision they make based on not this evidence.

William Howell:

That's right.

Anthony Fowler:

And presumably yeah, we have something better to tell them than what their gut tells them. And so maybe I wouldn't be as shy as you guys are in trying to inform that decision with some actual evidence.

William Howell:

While you continue to collect more evidence.

Anthony Fowler:

Of course.

William Howell:

You certainly earn that, of course.

Anthony Fowler:

And that campus will, so. And while you write down a model that tries to make sense of this. No?

Wioletta Dziuda:

Yes.

Wioletta Dziuda:

Thanks for listening to Not Another Politics Podcast.

Anthony Fowler:

Over the next season, we're going to be covering all sorts of topics such as political polarization, controversial policies like burqa bans and border walls. We're going to be talking about political discrimination, and although we will probably try to resist it, nonetheless, Donald Trump is going to make his way into our podcast.

William Howell:

He is going to be there. Our show is podcast from the Harris School of Public Policy and produced by Matt Hodapp. And thank you for listening.