Are Primary Elections Responsible for Polarization in Congress?

I'm Anthony Fowler. I'm Viola Giuda. I'm Ethan Buenner de Mosquida, and this is Not Another Politics Podcast. You know, as we talk about a lot, we have, in the United States, increasing polarization of our elites in politics. They are, in general, more extreme ideologically than the citizens they represent. So Democratic congressional representatives are to the left of the median voter in their district, and Republican congressional representatives are to the right of the median voter in their district on average. And there's a big question about what is causing this increased polarization of our elected officials. And one hypothesis you hear a fair bit is that with the decline of the party system, etc., the electoral environment has changed in ways that force extremism on people. And one of the most prominent versions of that is that elected representatives are worried that if they don't behave in an extremely sort of ideologically or partisan, pure manner, they will face a primary challenge from the extreme flank of their own party. So it is that they have become less worried as congressional districts have become safer and more worried about being challenged. If you're a Republican from your right, if you're a Democrat from your left in a primary that pushes them to sort of be more ideologically pure. But the political science literature has actually had a fair bit of trouble finding a ton of evidence that primaries have a big effect on the behavior of politicians or on the kind of voting or outcomes we get. but we have a guest today who's made some progress on that. But before we interview our guest, I don't know, Viola, what do you think? Just to elaborate a little bit on what you said, the story is that the electorate that goes and votes in the primaries is a different electorate. First of all, if you vote in the Republican primary, you draw from the right-wing side of the political spectrum, but also in particular there's this perception that not many people bother to show up for primaries and only the most extreme voters do. And then as a politician, you're really worried that if you're challenged by someone who presumably doesn't even stand much of a chance in the general election, you are going to lose the primary. The story seems very true, rings very true, seems reasonable, seems like the best and the cleanest explanation for the polarization in the US. And yet, as you mentioned, we looked at some papers and those papers seem to challenge this. So yeah, so I'm excited to hear what's the latest on this topic. Anthony, you've written a paper that's just coming out in the Journal of Politics. Yes, I wrote this paper with Xu Fu, who's in Shanghai. He's a former graduate student here in Chicago. And the paper is about primary elections and congressional polarization. The main idea of the paper is that if primaries were having this big polarizing effect in Congress, we might see members of Congress change their voting behavior throughout the year, throughout the term, depending on the primary election calendar. And so the main research design in the paper is right before or after the primary election date, we can look at how members of Congress are voting. And if the primaries are having this polarizing effect, we might see that once they've won their primary election, they're no longer worried about it. maybe now they're thinking about winning the general election, we might expect to see them really moderate their roll call voting. So that's the main idea of the paper. The theory there is sort of voter forgetfulness. It's a good question. One story would be something like maybe the voters are somewhat myopic, and so I can kind of, maybe I can trick the voters by being, you know, win the primary by being really partisan or really extreme, and then I'll go back to being a little bit more moderate for the general election. That would be one story. They won't remember it next time. And I think that, you know, I think some people in the political behavior literature would probably buy that story that like voters aren't paying really close attention and they're looking at what's happened in the very recent, you know, last couple months or something. Another story, you know, another story would be something like, I mean, there is just varying salience of these two considerations in the minds of the member of Congress. Like when I'm really focused on winning the primary, I'm going to be thinking about that. When I'm really focused on winning the general, I'm going to be thinking about that. So that's myopia on the part of the politician. So it could be, I guess it could be some kind of... Like if voters remember, they should be thinking about both all the time. It could be something like that. Well, yeah, we could talk about this. Somebody has to be myopic. We are, yes, we are... Or, I mean, or just impatient in a particular way. But, you know, like... It could be, yes. I mean, it could be like I'm imperfectly informed and I'm learning about the ideology of candidates on the basis of their recent behavior. We talk about this quite a bit in the paper because this is one of the, I think, one of the big reasonable objections to the paper is, well, shouldn't they always be worried about primary elections? And so probably the best answer we have on this is we can say, we can look at special cases where it's very unlikely that the member in question is going to be seeking re-election again. And in even those cases, we can say, okay, you're an 85-year-old senator, you're running for re-election. If you get this nomination and you win, will you, you know, probably when you're 91, you won't be running for re-election again. Don't say that to Chuck Grassley. And so... You're saying he's going to be president. So we can look at special cases like that where empirically it's extremely unlikely that you're going to be running again for re-election. And even in those cases, we get pretty similar results. So we can talk more about that. Absolutely. Your paper is about incentives and not about the selection. Both of which are interesting. And again, one could maybe make a story that I think you were trying to allude to that I should have those incentives throughout my career, not just before the primary and just after primary. But so what you are picking is only the incentive, what you might be picking, at least what you are searching for, is the incentive effect that's associated with this increased salience or recency of the primaries. Yes, that's correct. And I think, I mean, we don't study the selection effects. Partly, I mean, part of that is just there are other good studies that focus on the selection story, which is to what extent do voters tend to select more extreme candidates in primary elections? And the answer to that is they also don't. There's not a lot of evidence that voters prefer more extreme candidates in primaries. Extreme candidates, on average, don't do much notably better than moderate candidates in primaries. So there are other studies that do that. We don't look at that question. But also, I think, in terms of this debate about polarization, I think the incentive effects are often what people talk about the most. Members of Congress themselves say, well, I can't moderate because I'm worried about losing my primary election. And so they talk about the incentive effect as being particularly important for polarization. And so we focus on that. And you're right that the main design just focuses on the kind of incentive effect that varies over the course of a Congress. So if there is some incentive effect that's just kind of baked into the system and it's always there, then we wouldn't be able to pick it up. And that's why I think the analyses we do in the appendix with these senators who are very unlikely to run again and so forth, I think that's useful. You could say, well, maybe it's really all like baked into like you've selected for the kind of person who just never stops thinking about a primary election. And that's really a selection story. And maybe that contributes to polarization. Of course, the kind of people that run for office are either themselves extremists or they're thinking about their partisan base all the time. That could be true. And that's not what we're picking up here. So say a little bit of in summary about exactly what you do and what you find. Yeah, so the basic design, I mean, we're analyzing data at the level of roll call votes. So we look at every roll call vote that's cast by members of Congress in the last several decades. And we can see whether or not the roll call vote you're taking is taken before or after your primary election date. And the nice thing from a research design perspective is that different states hold their primary elections at different times. And so you could have a bunch of members of Congress voting on the same exact bill. maybe they're from the same party but for some of them their primary election date has passed and for others their primary election date hasn't happened yet and there's a lot of variation actually over the course of there's many months of variation where some states are having primaries some states haven't had them yet and so forth and so we're going to utilize that variation in a diff and diff design so we run a big regression that has we account for differences between different members we account for differences across different bills and what's left over is essentially we're asking how much do people change their voting behavior around the time of their primary election. And because we have other people voting on the same bills with different primary election calendars, we can do a really good job accounting for differences between the bills, differences between the members, et cetera. That's the idea. So we're looking within a given Congress. Within a given Congress. So within a given Congress, every member has one primary election date. And so most of the bills are before your primary election date, and then some of them are after. and exactly how many depends on what state you're from. And so the headline finding. The headline finding is that there is an effect that goes in the expected direction. So after your primary election date has passed you do become less partisan and less ideologically extreme It looks like there some evidence that you moderate a little bit once you you know you're less worried about winning the primary election. Maybe you've already won the primary election. Now you're thinking about the general. You do become a little bit more moderate. So that's in the direction that you'd expect. But the substantive magnitude of the effect is tiny. In the House, it's something like 0.2 percentage points. So your 0.2 percentage points are less likely to cast an ideologically extreme vote, that effect can explain something like 1% of the partisan polarization that we observe in Congress. So if you wanted to explain why the Democrats are so much more liberal than the Republicans, the possibility that Democrats are catering to their primary base and voting as if they're always worried about winning a primary, that could explain maybe 1% of that difference. This is a lovely paper. I really like this paper. This is the point where I disagree with you. I don't agree with your interpretation. This is a paper not, as you say very clearly, it's not a paper about, say, institutional reform. You are not asking if we changed primaries in this way or that way or eliminated primaries, what would happen to polarization? It's a mechanisms paper. We do have some evidence in the paper on top two primaries versus partisan primaries and whatnot. But yes, you're right. I agree. And the mechanism, that you, so I guess I would say two things about the mechanism. One is the headline finding, I interpret to say, there's a hypothesis out there that there is a mechanism which is congresspeople are thinking about their primary challenges in making their legislative voting decisions. And in particular, they are thinking about a challenge from their extreme. And you've shown quantitative behavioral evidence, as opposed to just asking them and hearing what they say, that that mechanism exists. That appears to be in their head and being translated into behavior. That seems very important. And so now we have sort of direct behavioral, quantitative, well-identified evidence that that mechanism exists and is affecting the behavior of Congressional Law. If we wanted to ask a question along the lines of... And then you have a source of identifying variation, which is like the salience of your election driven by timing. And what you can say is like, how much of the polarization could be explained by the translation of the salience of your primary driven by timing on your behavior? Once we know, and I agree that's like small, but once we know that the mechanism is real, that this is in Congress people's heads and they're acting based on it, there are now a million pathways by which the fact that that incentive is in their head and they're acting on it could translate into polarization not just the one that you have identifying variation for a ton of pathways they could it could affect the kind of votes we get the kind of bills we see it could affect uh and you just think about like the margins of like their behavior or all the like it could be that because that thought is in their head if we turned off primaries entirely there'd be huge general equilibrium effects that that are that are not here from the timing bit of the mechanism, but we know the incentives in their head. And so what we know, I think, like the way I would summarize it is like, you've provided good evidence that the mechanism exists. You've shown that one very particular identifiable pathway is small, but we have nothing here to suggest like the total effect of the fact that that incentive is in people's head is the aggregate of a bazillion things that you need a bazillion research designs to add up. We don't have any of those other research designs. And so had you found a huge effect, there's this asymmetry. You had a feudal effect, you would have been able to say, this thing's really, like, primaries are really important. Having found a small effect, but a real one, all you can say is, we know primaries have an effect, and the aggregate effect could be tiny, it could be huge, and you'd have to do a thousand other studies to identify all the other mechanisms to answer that question, and we're just not in a position right now to answer that question. And if I can add, or maybe like rephrase of what you said, in a sense, if you read the design of your paper, the prior, I think, should be that you're going to, if you're going to find anything, it's going to be small, because it's not, you know, you even have the discussion in your paper. It's not obvious, you know, how much the salience matters, like they should be thinking about primaries all the time, or maybe they shouldn't be thinking about primaries at the beginning of the Congress's session. So it's a priori, I think, even if I came with a prior that primaries have a huge effect on polarization, I would not expect you to find a huge effect. So just given this prior, I think I agree with Ethan that now my prior were upheld, that the effect might be large. Yeah, I want to think more about what those, I mean, I'm open to that possibility. I think those are good points. I'd want to think more about what are the reasons why this estimate could be really small, but in fact, the aggregate effects of primers could be really large. One we've talked about already, which is just the treatment itself is kind of weak. Maybe they're always thinking about primers. They're thinking about it all the time, and you're just getting a tiny... And then I would go back to this result I've mentioned already, which, you know, unfortunately, the referees made us put in the appendix, and that's just the way this works sometimes. But I actually think it's kind of important, which is if we focus on these cases where this is almost surely your last time running for election, we do actually have a pretty strong treatment in that case. We have a case where this is the last primary you'll ever have to run in again, and let's see how you change your behavior once that one's over. And even then, the effects, they're a little bigger, but not much bigger. So I think that, you know, if that was the big concern, if it was that they're always thinking about primaries, then I would expect in that case that this should be really big. But super entrenched incumbents may also be super safe. like it may be very very difficult like they may be exactly the politician right i'm an 80 year old senator i've been reelected seven times like i i've got a huge you know i've been selected on quality lots of times whatever and so like i'm just not electorally vulnerable like they could be it might not be so important for me to win this last primary because anyway i'm thinking about and we do find variation along those lines i believe if i remember correctly i think the effects are bigger for members from more purple, more competitive constituencies, things like that. So there is variation along those lines. But we can't find any case where these effects are huge, even in that pretty extreme case of this is likely to be your last. I mean, so just one thing to add to what Ethan was saying. Like, it's not only that you found an effect that goes in the direction you would predict, but, like, you sliced and diced the data. And, like, all the effects that you would predict theoretically seem to be there. They all go in the right direction. So I think it's a really pretty strong evidence that this is not just a fluke. You just happen to find a little bit of more polarization. It seems like everything that's consistent with this thinking about the extreme challenges during the primary seems to be borne out in the data. That convinces me that there's something real going on. I think I agree with you on that. In terms of how big it is, I don't know if that tells me, oh, this must be really important. I think this is like a deep problem with, like if you think about the credibility revolution, and it starts with program evaluation where we are directly answering questions about an intervention we're considering, and then we really want to know what's the effect of this intervention, right? Because like we might do this ed policy or we might not, and we want to know like literally is this education policy going to change test scores in this way or that way or labor market outcomes in this way or that way? And we get the insights from that, and then we pull it back into a much broader set of research questions where we're not doing program evaluation. You're not evaluating a primary reform. We're trying to learn about theoretical mechanisms, which might have lots of paths to... And one thing that thinking about this paper made me kind of come to is, no matter how good a job the empiricists do, I think we just can't answer the question, how important are primaries for polarization? What we can answer is, do they matter at all? And do we have evidence that the kinds of things that we think theoretically might be in people's heads are, in fact, translated into behavior. But we're just never going to have research designs for all the pathways. And so we're never, we're never going to know. Can I go back? I actually want to go back to the beginning. I want to go back to the motivation a little bit, which is both of you kind of asserted at the beginning that it just seems like if you just look with your own eyes at Congress, it seems obvious that primary elections play a big role. And I'm not sure I agree with that. And I think for a couple reasons. So, and so if you know, to think of, say, a recent presidential primary, for example, Like there just aren't that many Democrats around Elizabeth Warren. Right. The idea that like somehow some, you know, there's there's a lot of Democrats around Joe Biden and to the right of Joe Biden. And there aren't that many Democrats to the left of Elizabeth Warren. And so the idea that like if you're an incumbent with Joe Biden's ideology, that you'd have an incentive to move to the left to cater to those people. There's there's just there's not a huge mass of people over there. So that incentive isn't that strong. You don't need super extreme primary voters for it to have this effect. for it to have a polarizing effect. So if I just imagine that the median voter in the Democratic primary and the median voter in the Republican primary are to the right and left of each other, right? Can you imagine? Sure, sure. And on either side of the median voter. And then if I imagined just like a kind of a Downsian model with some uncertainty so that I gonna get two candidates symmetrically distributed on either side of the median vote of their primary and it going to be a coin flip between them Sure, sure. Half the time. That could produce some polarization. Quite a bit of polarization. Like more, I will get the like lots of politicians to the extreme side. But if I look at the typical member of Congress, and there are estimates of this, and we could quibble about the quality of the evidence, but there are estimates suggesting that the typical Democrat in Congress is well to the left of even the median Democrat from their constituency, and the typical Republican in Congress is well to the right of even the median Republican. And so it's not like they're catering to the median partisan in their primary election. They're even more extreme than that. Yes, we try to study this question of whether politicians are more extreme than the constituency that they represent, but it's also a very hard question to study because we ask voters what they think, and then when we measure politicians' positions, we usually look at some sort of behavior, which is all strategic. And so I just want to put it there. Sure, no, that's fair. I think it would be crazy to think that primaries are the only story here. And so, for example, politics becoming a worse business in various ways, so extremists being more willing to run than moderates is another force that doesn't have to run through primaries. And then I just want to add the one additional wrinkle, which is suppose you are an extremist voting in a primary election. You still really want your party to win the general. And so you have to be forward-looking to some extent, and you have to say, okay, on the one hand, maybe I would prefer Elizabeth Warren over Joe Biden, but if I nominate Elizabeth Warren, she's less likely to win. Unless you're myopic. And that also, which, you know, I mean, we can talk about how myopic they are. Yeah, I generally don't put a lot of weight on those myopia kinds of arguments. But I think that's another moderating force, that even within this primary election system that we might think is polarizing, there is some built-in incentive for moderation, which is you want your party to win the general. There is an incentive for moderation, but you still don't moderate because of the uncertainty of where the real median voters, you don't moderate all the way to the median voter, which you would if the median voter was already the median voter for the primary so there is a moderating effect the question, this is a particular question how big it is but we still could envision that it's not very big and the polarizing effect of the primaries is still large I think the question of do voters have incentives to moderate to win the general isn't quite the right question to respond to the critique so I think of course they do but if we're thinking about the question do primaries contribute to polarization? The right question is, do the Democratic voters in a primary have more incentive to think that way than the party leaders who would be choosing the candidates absent a primary? And I think the answer is obviously no. The party leaders are better incentivized and face fewer collective action problems to select candidates who are electable than voters who face all sorts of collective action problems. Depends what the party leaders care about. I mean, of course, there could be the corruption and whatever, but I'm just saying, like, the right... We could debate who's better incentivized, But the right question is not do voters have an incentive to think and moderate to win the general. The question is comparing the voters who vote in the primary to whoever else would be deciding instead, let's say the party leaders, who has the better incentives and can better act on those incentives. And my guess is party leaders are better incentivized to win in the general than voters. And I certainly think fix the incentives for achieving the goal. Party leaders are better able to act on their incentives. So my guess would be the opposite. And the evidence would be on my side. This is the Hirano et al. paper that I think we've talked about before on the podcast, is that when states switched from the old smoke-filled room system to partisan primaries, we actually saw a reduction in extremism and polarization. And I think the story I would tell is probably the opposite of the story you were telling. Did we see an increase in electability? You see more moderation under partisan primaries than under, you know, nominations from party leaders, which is consistent with this idea that either it's just the voters are themselves more moderate or they're more forward looking and the party leaders. And, you know, having met a handful of these party leaders, I'm not shocked by this result, actually. But, you know, one thought that's been in my head that, you know, there's just no way to figure out whether it's correct or not. But, you know, we tend to think about this as a property of the system that either primary is polarized or not. But I think there might be some dynamics going on. Initially, when we switched to the primary system, it might have been the case that actually, you know, now anyone could run. And you didn't really need strong support of your party. And maybe we did have more moderation. But over time, this threat of extremism arose and started being salient in people's minds. And then over time, they became more and more attuned to that. It's completely possible that the story is not static and people are learning throughout this entire period that we are referring to. Can I mention, I want to mention one other set of analyses in the paper. This is another one that, unfortunately, the referees and the editor made us put in the appendix. And, you know, I don't know. That's just how the publication process works sometimes. We're so thankful to the editor and the referees. But I actually thought this was really interesting. And you'll see why they made us move it to the appendix, which is the results didn't come out the way that you would have expected. So this is kind of a sad publication bias story. But we did a bunch of analyses with filing deadlines. So the other important date is the date by which any challenger, any primary challenger would have to officially declare that they're running against you. And so we thought that was also an interesting date. Those also vary wildly across states. Because then you learn whether you are being challenged, whether you're being challenged by an extremist. So that was the idea. So one thing we found is that on average, you see more extremism and more partisan voting after the filing deadlines. One thing you would theoretically expect, at least I would have theoretically expected, is that if you learn that you have a challenger, maybe you become a little more extreme and partisan. If you learn that you don't have a challenger, maybe you actually become less extreme and partisan. Like maybe you were voting in a partisan way to fend off potential challengers, and then you find out, oh, we didn't get any challengers, and now I can kind of go back to voting the way I wanted to. And we don't find any evidence of that, actually. So just on average, you become a little bit more partisan or more extreme after filing deadlines. So you don't find significant effects, but the sign flips. And the sign does not... So there's kind of a positive effect of the filing deadline passing on extremism. And it doesn't depend strongly on whether you did or didn't get a primary challenger. But the thing we don't find is people moderating once they learn for sure that they don't have a primary challenger. I would have expected to see something along those lines if, in fact, there was this big effect. And I think that's an interesting result in that you would have expected this if we didn't find it. If you put those results together, it suggests that members, maybe members are voting in more extreme ways with the goal of winning primaries, but they don't seem to be voting in more extreme ways with the goal of fending off primary challengers. Anyway, I thought that was interesting and complicates the story a little bit. Does it just look like in a purely correlational way about who gets a primary challenger? Do we know whether more moderate or more extreme? I think one of the strong results is that you're much more likely to get a primary challenger if you're in a safe seat. If you're in a safe seat. And people in safe seats tend to be more extreme. It would go in maybe the opposite direction that people would expect. Extreme vis-a-vis the median voter in the U.S. or vis-a-vis the median voter in their decision? Vis-a-vis the country, right? Yeah, that's right. It is very rare for, I mean, another thing worth mentioning, it is most members of Congress do not lose primary elections. It's very rare, right? A couple per election cycle. It happens more often in redistricting years where two incumbents have to face each other in a primary, and then obviously one of them wins and one of them loses. But it is very rare for an incumbent to lose a primary to a challenger. How rare is it to have an expensive primary? Like, so even if you're going to win, if you have to spend a bunch of your war chest on? Yeah, that's a good question. not very expensive. We looked at the, we have the campaign finance data. We use that as, you know, one of the things to look at. Like maybe you had a, maybe you had a serious primary challenger who themselves spent a lot of money, but even our measure of serious primary challenger, I think, is pretty low. Like if you had a primary challenger that raised $10,000, I think we count that as a viable challenger because there just aren't that many cases of really serious challengers running, and there aren't even that many really close primary elections. I think the argument against what I'm saying is that could be all in equilibrium where these members are really good at fending off primary challengers and they're catering so strongly to their partisan base that that's why they win their primary so often. But just anecdotally looking at the data, you would also say like, this probably isn't a huge force because it's most members of Congress just aren't that worried about, they shouldn't be that worried. They might say they are, but I don't know. One possibility is they're rationalizing their extreme behavior. They say, oh, don't blame me. Like, of course, of course I took all these extreme votes and I can't support your moderate bill, but it's not because I don't like your bill, it's because my primary voters won't go for it. That's one possibility. I mean, I guess I'm off two minds. Now, on the one hand, I think, I buy what you are saying that you slice and dice the data and it seems like you can get a big effect and sometimes the same story would predict a different effect that you find a sign even flips You talked about those results in the appendix. And in some sense, we do see polarization in other countries also arising at different points in times and not associated with any primaries. so obviously there must be a lot of different sources of polarization but on the other hand I agree with Ethan that when you just read the main body of the paper the results seem to be so clean like every question that you ask when you are motivated by theoretical reasoning it seems to be borne out in the data the magnitude is small but it's there and it's significant the paper is consistent with the possibility that congresspeople believe that they are under threat from extremist challenges in primaries and that they are wrong. They always win when they haven't written all the things. But they might be telling the truth in their memoirs. Like they may really believe it and be acting on it, kind of never get to learn that they were wrong because they act on it. And so they never run in a primary against an extremist when they haven't themselves shaded to the extreme. And so there's a self-confirming thing going on there in the same way that they never get to learn that their campaign ads aren't very important. Yes. I think that's plausible. But they have a lot of incentive to learn, and we had all these podcasts, so why wouldn't they listen to our podcast? Wait, why do they have such a strong incentive to learn? Assuming that they do care about policies, they would like to know that. What if mostly they care about power? Well, they still want to win in the general election, and you would think... They're mostly in safe seats. They could also be extremists themselves, and so what we want to see, I don't know, what would we want to see? I mean, we'd want to see somebody break the trend and show that you can, in fact, do really well in both primaries and general elections by moderating. We know that, no, that you can do pretty well by moderating. We have papers. We have examples. Certainly in generals. I mean, I think there's very clear evidence, I think, that moderation is good for you in the general election. In the primary election, it's more mixed. it's like probably there's probably there's not a really strong effect either in either direction is what it looks like to me. If you are in a very partisan district where you're assured of winning the general anyway, then maybe there's just not a very strong incentive to moderate and you just, you should just do whatever you want to do. I don't know, you tell me. It seems to me casually that congressional leaders are avoiding compromise bills that would put their, that would force their members to reveal moderation. Now, that could be because the congressional leaders themselves are extremists and they don't want to compromise. That would be my first story, probably. But yes, yes. And maybe that is, maybe they are worried. Yes, they're worried that in the most extreme, in the most partisan districts, members could lose primary elections by taking a moderate vote on those. I mean, if you think about this salient story, actually, that could be another thing to look at. Like if we look at bills that are going to be more bipartisan than average versus bills that are less bipartisan on average? What is the timing by which that, like, you know, there's an intuition and I think a kind of a storyline that we vote on bipartisan bills early in a Congress, not late in a Congress. Or this story would be you do it really late, like after most of the primaries are over, or you do it really early before, right, and you don't do it. Do we have papers on that, on sort of when do bipartisan bills come up? I don't think so. I mean, there is a literature on, like, when do you get significant legislation, that kind of thing, but I don't think that's the kind of thing that has been investigated carefully. This episode is just full of ideas for papers. All right. Lots of papers to write. Overall, I think I'm convinced that it found something. You know, there's still a huge question mark to what extent primaries contribute to the polarization in the U.S., but I think by now we have had so many papers that point in the direction of saying defect is not huge, and as we talked about, it's really hard to find another variation that's going to give us a better answer to this question. So we have to move on. That's my bottom line. Just move on. The bottom line is at least you found something and we should move on. Move on means never study this topic ever again. It's the final word. Never talk about this. Final word, yes. You killed the literature. My bottom line. My bottom line, I think, is The paper got me thinking again in a way that I find, like, a little disheartening, I guess, to tell you, about our ability to ultimately answer the big questions we care about. Like, I think this paper is an example of as nice a job one can do of identifying a mechanism we care a lot about. And it's really important to know that mechanism. And then like when I think about like the kind of case that like, say, Scott, Chris and I were trying to make in theory and credibility is like lots of doing that is going to add up better to answering the big questions we wanted to answer than anything else we can think about. And I still think that's true, but I also think I've convinced myself that we're not going to add up to having a definitive answer to the question, do primaries have a big effect on polarization or not? Because I just think what we can do is this kind of paper that nicely identifies a mechanism, but where it's extremely hard to aggregate up to what would be the effect of the institutional reform. And I just think that's not a problem with the paper. that is, I think, a limit to the enterprise we're engaged in. Like, you know, we have to be a bit epistemically humble and, like, there's questions we can answer, there's questions we can't answer, and we should answer the questions we can answer. I don't think we should move on. I think you should write more papers on primaries. But I don't, I have trouble imagining a world where even if you wrote 10 papers this good, I say to myself, I know whether in the end it adds up to a big number or a small number. I think, like, I just, I feel like there's the exploitable variation in the world and the nature of institutional reform and equilibrium makes that really, really, really hard in a way that just means I think, you know, I have to tamper my expectations of what political science as a discipline can do with respect to answering this kind of question. That is very close to let's move on. I know you tried to say it. No, because I'm saying there's nothing to move on to. We should just keep doing this. If we can do this, let's do this. Everything else is going to be like that too. This is not about this paper. It's about identifying mechanisms in general. I think I'm more okay than you are in saying that we've learned more than that. And yes, we've identified that there is a mechanism in the world that looks like you might expect, that members of Congress do have in their heads that they have to worry about their primary election. Except for those results that you buried in the appendix. And that's leading them, yes, that's right. And that is leading them to take more extreme votes during the course of their primary campaign. And then they go back to being a little bit more moderate afterward. but the effect is so small substantively and it's still really small even for those members who aren't going to be running for another primary that it makes me think the the the that can't be a huge that can't be a big part of the story that can't be a big part of the story about why democrats are so liberal and republicans are so conservative in congress it's got to be something else it's got to be things like they themselves are extremists and other i mean there's there's there's other things that we could talk about. They're not in very competitive constituencies to begin with and so forth. There's more things that I think are worth studying if we want to know why is there so much polarization in Congress. We've studied the primary question from a few different angles, not to say that we're done, but the estimates in this paper are substantively very small. And I think overall, looking at the literature, I don't think primary elections are having a huge effect on polarization. You think we should move on? I don't know. I think there are more papers to be written. and I think we came up with a couple good ideas even in this conversation, but I'm willing to go further than you are by looking at the evidence and saying, I think we do learn something from the fact that these effects are tiny. If I describe my research design to the people who say the primary effects are huge, I think they would have expected much bigger effects. If you're enjoying the discussions we're having on this show, there's another University of Chicago podcast network show that you should check out. It's called the Chicago Booth Review Podcast. What's the best way to deliver negative feedback? How can you use AI to improve your business strategy? And why is achieving a soft landing so hard? Well, the Chicago Booth Review Podcast addresses the big questions in business policy and markets with insights from the world's leading academic researchers. It's groundbreaking research in a clear and straightforward way. Find the Chicago Booth Review podcast wherever you get your podcasts.