Do polarized politics leave anyone left in the middle? Anthony Fowler finds that most Americans’ political views fall between the opinions of Democratic and Republican elites. And that’s not because they don’t understand politics in the same way. Most Americans’ views fall into the ideological continuum from left to right; they’re just somewhere in the middle. These moderates matter for election outcomes. While they participate a bit less, they are the consequential swing voters. Fowler also finds that selecting candidates on policy grounds could matter more to voters than the power of partisan identity. We may not be giving voters enough credit.

Guest: Anthony Fowler, University of Chicago

Studies: Moderates; Partisan Intoxication or Policy Voting? 


Matt Grossmann: Moderate voters matter, this week on the science of politics. From the Niskanen Center, I’m Matt Grossman. Does our polarized politics leave anyone left in the middle? In fact, most Americans political views fall between the opinions of Democratic and Republican elites, and that’s not because they don’t understand politics in the same way most Americans’ views are well summarized by the ideological continuum from the left to the right, they’re just somewhere in the middle.

This week I talked to Anthony Fowler of the University of Chicago about his co-authored American Political Science Review article, Moderates. They distinguish between true moderates and those giving opinions that don’t line up with the left right spectrum at all, or can be better explained with more than one ideological dimension. They find that true moderates matter for election outcomes, although they participate a bit less, they are the swing voters. We also discuss his quarterly journal of political science article, Partisan Intoxication or Policy Voting. He finds that many claims about the power of partisan identity to voters could be explained instead by selecting on policy grounds. Perhaps we’re not giving voters enough credit. Here’s our conversation. If you could just start with a summary of your recent article, pithy titled moderates. What did you find, and what were the big takeaways?

Anthony Fowler: Sure. Yeah, I wrote that article with a great team of co-authors. Let me make sure I remember all of them. It includes Seth Hill and Jeff Lewis and Lynn Vavrick, Chris Warshaw and Chris Honovich. The six of us worked together, which six is not the ideal number probably, but it was great to have such a great team to work with. One of the things we sought out to understand was we know there are a lot of people in surveys who give a mix of liberal and conservative answers to policy questions. And we wanted to understand are those people genuinely moderate or are they actually ideologically in the middle or are they just conflicted? Maybe they’re very liberal on some issues and very conservative on some issues, or are they just not even paying attention to survey questions? Are they just giving random answers?

And so our goal was to try to look at the data, and even with the available data, try to see if we could come up with a strategy that would distinguish between those different possibilities. And it turns out that the vast majority of people do have opinions that are well summarized by a single ideological dimension. Something like three quarters of survey respondents are well summarized by that one dimension. And if we look at their ideological scores, they look more or less like a normal distribution with most people close to the middle. And so the conclusion is that most Americans do have a coherent ideology, it’s well summarized by a single dimension, and most of them are in the middle.

Matt Grossmann: So let’s go through those three types a little bit. Distinguish what a prototypical person might look like, who’s truly in the middle versus someone whose opinions aren’t ideological at all, versus someone who’s maybe liberal on social issues and conservative on economic issues.

Anthony Fowler: Sure. Yeah, I mean that’s the exact typology. You’re right. And I think when you just observe that somebody sometimes gives liberal answers and sometimes gives conservative answers, it’s hard to know which of those three possibilities explains their answers. Our strategy was to essentially look at the pattern of responses across a bunch of questions, and there’s different kinds of patterns that would be more or less consistent with those different typologies. So a simple example would be, suppose there were a bunch of questions about minimum wage. Suppose there was a question like, should we raise the minimum wage to $9? Should we raise it to $10, $11? Suppose you had all those different questions. And so then presumably most people, if they have a coherent ideology, most people will answer this similarly to those questions in a predictable way such that if you answered yes to raising it to $15, you probably also would answer yes to raising it to $14.

And so you could imagine a particular pattern that would be consistent with someone being explained by a coherent ideological dimension on just that one issue. And you could distinguish between the extreme liberals, extreme conservatives and the people in the middle. If somebody was just answering randomly, there might be no coherent pattern at all. Any pattern of responses would be equally likely. So they might say yes to 15, then no to 14, then yes again to 13. And although there could be some strange set of preferences as consistent with that, it’s more likely that they’re just answering randomly and not paying close attention. And then if we bring in other issues, you could imagine that somebody is very liberal on minimum wage, but they turn out to be very conservative on abortion or vice versa. And that would be consistent with this. We have this catchall conversion category of people who they’re giving real answers, but those answers are not well summarized by a single ideological dimension.

Matt Grossmann: And roughly how many people fall into each of these categories and how do they compare to the fairly strict conservatives and liberals?

Anthony Fowler: Yeah, so it’s something like three quarters are what we call downsians, that fit that one dimension. Maybe one fifth are these conversions who are giving real answers but are not well summarized by single dimension. And then very few people, maybe 1 in 20 people are giving seemingly random answers that seem to suggest no pattern at all. And then of course, within the downsians we can then ask among the downsians how many of those are liberal versus conservative versus moderate? It’s obviously continuous distribution of ideology, so it’s all in the eye of the beholder a little bit. But what you get is this very unimodal, mostly normal Gaussian distribution with most people close to the middle. You don’t get this bipolar distribution that you would get if say you looked at members of Congress, which suggests that most people are close to the meme.

Matt Grossmann: So you mentioned that it would be ideal to have all these different minimum wage questions, but I know that you had to deal with the questions that people were asked and most of them were yes or no questions about policy proposals. So give us a sense of how those were aligned across this liberal and conservative spectrum. Were there examples where you really distinguished the true liberals or the true conservatives from other folks? And what kinds of patterns did you usually see? What kinds of policy positions did these moderates have?

Anthony Fowler: Yeah, that’s a good question. We do, in fact, in one part of the paper, we actually were fortunate in that one of these surveys, they actually did ask multiple minimum wage questions to the same respondents. And that was a nice way of validating our estimates and making sure that they were sensible in what you’d expect. But by and large, researchers don’t do that. They don’t ask 10 different versions of the abortion question with different cutting lines. And the other tricky features that when survey researchers do ask these binary questions, they usually try to write the question in such a way that roughly 50% of the population will give a liberal answer and 50% will give a conservative answer. If you ask one where 85% give the liberal answer, survey pollsters often think of that as a bad question. But of course those would actually be helpful for us in distinguishing the extreme liberals from the moderate liberals and so forth.

But there is still variation. There are still questions where say 65% go one way and 35% go the other way, and they move around in different places. So the moderate that we’re thinking of is somebody who probably holds a moderate position across a lot of issues. And it might be hard to glean from any one survey response whether they’re moderate, but you look at their pattern across lots of questions, and you see that they tend to give the conservative answer when the conservative answer is a little bit more popular in the general population. And they tend to give the liberal answer when that one’s a little bit more popular. Suggesting that they really are close to the middle. And had you asked the question a slightly different way, they might’ve changed their answer.

So they probably want taxes to be somewhere in between what Mitch McConnell and Nancy Pelosi want. They want the minimum wage to be somewhere in between. And so you go down the line of things. It’s not hard to imagine such a person, but we don’t often ask questions that would easily identify those people unless in our case we have a lot of questions where we can sort this out.

Matt Grossmann: And what’s the relationship between these policy moderates that you’ve identified and self-identification as a moderate or on international scales, even if you don’t ask liberal, conservative or moderate, but you just ask, are you on the left or the right or 10 point scale or seven point scale? There are a lot of people who place themselves in the middle. Are those the same people who are giving this mix of opinions?

Anthony Fowler: I think by and large, there’s a very strong correspondence between the self-reported ideology and the ideologies that we infer from people who serve policy positions. We do look at that. I mean not surprisingly, the people who self-identify as independent or moderate are much more likely to come out as downsient moderates in our analysis. I’m sure there are definitely some mismatches. There are definitely some people who probably think of themselves as liberal or conservative that end up looking pretty moderate on our scale, and some vice versa. But by and large, the correspondence is pretty good, such that if you just used say someone who identifies as independent or moderate as a proxy, you would do a pretty good job of identifying those people who really are in the middle on policy. And I think a lot of political scientists have previously argued that the ideology question, say in the NDS, is not very meaningful and that they don’t have a lot of faith in that. But if you just look at the data, it does a pretty good job of predicting where people would come out in on our measure.

Matt Grossmann: So one reason that people have been somewhat skeptical of that liberal conservative spectrum is because there have tended historically, at least to be a lot of people who place themselves on the center right of the ideological scale, but give policy positions that are more center left. So they say that they’re slight conservative, but they favor increased spending in most areas. And I know you don’t have the same scale on the policy side, but can you say anything about that? Are there people who might self-identify on the right and might have policy positions that look a little more left? Or does the center of your scale correspond more to center left policy positions?

Anthony Fowler: Yeah, that’s a good question. We are mostly thinking of your ideological placement as being relative to the rest of the population. And so it could be the case that the moderates that we’re identifying actually do favor a little bit more government spending, things like that. We haven’t looked at that question very closely. I will say that there is kind of… And I’m going to sound like a conspiracy theorist or something like a Fox News cast. There is a liberal bias in most political science survey questions. The liberal answer, the progressive answer is usually the more popular answer.

And this is not a topic that we’ve delved into closely, but it’s just something we notice in the process of working on this project. And I don’t know if that’s because the liberal position is indeed just more popular among the people in the middle of American society or if these questions are written by mostly liberal academics who write the question in such a way that the liberal answer just sounds like the better one. But had conservatives written the questions, maybe we’d be talking about the opposite pattern. I don’t have a strong view on that, but that is something we noticed and we are mostly thinking about ideologies being relative to other people in the population rather than relative to some benchmark that we have in mind.

Matt Grossmann: So you also find that people who are consistently or more consistently on the left or right side of the spectrum are more likely to participate. They’re more interested in politics and they’re more knowledgeable about politics. So that does seem to be some evidence that we might hear from them more often and we might gain our perspective on the public more from these people who are on the left or right side. So give us a sense of what you think the implications of those findings are, and maybe a little bit on the causal direction? Is this that once people become informed and involved and engaged in politics, they pick a side or is it that those people who pick a side are the ones that we hear from?

Anthony Fowler: Yeah, that’s a great question. You’re absolutely right. There is a correlation. I will say the moderates do participate, they do vote. They’re not completely uninvolved in the political process. They do vote, and they are a very electorally important group. We actually show that they’re perhaps the most electorally consequential group. But you’re absolutely right that the extreme, ideological extremists-

Matt Grossmann: Just to clarify, that’s because they change sides often, not because they participate at higher rates.

Anthony Fowler: That’s right. So they participate at lower rates, but they still do participate and they’re willing to change their votes. They’re willing to switch the parties they support depending on which candidate they think is more competent or more moderate, et cetera. And so they do turn out to be perhaps the most consequential group for elections. But if you just look at sheer rates of participation, and I’m sure if you looked at say social media usage and how loud those voices are, if you looked at donations, all of those things, the ideological extremists are much more likely to be participating in democracy in those ways. And so when you go on social media, when you turn on cable news, I’m sure if you also looked at journalists, you’d find that journalists tend to be more ideological extreme than the general public as well. It gives us the impression that there’s so much polarization out there and that everyone is fighting each other, and Democrats and Republicans hate each other.

And I think that’s by and large a wrong impression. I think if you look at the broad survey of the whole population, you would find that most people are in the middle and most people probably aren’t that happy with either the Democratic or Republican Party. And they don’t, by and large, hate each other, and they don’t, by and large, disagree with each other so much. So I think there is this big disconnect between what you would think from say, looking at social media or cable news and what’s actually going on in the American public.

As far as the direction of causation, I think it’s a really interesting question. Obviously, I haven’t seen a really compelling study that gets at that either way. If I had to guess, I would guess it mostly goes in the direction of extremism leading to more participation, rather than participation leading to extremism. I mean, if I think of a few bits of evidence that inform that opinion, Australia for example, has compulsory voting. That’s one case I’ve written a paper on, compulsory voting. I don’t think you see drastically more polarization or more extremism in Australia just because you compel people to vote or you highly incentivize them to vote. You don’t see people all of a sudden becoming more partisan than they are in other countries.

And if I just think about it from introspection for a second, it’s hard for me to imagine a moderate who is forced to vote all of a sudden becoming more extreme. Whereas it’s pretty easy for me to imagine someone who already has extreme views feeling so strongly about something that they have to get out there, they have to vote, they have to donate, they have to scream on social media and so forth. That direction of causality just makes more sense to me. But I would love to see some studies on this to better answer that question.

Matt Grossmann: So as you mentioned, you also find that the moderates are more likely to respond to candidate factors in elections. The candidate ideology is included in that, and that makes sense. But talk through that finding but also connected your experience findings. Why is it that the moderates are also more attracted to incumbents and experienced candidates?

Anthony Fowler: Sure, yeah, I mean, the moderates are the ones who are the most indifferent between Democrats and Republicans on policy. And of course we know that the Democratic and Republican candidates, in a typical election in the US, are quite far from each other ideologically. So if you’re a liberal or a conservative, you probably know which way you’re going to go. And the difference in terms of the quality of the candidates or the competence of the candidates or some other valent characteristic would have to be so big for you to change your vote. I’m sure a lot of listeners can even relate to this themselves. I’ve certainly had the experience of voting in an election in Illinois where I’m pretty unhappy with my incumbent Democrat and maybe tempted to vote for the other candidate. And then you look at the other candidate and they’re just so far from you on ideology that you say, okay, am I really going to vote for the other candidate?

And you’re not happy with that, but if you are a little bit to one side of the ideological spectrum, you’re probably not going to change the party you support, not because you don’t care about those other things, but because you’d have to sacrifice so much on the policy dimension to vote on some other dimension. Whereas the moderates, they are closest to being indifferent between the democratic and Republican candidate, typically on policy, which means they can put more weight on things like competence and experience and so forth. And so empirically, that’s exactly what we find.

We find that these moderates, they’re most willing to change their votes, change the party they support between elections, and they’re most responsive to things like the ideologies of the candidates, of course, but also non ideological characteristics that matter to them, like experience and incumbency. And I’m sure if we had better measures of quality, we could see this happening in other realms as well. But they’re the ones that are voting for the high quality candidates. They’re the ones that are contributing most to selection and accountability, because they’re ideologically close to being indifferent between the Democratic or Republican.

Matt Grossmann: So talk through that research a little bit because I know that people will be familiar with just, you look at people who say they voted for different presidential candidates and you find that correlation, but you have a more extensive analysis in congressional elections. So how do you show that these moderate voters are responding to these candidate characteristics?

Anthony Fowler: Yeah, we have data from basically a full decade or a full redistricting cycle of congressional elections of US house elections. And we also have some reasonably good data on characteristics of these candidates. We obviously know who’s an incumbent, we know who had prior political experience, and we have measures of the ideologies of the candidates as inferred from campaign contributions. But this is largely worked on by other people. But we can infer how far left or each of these candidates are based on who their contributors are. And so we have these different measures of the characteristics of these candidates and we can see how do voters on average respond to those characteristics. So we have our CCS survey data, we can construct a panel where we look at a bunch of voters in the same district, how they change their votes over time.

And we can see, on average, how responsive are people to things like experience, incumbency, ideological moderation, and on average there is a pretty strong relationship there for all of those. And then we can look across different types of people, and we’ve categorized these different types of people using our method where we’ve got say, downsient liberals, and downsient conservatives, downsient moderates are conversients and the inattentive group. And we can ask, how responsive are these different groups relative to one another? And by and large it’s the conversients and the downsient moderates that are the ones that are changing their votes and they’re responding to the characteristics of the candidates. Whereas the ideal ideologues are by and large, mostly supporting the same party the whole time.

Matt Grossmann: So this sounds like there’s true swing voters, but you have a supplementary analysis where you show it’s not just partisanship. So it’s not just that there’s independence or moderates that are always up, but there really is these policy views that are making the difference?

Anthony Fowler: I guess I’m not sure I fully understand the question. I mean, I think you’re absolutely right. There are swing voters and I think we’ve come up with a pretty good way of identifying who those swing voters are likely to be. I think a lot of people could have crudely just assume that maybe if you’re an independent, you’re a swing voter, and if you’re a self-identified Democrat or Republican, you’re not a swing voter. And that’s probably a reasonable approximation. I think that’s probably not perfect. I’m sure there are still plenty of self-identified Democrats who are pretty ideologically moderate and vice versa. And I’m sure there are some independents who are closet partisans, all of those possible categories. But I think you would do a pretty good job with party ID. Just trying to guess whether someone’s likely to be one of these swing voters or not. Of course, that doesn’t in and of itself suggest that party ID is important per se. It might just be that how people answer that part ID question corresponds pretty closely to their policy positions.

Matt Grossmann: So you also look at the demographic correlates of moderation and you find that women, racial minorities, young people, the non-rich and the less educated are all more likely to be moderates. So what does that demographic pattern tell you, how does that align with people’s perceptions of who’s a moderate in the US?

Anthony Fowler: Yeah, that’s a good question. I’m not sure if I had strong predictions about who would end up looking the most moderate demographically. I think by and large American politics scholars have spent a lot of time studying the correlates of being liberal versus conservative, and they’ve studied the correlates of participating more or less in politics. There aren’t a lot of studies on the correlates of being in the middle, but it could be that there is in fact a unique subset of people who are just much more likely to be in the middle. And it is interesting that it’s younger people that are more likely to be in the middle, racial minorities are a little bit more likely to be in the middle and so forth.

We could have maybe guessed some of this based on other things we knew about what’s correlated with being liberal versus conservative and what’s correlated with participating more in politics. But by and large, I didn’t have any strong expectations about this. I think that’s interesting, and I think demographics are always going to be a pretty crude measure of who actually comes out to be in the middle. But I think it would be great to see politicians exert more effort to try to identify those people in the middle and target their campaigns toward those people, because those are important people that should in fact be catered to.

Matt Grossmann: And I know didn’t do a lot with group identity as an alternative way that people might relate to the political system, but you also didn’t find any big differences in terms of who’s responsive to the incumbency or ideology of the candidates. It’s not that some demographics are less responsive.

Anthony Fowler: We didn’t look at that question specifically. We compared groups according to their ideology. We did not specifically look at other demographic groups. If you were trying to predict who is the most responsive, it could be that there’s other interesting ways to predict who’s the most responsive. But the one that’s the most theoretically motivated is the idea that people who are ideologically close, they should be the most responsive. And that is more or less what we find. Other demographics. If it’s true that other demographics are more responsive, I would guess that it’s largely because of their ideology and not because of other idiosyncratic features of that group. But I could be wrong, of course.

Matt Grossmann: So you also pursue a two-dimensional model and you find at least that there’s some people who may be better explained by that two-dimensional model, but not by a lot. So say a little bit about that analysis, if you’re able to define those dimensions at all, and what does it mean that we’re increasingly, or that we are at least currently unit dimensional?

Anthony Fowler: Yes. So like I said, roughly three quarters of the population is well summarized by one dimension, and then we have roughly another fifth that looks like this conversion category, that looks like they don’t fit the one dimensional model very well. And one question was how many of those people might just be roughly two dimensional people? Of course these are just models. Nobody’s exactly a one-dimensional or two-dimensional person, but we’re trying to understand to what extent can we understand people’s responses by these different ideological dimensions. And so, one of the things we did to assess that question was re-estimate a version of our model with two ideological dimensions instead of one. And indeed you do find that the estimated share of conversients goes down when you do that, suggesting that at least some of those conversients are probably people who are well summarized by two dimensions rather than one. But it’s not that their preferences are completely idiosyncratic.

And if you just look at some of those, the question responses, it does look like there’s a big chunk of Americans who are more or less described as being conservative, maybe not really conservative, but moderately conservative on social policy and moderately liberal and economic policy. They end up being classified as conversients by our model. But they’re pretty well summarized by that. That two-dimensional summary that I just gave you.

Matt Grossmann: And we’re observing this unit dimensionality potentially at the end of a process. There’s some historical research that suggests that the mapping of these two dimensions or other dimensions onto the one took a while and there’s debates about the extent to which it was driven by political elite coalition building or something related to partisanship. So I guess what do you think about that as, I don’t know if it’s an alternative explanation, but just the idea that you’ve observed it at the time when this fits together, but it fits together because of the way that political elite coalitions have fit these issue positions together?

Anthony Fowler: Yeah, that’s a great question. I think you’re absolutely right that for some of these issues, it’s not obvious how they should go together. It’s not completely obvious that your views on abortion policy would be so strongly correlated with views on say, tax policy or something like that. And yet that’s the way it looks in the US. Not entirely, of course there are these exceptions we just talked about, but there are these strong correlations. And a lot of people speculate that is because of elite behavior that people are taking cues from elites and party leaders and so forth. And I’m sure that is right to some extent. There is a lot of good research on this topic. I’m sure you’ve covered a lot of this research on this show. I would say that some of that research is mixed, that there is very clear evidence of what you might call learning and opinion change or issue projection, party Q effects and so forth.

There is clear evidence of that, although most of the evidence of that is on issues that are lower salience issues. So think of Gabe Lens’s great paper. One of his cases is on social security privatization during the 2000 presidential election. I think that’s a great paper. It’s a great design. That’s a great result. Social security privatization is not the most salient issue to most Americans, and that may be precisely the issue where you’d expect to see lots of projection or lots of following of leaders, whereas I haven’t seen any really compelling evidence that people are just doing whatever the party leaders tell them to do on say, taxes or things that are really core to their interests. Michael Tesla has a really nice paper where he looks for the same thing across a bunch of issues. And again, you see this opinion change on some issues and not on other issues.

And maybe it is well summarized by, well explained by what are the issues that people tend to care about the most. So I’m sure there’s a little bit of both. I’m sure some of these issues do just naturally go together in people’s minds. And some of the one dimensionality is a result of elites and cues and so forth that’s been happening over decades and it’s hard to tease out exactly how much. But I generally don’t look at the one dimensionality and say, oh, look at all these naive sheep out there who just do whatever the party leaders tell them to do. And one reason I say that is because almost nobody does what the party leaders tell them to do. Most people are in the middle. And so most people are in the middle on economics and they’re in the middle on moral questions and so forth. So almost nobody’s just naively following the party leaders. It just so happens that people who are moderate on one issue tend to be moderate on another issue too.

Matt Grossmann: So you also have this QJPS article where you go through some of this distinction between the possibility of policy voting and partisan voting. So why is it so hard to tell those things apart and what do you do to try to do so?

Anthony Fowler: Sure. Yeah. I mean this is such a big literature, this idea that partisanship or identity or some group attachment drives political behavior. It goes at least back to the Michigan School, at least back to research starting in the 1950s and so forth. And it’s, I think, been such a common finding that everyone just accepted it is at face value. This is of course, party ID drives vote choice. If you open up an American politics textbook, you’ll see that people just say, of course, your party identity determines who you vote for. And I think I’ve always been skeptical of this. It’s just never made a whole lot of sense to me. What are they even saying when they say this? And I’ve been at conferences where I’ve pushed people on this, and I think after years of doing this and being bothered enough, I thought like, okay, I’m going to write a review article about this.

And essentially the idea was let’s take this claim seriously and let’s look at all the evidence that’s put forth in favor of this what I call the partisan intoxication hypothesis. And by and large, that evidence is just not very convincing. It doesn’t distinguish between partisan identity driving vote choices and policy preferences determining vote choices. And the reason it doesn’t is, of course, that party identity is very correlated with policy. How do you answer the party ID question if somebody came along and said, which party do you identify with? You would of course just say whatever party you typically vote for. And the party you typically vote for is probably the party that you probably agree with most on policy. And so these things are all correlated with one another.

And then of course, political scientists observe, oh my gosh, party ID is so correlated with vote choice. It must be that people have this arbitrary attachment to identity. No, it could just be that party ID is just the party you normally vote for, that you normally agree with on policy. So that’s largely the point. Is that these are very difficult to distinguish between. And I think it’s irresponsible of us to go around saying, oh, look at these dumb Americans. They don’t actually have real policy preferences. They just go vote with their arbitrary identity, when we don’t have any evidence that it is an arbitrary identity.

Matt Grossmann: So the two of the cases that you focus on in that article, the 2016 election and especially the southern realignment, are typically chalked up to racial attitudes. So I wanted to get your sense of the relationship between what you’re calling policy views and maybe something that looks more like group prejudice, that they might both combine those two political and social views. And to what extent do you really think this finding that the southerners were conservatives and made their way to the Republican Party should challenge that widely held idea that the southern realignment is about racial attitudes?

Anthony Fowler: Sure. I mean, of course your racial attitudes do influence your policy preferences. So if there are a bunch of people who have taste based discrimination toward black people, that will affect their policy positions on civil rights issues and so forth. And so in some sense, those things are intertwined. So I mean, one thing to say about that is just because vote choices are… If it is true that vote choices are largely determined by people’s policy preferences, that doesn’t necessarily paint a normatively positive story. That could be that people’s policy preferences come from all kinds of bad places. So I’m not making a strong normative argument that I want to live in a world where only policy matters and nothing else, because of course, people’s policy positions might not be very enlightened. So there is that obvious point. But I think the Southern realignment is a good case that challenges the conventional wisdom of party identity, which is the South was solidly democratic for many, many years.

Southerners did identify strongly with the Democratic Party and surveys. They actually continued to identify strongly with the Democratic Party and surveys for several decades into the Southern realignment, but they switched and they voted for Republican presidential candidates when the parties change positions on civil rights policies. That seem to suggest that they are willing to change their votes when the policy positions of the parties change, even if their identity that they attached to the parties didn’t change. So that seems like a challenging case for the people who want to say that party identity drives vote choice. And so that’s a big reason why I talk about it in the book. Because it’s a rare opportunity where the national party platforms shifted so dramatically on an issue of importance to people so that you could actually see whether or not did people change their vote choices to go along with their prior policy positions, or did people change their partisan attachments to go along…

Maybe I get that wrong. Did they change their vote choices to go along with their prior policy positions? Did they change their policy positions to go along with their prior partisan attachments? So I think that’s a nice illustrative example, and there’s more to talk about there. A lot of people will follow up and say, oh, but those southerners still voted for Democrats for lower offices. They still voted for Democratic governors and state legislators and members of Congress. And that’s true, but that’s seems to be by and large because those other Democrats were pretty conservative Democrats, and so I have some analysis of that in the paper as well.

I look at voting in the US house, and I show that to the extent that Southerners continued voting for Democrats in the house, it was largely only voting for conservative Democrats. And as soon as those conservative Democrats retired and were replaced by more typical national Democrats, that’s when you see people really shifting toward the Republican Party congressional election. So it looks like policy plays a very important role. Of course, people’s policy preferences are not always enlightened, and they’re not always what we want them to be. I mean, of course people inherently disagree on policy. That’s the nature of the game. But the evidence clearly suggests that policy is playing a lot more of a role there than partisan identity, per se.

Matt Grossmann: And so it seems like both sides of the southern realignment debate have actually relied on partisan identity, the side that thinks that the change was driven by racism, talks about racial attitudes driving people toward the Republican Party, but even the side that doesn’t think that was true, talks about racial identity was what was attracting people to the Democratic Party in the first place. And so it was the softening of that racial tie. But it sounds like you want to say all this is put together and is not… You don’t want to try to distinguish the weather, racial attitudes were the one piece of it, or whether it’s a broader conservatism across different kinds of policy issues.

Anthony Fowler: No, that’s right. I mean, I know that there are interesting debates about that. I’m not taking up that particular fight. I think to the extent that… It must be true that your racial attitudes do affect your broader policy views and perhaps vice versa. And so I’m thinking of all of those things as being related to each other. And again, I’m not saying that voting on the basis of policy is always an enlightened thing to do. Of course, of course we disagree with people’s policy positions sometimes. I mean, I’ll say one other thing about the 2016 presidential election, which is of course that’s also there’s ripe debate about to what extent were racial attitudes driving people’s vote choices. There’s a lot of good work on that. It does seem like attitudes about immigration was one of the strongest predictors of switching from voting for Obama in 2012 to voting for Trump in 2016, and that was a big group of people.

And so it seems like racial attitudes are in some way related to that. And I do at least discuss that case in the paper. And I think it is important to say to academics that although all of that is right, and I’m sure racial attitudes did play a role, just because someone voted for Donald Trump, that doesn’t necessarily mean that they’re a racist or that they voted for Trump for bad reasons or for identity-based reasons, things like that. There are all kinds of reasons that somebody might have voted for Donald Trump. For non-racist reasons or for non identity-based reasons, including just they have different preferences over things like tax policy.

That seems like an obvious point. And yet I think a lot of people in the immediate aftermath of the 2016 presidential election looked at that and said, oh, it must be that America is racist, and it must be that they’re all driven by identity and they’re unthinking partisans and so forth. And I just don’t find that to be very compelling. And I think a closer look at the data and with maybe some years of reflection, I hope more people can agree with me on that.

Matt Grossmann: So in the initial converse discussion of ideologues, it’s not just that they hold consistent positions, but that they have some super structure that they can articulate that is driving their political views. And there are a lot of things beyond just ideological self identification that people have looked at, such as core values like traditionalism and egalitarianism or personality traits like openness to new experience. Where do these things fit in your view? You know, not just in policy voting, but in thinking about what a moderate or what a liberal or conservative is. Is there anything that is holding these views together and is there any piece in the middle that you think might be independently important? So just to, I guess, put out one model, even if you had a full spectrum of policy views, often you can still help predict vote if you add things to that like identification and some of these moral traditionalism and other kinds of super structure or things we use to think about or connected to those policy views.

Anthony Fowler: Yes, I’m sure all of that’s right. I mean, it must be the case that almost everybody has core principles of some kind. I mean, I don’t buy into this view that people really have no ideology or something. Of course, everyone has core views, maybe they can’t fully articulate and maybe they couldn’t tell you like, oh yes, I’m an egalitarian or whatever. But they have these core values that are in fact largely determining how they think about policy. And they might also be torn, of course, they could say, look, there’s these competing considerations. I can see where the conservators are coming from and this, and I can also see where the liberals are coming from on this, and that’s why I end up being in the middle. And I think a lot of those people that converse concludes are unsophisticated, are in fact just people in the middle.

I think a big reason why you observe a low correlation between at least a lower correlation for regular members of the public than for elites in converses study is because there are a bunch of people in the middle who they might have given the conservative answer had the question been worded a little differently, and they would’ve given the liberal answer would’ve been worded differently. So I think these core values are of course, important to different people and they’re turned up at different levels for different people. The liberals care more about some of them and conservatives care more about some of them. And the majority of people find themselves somewhere in the middle and they find both arguments to be persuasive a little bit. So I think that must be right. And to the extent that you can measure those core values better than we can measure them through binary policy questions, I would expect them to strongly correlate with vote choice.

And I think of that as all related. So a lot of political scientists seem to have this very narrow view of ideology means a nominate score or something like that. And it obviously is more complicated than that. Ideology means some combination of your core values and your beliefs that turn out to influence how you feel across a range of policies. The fact that ideology broadly is determining vote choices is different from saying, I can predict your vote choices perfectly with a single number. And it turns out the single number does a pretty good job as we show in our paper, but of course it’s more complicated than added. So I think I agree with everything you said in this question.

Matt Grossmann: And what about the alternative views of politics that Converse originally articulated? So some people think of politics as just who’s in charge and are they doing a good job or not? And some people think of politics in terms of which side is… Are the groups that I support on in the political system? Is some of what you’re observing in these alternative views or even the moderates, partly people who just don’t think about politics in terms of policy views but in terms of these other things that even they articulate on the open-ended responses?

Anthony Fowler: I think it must be that both things are important, that the ideological left right differences between candidates, those are really important. But also there’s valence characteristics that matter a lot too. There’s effort and quality and corruption and competence and all those things matter. I suspect all those things do matter to most voters, including to the moderate voters. It just so happens that the moderate voters are in the middle and they aren’t that happy with either the democratic or Republican Party most of the time on policy. And so they end up voting more on the basis of these valence characteristics. But both of them matter. And there is of course, in this era of high polarization and high divergence between Democrats and Republicans, there is a bit of a dilemma, because you can’t simultaneously always hold your politician accountable on effort and call quality and corruption and competence while simultaneously voting for the candidate that you agree with on policy.

You often have to pick one or the other. And that’s an unfortunate feature of elections. And maybe in a world in which there’s more convergence on policy than we can put more emphasis on these valence characteristics. And I think that’s one reason why a lot of people are worried about the extent of polarization in America today, is that at least polarization among the candidates is that we end up having to vote on the basis of policy because the policy divergence is so great when we would like to be selecting more competent, more qualified candidates.

Matt Grossmann: So overall, this seems like a pretty positive view of the American public and the American political system, at least to what I usually hear from political science, that voters have coherent views, not very many are unengaged and unresponsive. And the voters that are in the middle punish the candidates that are too far on one side or the other and vote on these other characteristics that we might think they should take into consideration. So how positive should we be about the American public and the political system, and are moderates successfully playing this balancing role in the system?

Anthony Fowler: Yeah, I’m pretty positive on the American electorate. I think that it’s a popular thing among academics to look down on the voters and say, look at the stupid voters. They did something I didn’t like. Which of course, I mean the voters also do things I don’t like all the time, but that’s how… That’s democracy and liberal society means that we sometimes we have a process and we do things that we don’t always like, and that’s better than killing each other in the streets. So by and large, I’m pretty happy with the American electorate in terms of they’re paying attention, they have real positions. Electoral selection and accountability seem to work more or less the way they’re supposed to theoretically. Of course, things could be better. I’m sure the electorate could be more informed than they are. I wish there wasn’t so much anger and vitriol on social media and so forth.

There are lots of things I wish were better, but by and large, when I look at the evidence, I often find that the electoral system is working better than a lot of other people think. And so I’m perfectly happy to defend it when the evidence goes in that direction. If I think of some of the big problems in American democracy today, one of the big problems that we’ve already been talking about quite a bit is the divergence between democratic and Republican candidates. So even though the voters are pretty modern and pretty sensible and reasonably informed, and there’s not much evidence that they’re totally irrational like a lot of people claim, I’ve written papers about shark attacks and college football games and things like that, and the voters turn out to not be as irrational or incompetent as people claim. Nevertheless, the result of the election can only be so good as the candidates going into that election.

And if none of the candidates are doing a very good job of catering to the voters in terms of policy, then there’s only so much the voters can do. And I think most of the evidence suggests that voters prefer the more moderate candidates when they’re given the option, but they’re almost never given a moderate option. And so in some sense, that’s one of the biggest puzzles in American democracy today. Which is why are the voters so moderate? Their vote for moderate candidates when they arise? Both parties could do better if they feel they’re more moderate candidates and yet they’re not doing.

Matt Grossmann: So you are also the co-host of the Not Another Politics Podcast, and are engaged in bringing political science to the public conversation. So give me a sense of where you think that is now. I know we’re speaking at a time when the political science has been pretty increasingly cited, at least, in media and public conversation, but Monkey Cage is leaving the Washington Post, Twitter is doing whatever Twitter is doing at the moment under Elon Musk. So what is our current role in the public conversation, and what are you trying to do to improve its role?

Anthony Fowler: Yeah, that’s a good question. Thanks for mentioning the podcast. So yeah, I am a co-host with Wioletta Dziuda and Will Howell, on Not Another Politics podcast. We’ve had you honest guests before. We really appreciate that. And we enjoyed doing it. We talk about research papers just like you do. I mean we were big fans of your show and we’re trying to do essentially the same thing. We’re trying to talk to researchers, talk about research in depth. We’re trying to be a little bit more in depth than say your typical NPR coverage of studies, that’s usually just at the very high level and very credulous. We’re trying to actually dig into the details and say, what does the evidence really say? What can we learn from this? And sometimes we disagree with each other, so I enjoy doing that.

It’s fun. And we get listeners, which is nice. I think a lot of our listeners are other academics, which is actually fine with me. I’m perfectly happy with that, because if we’re having a contribution that way by engaging academics in this conversation, I’m fine with that. And I think by and large, our most important job is to train students and to do good research and to hope that that research turns out to have an impact. But I don’t actually think of the public engagement part of my job as being the most important part of the job, which is good for me because that’s not the part that I’m good at anyway, I don’t think. But I try not to worry too much about that. My hope is that if we do lots of good research, that it will make its way into the public conversation.

Of course, none of that never happens as much as you would like or as well as you would like it to happen. But I think I’ve accepted that and I think I’m happy to just keep doing what we’re doing and hope that we make some progress. I will say that there is of course, always resistance among when I do talk to campaign people or policy people, advocates and politicians and so forth, there’s always resistance to your academic findings. Especially when they don’t go with what they were hoping was true, they don’t go with their agenda and so forth. But I think in the long run, you do see the kinds of lessons coming from political science as eventually impacting the conversation. Even in the last 10 years or so, I remember a lot of resistance from politicians and from campaign people saying, there’s no way that moderation is a good idea in elections.

You really have to stay to the extremes to make sure you turn out the base and so forth. And they would make these arguments very vehemently. And I think the amount of evidence that keeps coming in on this question has gradually changed the conversation so that you’re never going to get say AOC or Marjorie Taylor Greene to admit like, oh, yes, you’re right. It would be better if we moderated more. But I think more and more people are becoming aware of that. And I hope that in the long run that will change the calculus for politicians and for other policy people and so forth. And that’s just one example, but I think there are lots of examples like that. So the long and short of it is, yeah, sometimes it doesn’t go as well as you hope, and sometimes it’s unpleasant trying to get our research findings out there, especially if they’re not what people want to hear. But my hope is that in the medium to longer run that we are having a positive impact.

Matt Grossmann: And anything you want to tout about what is next for you?

Anthony Fowler: Sure. Yeah. Actually, I’m really interested in this puzzle of elite polarization. And so I actually have three working papers all related to elite polarization in some way, mostly congressional polarization. One of those papers is with Jeff Lewis, and it actually applies some of the things we learned from this paper on moderates and the public to studying members of Congress. We’ve assumed that the one dimensional model works really well for members of Congress because there are so many extremists in Congress and you do predict their roll call votes and so forth really well with that one dimensional model. But there are some exceptions to that. For example, the squad in the house looks like moderate Democrats on these standard roll call scores. And a big part of that is probably because they are not voting perfectly consistent with the one dimensional model. They sometimes vote like extreme liberals, which they probably are, and they sometimes engage in what we’re calling for this project protest voting, where they sometimes vote no on democratic proposals that they think don’t go far enough, or they want to signal their dissatisfaction with Democratic Party leadership.

And so we are estimating a mixture model, like we did for the public, but for members of Congress, where we allow for this possibility of protest voting, and we come up with what we think are better ideological scores for members of Congress, and also estimates of who and how often people are engaging in this protest voting. So that’s fun project. I have another project with Shufu on the extent to which primary elections are exacerbating congressional polarization. And then I have another project related to party leaders and change in constituencies and to what extent that is explaining polarization and to what extent those things are explaining the rise in polarization over the last 50 years in Congress.

Matt Grossmann: There’s a lot more to learn. The Science of Politics is available biweekly. From the Niskanen Center, I’m your host, Matt Grossman. If you like this discussion, here are the episodes you should check out next, linked on our website. If moderates are electable, why are ideologues winning? How donor opinion distorts American democracy. How political values and social influence drive polarization. Is demographic and geographic polarization overstated? And reducing polarization with shared values. Thanks to Anthony Fowler for joining me. Please check out moderates and partisan intoxication or policy voting, and then listen in next time.