Are political elites in bubbles, out of touch with the American public, not recognizing how their views and conditions are not reflective of most people’s experience? Prior research found that elites tend to overestimate conservative policy positions in the American public, but there are wider misperceptions across the political spectrum. Alexander Furnas finds that unelected political elites—from government officials to lobbyists to media figures—all assume that public opinion more closely matches their own opinions than it really does. Adam Thal finds that politicians overestimate the level of financial struggles facing constituents. But correcting those misperceptions does not change their opinions.

Guests: Alexander Furnas, Northwestern University; Adam Thal, Marymount Layola University

Studies: “The People Think What I Think“; “Do Political Elites Have Accurate Perceptions of Social Conditions?


Matt Grossmann: When elites misperceive the public, this week on The Science of Politics. For the Niskanen Center, I’m Matt Grossmann. Political elites are often accused of being out of touch with the American public, not recognizing how their views and conditions are not reflective of most people’s experience, and caught in their own bubbles. Prior research found that elites tend to overestimate conservative policy positions in the American public, but today’s episode focuses on wider misperceptions across the political spectrum. This week I talked to Alexander Furnas of Northwestern University about his new paper with Tim Lapira, The People Think What I Think. He finds that unelected political elites, from government officials to lobbyists to media figures, all assume that public opinion more closely matches their own opinions than it really does. It’s not just conservatives whose perceptions are off. Everyone overestimates how much the public agrees with them on policy issues.

And elites might misperceive not only policy opinions, but also the circumstances the public faces. I also talked to Adam Thal of Loyola Marymount about his new British Journal of Political Science article, Do Political Elites Have Accurate Perceptions of Social Conditions? He finds that politicians overestimate the level of financial struggles facing their constituents, but correcting those misperceptions does not really change their opinions. Let’s start with my interview with Furnas, which develops from research suggesting politicians have biased perceptions of public opinion. So tell us about the biggest findings and takeaways from your latest paper on elite perceptions of public opinion.

Alex Furnas: I think the thing that was really striking to me about these findings is how consistent they are. So essentially, we find that across a whole range of policy issues, elites believe that the public’s opinion is more in line with their own opinions than they actually are. Essentially, we asked the public in a likely voter survey that we conducted, whether they support or oppose 10 different policy areas. And then we asked a set of elites in a similar survey, we asked them both whether they supported or opposed these policies and what percentage of the population they believed supported those policies. And we found that for policies that elites themselves strongly favored, they overestimated public support by about 12 percentage points. And for policies that they themselves strongly opposed, they underestimated public support by about 12 percentage points. So there’s sort of a 20 to 25 percentage point difference in elite evaluations of public support for policy, depending on whether the individual elite strongly supports or strongly opposes that policy.

If they sort of weakly support or weakly oppose, their over underestimation isn’t as strong, but it’s in the same direction. And so we found this is true across these policy issues. It’s true across different types of elites. So it’s true for folks who work in political media, it’s true for lawyers and lobbyists, it’s true for law clerks and judges. It’s true for state and local elected officials and state bureaucrats. It’s true for bureaucrats in the federal government. Across a whole bunch of categories of political elites, it’s true. And then we did a bunch of things in this paper to try to say, where might there be heterogeneity here? Is this effect stronger among strong partisans and less so among weak partisans because maybe there’s some sort of kind of partisan echo chamber here where they think, “Oh, I’m a really strong partisan and these are policies that have some partisan valence to them. And so I’m kind of anchoring on my own co- partisans.” And nope, there’s no difference based on the strength of partisanship.

One thing that we were really interested in was about how there are different professional norms across categories of elites about how important one’s partisanship is to their professional identity. If you’re a congressional staffer or you are a lobbyist that has strong revolving door ties to the Democratic Party or something, there are a bunch of types of elites where being partisan, if you’re working campaigns, being a member of a political party’s part of your job, and there are folks like law clerks or often political journalists who for them, they may have a private political identity or partisan identity, but there’s a professional norm that that is separate from what your job is. And so we developed a question battery to measure that and thought maybe these effects are stronger among people whose professional partisan identity is really important to them. And there’s no difference there either. We kept coming up with sort of…

Another was, well, maybe if you really trust copartisan information sources and really distrust information sources from the other party or sort of other ideological perspective, looked at the difference in your kind of partisan information trust. Because we didn’t actually have a direct measure of information use, so we couldn’t kind of directly measure kind of your information echo chamber stuff. So we looked at the differential trust in partisan information sources to see if you’re kind of more insular in that kind of way, does that lead to stronger effects? And again, it doesn’t. It sort of just, every way we sliced it, this what we call a false consensus effect. And it’s not our term. That’s a term out there from the literature. But what we find going on here is essentially this really persistent and pretty substantively large false consensus effect where elites believe the public believes what they do.

Matt Grossmann: So tell us about the prior findings in this literature that are mostly about politicians. It seems like this one is confirmation of part of that literature, that there are these biases based on people’s own opinions, but less confirmatory of the other finding, which is that there’s usually an overall conservative bias in public opinion, at least among politicians in the US and elsewhere. So how does your research fit into that and how are you thinking about it?

Alex Furnas: Yeah, so that’s a great question. We were certainly inspired and motivated by some of those studies. There’s the Brockman and Skavrone piece and the Miltenberger Hurdle Fernandez Stokes, et al. piece. And they have tended to find, and there’s been other work there too, those studies have shown this sort of conservative bias in perception of public opinion. And those have been studies of elected officials or the Hurdle Fernandez, et al. paper was congressional staff, but that’s a very similar information environment to electeds. They’re sort of serving as agents of elected principals. And what we looked at here was unelected elites. So we’re looking at a different population and we don’t find that same kind of directional bias. When we look at aggregate differences across all issues, we find that Republicans overestimate conservatism a little bit and Democrats underestimate a little bit, more in line with the directional bias of their own opinions. And I think that there’s more work to be done here to try to reconcile these findings.

My hypothesis, which we don’t test here, but I think is an interesting one for future work, is that there are important differences in the information environments that electeds operate in compared to unelected political elites. And so in particular, I think, elected officials have various channels to get feedback directly from the public, and these are the calls that their offices get, letters that are written to them, who shows up in town halls that they hear from. There are a bunch of ways that it’s part of their job to hear from the public. And that’s not true for unelected political elites, for the most part.

They have reasons to want to care about public opinion because they want to know about the potential feasibility of policy recommendations they’re going to make. Or if they’re advocating for particular policies to elected officials, if they’re lobbyists or interest groups or something, they want to be able to try to speak with some authority about the political landscape that this is all operating in. But they don’t have those same feedback mechanisms unless they’re commissioning polls or something, which is relatively rare because it’s a resource intensive task.

But I think that the types of feedback mechanisms that elected officials have, and this is stuff that’s well documented in the literature, tend to have biases on who shows up and how vocal different communities are. So being politically engaged and vocal, it correlates with free time and resources. And so you tend to get older, whiter, typically a little bit more well off people showing up to meetings or being more engaged in those sorts of ways. And those populations skew conservative typically. And so I think that elected officials have these really important mechanisms of getting feedback from the public, but the sample of messages they get is from a population of people that skews conservative and that that’s a plausible explanation for why then you see a kind of conservative tilt in electives’ estimates of public opinion.

I also think it’s certainly possible that some of the differences we’re seeing here are idiosyncratic based on the selection of issues that we happen to poll on. In our study, there were more sort of issues coated with the Democratic Party than Republican Party. I think it was like eight and two or something. It was definitely more. And we do subset the results by whether it’s Conservatives responding to Democratic coated issues or vice versa. And we still see consistent results there. But I do think it’s a non-random set of issues that we pulled on.

In part it was a little bit hard in the study design to think of issues to pull on that were Republican because we were designing this around the time when the Republican Party didn’t have a party platform, still doesn’t have a party platform officially. And so sort of not entirely obvious to me what policy proposals to. It’s not a super policy proposal rich ecosystem on that side right now. But that could also be part of what’s going on here. But that’s how I think about the relationship between these findings, but I think there’s more work to do.

Matt Grossmann: So tell us about these categories of political elites that you surveyed, why it’s important to understand their views and how you got ahold of them.

Alex Furnas: Sure. So we used Leadership Connect as the contactless provider for the survey that we ran. Leadership Connect is a sort of online directory that often serves for government affairs, political intelligence, kind of relationship management stuff for corporations and lobbying organizations and things like that. But it’s grown out of what used to be the Yellow Books. And so these were who used to be in the Congressional Yellow Book, the Bureaucracy Yellow Book, these kinds of things. And now they’ve created one sort of integrated contact list platform. And so that’s what we used as our contact list. We went through that, and they essentially have eight verticals of what they call communities. And we essentially used their typology, and so it’s lawyers and lobbyists.

This [inaudible 00:13:05] does not use the Congress because we were looking at unelected elites, but they have a Congress vertical, which is members and mostly staff. So lawyers and lobbyists, courts, that’s judges and clerks, federal bureaucracy, associations and nonprofits, which that’s where a lot of think tankers and things are. It’s not all associations and nonprofits in the country. We were looking at sort of D.C., Maryland, Virginia based. Their whole focus of the directory is on folks adjacent to government relations. And so that’s where we started. We went through their directory and picked job responsibility and job title categories that we felt were sufficiently senior. So these are people who head programs, there’s also a corporation, these are c-suite people or people who head the government affairs departments of corporations, executives at nonprofits. For the sort of nonprofit think tank vertical, we included research fellow type of things.

And so yeah, we went through for each vertical and sort of set a threshold of what’s the level of seniority and job responsibility that we think falls within the scope of what we’re considering political elites here. And that’s all kind of stuff that’s detailed in the appendix of the paper, of course. And that left us with a pretty good contact list that we reached out to. And we think that this is a really important population because there’s a lot of agenda setting that happens in federal policymaking. Only a few things make it onto the agenda at any given time. And policy windows open idiosyncratically, and then people try to move on things. And so there are lots of folks in the background kind trying to push things onto the agenda, making the case that things are important or sort of building up portfolios of evidence or policy proposals, possible solutions that can go into the garbage can model when that window actually opens.

And we think the opinions of those folks doing all that stuff, even if they’re not the ones actually doing the voting or informing those who do the voting, the actual legislators, when the time comes, they’re all sort of creating the information environment in which that’s going to happen and making the case for why some things are important and others aren’t, or why some solutions are considered reasonable and things we should be looking at, and other solutions aren’t. And so what those folks think are a big part of that, we believe. And so it matters then if those folks have a reasonable perception of what the public is going to want or not.

If we expect or hope for some form of Democratic responsiveness, all of the parts of this policy knowledge, production advocacy, ecosystem or environment, all of those parts are combining to make that happen. And if some of them have really distorted views of what the public wants, we should have lower expectations for responsiveness. And we see all the time, pundits. And pundits are like a subclass of people that show up in this data. Political journalists and opinion writers were part of our sampling frame. And all the time we see in major media outlets, pundits making the case that something is radical or something’s too far, is not a reasonable policy option, or the voters won’t like it, or we should be paying more attention to what Trump voters think about this, or what Biden, who are the constituencies we should be listening to.

And those are all informed by their implicit assumptions of who is the public and what do they think and what do they believe. And it turns out a lot of those people probably think the public believes what they believe or much closer to what they believe than they actually do. So we should maybe be a little incredulous when folks are making those kinds of claims that we see made in public a lot.

Matt Grossmann: So one explanation for this is elites aren’t any different than anyone else, and this is just a classic of human psychology. Is that your view or are there other reasons why these elites, maybe their personal incentives or their social networks might be either particularly bad at this or might have something that goes against another mechanism that might’ve moved them closer to the truth?

Alex Furnas: Sure. Yeah, that’s also a really great question. I think it’s a little bit of both. I’m generally of the belief that elites are people like everybody else, and we all have basically the same set of cognitive tools. And we’ve developed various through evolution and through… We use these heuristics because our information environments are too complicated and we’re cognitively bounded. And so we engage with the world through a bunch of various kinds of heuristics, and those lead to biases, sort of no matter what, and we got to work with what we have. And that elites are like everybody else. We see in the literature there are false consensus effects observed in the normal population all the time. In some sense, I don’t think this is any different than that.

I do think that there is sometimes a belief that elites are different, that they’re more highly educated, they pay much more attention to these kinds of issues than the general public does, and so we should maybe hope or expect for better from them. Their perceptions of public opinion are probably better than the… We didn’t ask the public to estimate how popular these things were in the public, but elites are probably better than the public, if I had to guess, although still pretty bad. But sometimes the sort of higher political sophistication or higher education, various things elites can push in the other direction. They can be better at rationalizing their own beliefs. They can be better at coming up with counter- arguments for counter attitudinal information. And so I think it is a little bit of an open question to me how those play off against each other. They’re certainly more aware of salient policy issues and paying more attention than the average voter, but they’re also, I think, better at telling themselves compelling stories to justify their own positions.

Matt Grossmann: So you said you didn’t quite find a coherent kind of, here’s who’s bad and here’s who’s better at these kinds of perceptions, but what differences did you find across issues or across parties or across occupations?

Alex Furnas: Yeah, so really it was mostly a story of no differences across parties, occupations, political information preferences, professional norms. It’s really a quite universal effect, is what we tended to find. So when we look at the variance by issue here, or differences in how accurate folks are, both Democratic and Republican elites underestimate how popular clean energy subsidies for low income folks are, but overestimate the popularity of a carbon tax, for example. They overestimate the popularity of a pathway to citizenship, both Republicans and Democrats, and underestimate the popularity of a wealth tax. For most other issues that we pulled, we see Republicans underestimating the popularity of a progressive policy and Democrats overestimating and vice versa. Usually we see kind of partisan asymmetry there where people overestimate copartisan opinions and underestimate the popularity of out-partisan policies. But for those couple issues, both Democrats and Republicans are wrong in the same direction.

Matt Grossmann: So the other interview is about elite misperceptions of economic conditions that constituents are facing. And so most of the literature has been about this kind of beliefs of policy positions, but is this kind of a reason to expect that perceptions might be off in other areas as well, that elites just may be not as well- informed about not just what the public thinks, but the actual circumstances that they’re encountering?

Alex Furnas: Yeah, totally. I would expect you to see this in elites’ perceptions of economic conditions or of their beliefs about the economic perceptions of the public. If you’re an elite who believes the economy is doing well, I bet you overestimate how much people think the economy is doing well or overestimate the particular economic conditions that some sample of folks are in. And I think that’s certainly really timely in the sort of really low unemployment, falling inflation, Biden boom. And yet consumer sentiment is low and public polling about the state of the economy is poor despite a bunch of indicators looking really good. And so I bet you have sets of elites, probably partisan elites, who believe the indicators are more right, or who believe the consumer sentiment or that kind of polling stuff is more important, than you would see their perceptions diverge similarly.

Matt Grossmann: So you have also done a lot of work on legislative staff, and some of it tries to explain some of these previous findings among elected officials. So tell us about that and how you’re thinking about information environments that political elites have as a possible reason for some of these biases.

Alex Furnas: Sure. Yeah. So I’ve done a couple of surveys prior to this one. Did a few surveys of just congressional staff and asked them a bunch of questions about what information they use and trust and a bunch of other things about their political knowledge and their career trajectories and things like that. And a theme of my work in those areas is that I’m particularly interested in differential use and trust of partisan information sources. So to what degree are, in this case, congressional staffers, often committee staffers who are essentially tasked with providing their members with policy briefs and questions for committee hearings or recommendations on how to vote, these types of things, how much are they drawing from just ideologically consonant sources? And to what extent is that maybe part of polarization that we see and things like that. And we do find consistently really strong selection of partisan sources over non-partisan sources or certainly over out-partisan sources.

And that’s true both at the aggregate when we look at just categories of sources and also when we kind of look at individual sources, look at specific think tanks and things like that. We’ve done that both sort of observationally, just asking batteries of how much do you trust or use these different sources? And also in survey experiment settings, we’ve done sort of a vignette experiment and a couple conjoint experiments, where the conjoints aren’t published yet, but the vignette was at BRQ, I think, a couple of years ago. And yeah, there’s a sort of large effect of how much a staffer is likely to use an information source if it comes from no line source.

In the conjoint experiments, we manipulated both the ideology signaled by the source, and also the ideological content of the message. And we find that the results are, it’s about the same magnitude of seeing a consonant message or a consonant source, and they operate pretty independently of each other, as about 15 percentage points more likely to choose that source. These were hypothetical white papers that staffers were choosing between.

And we have some more work that we’re designing right now to sort of get into those mechanisms a little bit more. I’m particularly interested these days on, most of my previous work was on think tank type, sort of beltway institutions, including nonpartisan CRS, CBO kinds of sources and the trade-offs between them. But I’m now really interested in the use of scientific evidence specifically. I’ve been looking at that a lot and finding really stark differences in the rates at which Republican and Democratic factions use science in their policy materials, in committee reports and in think tank documents, the rate at which they cite science. Democrats use science much more, and the difference is growing over time. And those results are really universal, actually. We see them across all areas of science and across all policy areas. That’s true. Some future work. Hopefully submitting a paper today detailing those results from a large corpus of policy documents.

Matt Grossmann: So part of the proposed explanation for some of the previous findings among elected officials was that these networks among staffers and the reliance on conservative sources among staffers, that might inform their elected officials, but there has been a change in the infrastructure on the left as well. So I wonder if you would comment on the possibility that maybe the information environments have become bubbles on both sides now as one potential explanation for your findings. That is maybe these people are just seeing a lot of issue polls where their side wins because there’s a pretty big infrastructure for doing that.

Alex Furnas: Yeah, I think that’s absolutely possible. So another sort of working paper that we’re doing in my center right now is looking at citations to think tanks between each other and over time. And we see that sort of liberal think tanks cite non-partisan, or neutral think tanks. I mean, obviously these boundaries are a little bit squishy. But there’s much more sort of policy to policy citations between liberal and neutral or centrist think tanks than there are between conservative and centrist or neutral think tanks. So I still think, at least insofar as the kind of think tank network is indicative of the information infrastructure environment, and it’s obviously only part of that world, but at least insofar as that’s concerned, I think the ecosystem on the right is still a little bit more isolated.

But I think you’re right to note that that trend, that that is less so now than it used to be, that there is a little bit more of this separation. And I think that both sides seeing issue polls that tend to support their own conclusions could be part of the story that we have here. In this study, we do look at, it’s individual variation, not just party. For that to be entirely the explanation, everyone would just have to have consonant partisan opinions. And there’s a decent amount of variance in there because we’re exploiting within respondent variants across issue areas, because there are individual fixed effects, essentially, in the models that soak up kind of the partisanship component.

Matt Grossmann: So one potential implication of that is that you could have some issues where it isn’t just elite partisan disagreement, but elite disagreement with the public where elites would not necessarily perceive that agreement. So classic examples are on foreign policy like free trade, where elites on both sides are more in favor than the public on each side. So any sign of that in the data and any comment on that implication? Because it’s not just pure polarization in your paper, it might mean that elites are insulated from public opinion even when there’s kind of an elite opinion that’s different than the public opinion.

Alex Furnas: Yeah, I think that’s totally right. I think that’s what we see for issues like carbon tax or wealth tax, is that elites are all on the same side in their misperceptions of the public. And I bet if we tried to select only issues where we believe that that’s true, and so some of these foreign policy issues are a great example as well, I bet we would still see the same results that we see here where they would all overestimate things that they themselves support.

I think we see more support for a carbon tax, they underestimate the wealth tax support because they mostly don’t support wealth tax and they overestimate carbon tax support because more elite support carbon taxes than the public. And that’s true in both parties. Democrats relatively more, but even Republican elites more than the public because that’s a kind of market mechanism for dealing with a problem that tends to be appealing to elites. So yeah, I think when it’s at the level of individual support for a policy leading to overestimation of public support for that policy, that can happen regardless of partisan differences. It can be everybody supports it more or one party supports it more and one party supports it less.

Matt Grossmann: Anything we didn’t get to that you wanted to include or anything you want to tout that you’re working on next?

Alex Furnas: I mean, just what I mentioned right now is I’m really excited about looking at sort of the use of science in policymaking specifically, and we’re finding some really striking results about those partisan differences and also elite differences in trust of science and scientific institutions as well. So keep an eye out for some of that stuff. I’m at this really interesting center for Science of Science & Innovation in the business school at Northwestern, and we’ve been working for a couple years now on a research agenda of sort of the intersections between science and politics and the science of science and politics. And so we have several papers that are close to fruition now that should hopefully be sending out. So yeah, just stay tuned for that stuff if that’s of interest.

Matt Grossmann: While Furnas pursues the traditional analysis of policy opinions, Adam Thal wondered whether politicians are out of touch in another way. Maybe they don’t know the economic problems their constituents are facing. But the results surprised him. They actually overestimated those economic problems. So what are the biggest findings and takeaways from your new article on elite perception of social conditions?

Adam Thal: Sure. So I think the big takeaways relate to this idea that a lot of people have that politicians are out of touch with reality, particularly with respect to economic problems. I think we have this idea that politicians don’t understand a lot of problems faced by low income people in the United States. I don’t find that that is true. So I find very little evidence that politicians underestimate problems faced by the poor. If anything, they tend to overestimate the scale of those problems.

Matt Grossmann: So tell us more concretely what you were up to here. Who were you talking to? What did you ask them and how did you confirm this view?

Adam Thal: Sure. So I started off trying to understand part of what might explain broadly speaking political inequality in the United States. So we know we have this pattern with policymakers where they tend to make policy decisions that benefit the rich and go against the interests of the poor. That doesn’t always happen, but it does happen maybe more often than not in American politics. And so I wanted to help us understand why. Why do politicians sort of ignore problems facing low income people in the United States? And so I wanted to basically look at one potential mechanism that could help to explain that, which is the idea basically that politicians are out of touch, that they do not understand the scale of problems affecting low income people in the country. I think there’s good reasons to expect that that would be the case.

Politicians on the whole are much more affluent than the average person. And so there’s good reason to think that because they don’t experience a lot of economic problems themselves, they might underestimate the scale of those problems, not really understand how bad things are for low income
families, and maybe that could explain why they’re not doing all that much to address these problems so that was the theory.

To actually test that theory, I worked with a number of colleagues who were working on their own projects to field a survey of state legislative candidates. So these were politicians basically running in primary elections in 2018 for state legislative offices. And I basically surveyed these folks and I tried to do two things. One is I tried to measure their perceptions of the scale of problems affecting low income people in the states where they’re running for office. And then I also ran this experiment to see if we correct those perceptions and get them more in line with reality, will that change how they feel about policy?

Matt Grossmann: What is the kind of level of economic insecurity that Americans were facing in these places and what were the levels expected by the politicians?

Adam Thal: Sure. So I looked at three different problems. So one is the level of financial insecurity, another is their ability to access healthcare, and another is their ability to pay for higher education. So all three of these problems, generally speaking, are reasonably severe. So we can focus just on the problem of financial insecurity. So there’s a statistic out there that 40% of Americans don’t have enough money on hand to cover a $400 emergency expense, so they’re totally unprepared for any kind of financial emergency. So the actual level of these problems is quite high. And I asked politicians sort of what are their perceptions of these problems in the state to get a sense of whether they understand that reality.

Matt Grossmann: And then before we get to the experiment, let’s talk a little bit more about how this fits into previous work. So a lot of previous work on elite misperceptions, including another interview for this episode, is about misperceptions of policy positions of the public. And it has found, as you know, that elites kind of think the public thinks more like they do, and also in some cases that there might be a conservative bias in their perceptions of public opinion compared to real issue opinion. So kind of fit what you’re doing in with that literature, and do you think it’s kind of consistent or inconsistent with the findings there?

Adam Thal: Sure. So I was very much inspired by that literature, which shows that politicians, including the same kind of politicians that I’m serving, like candidates for state legislature, they tend to misperceive public opinion and they have this conservative bias. That kind of inspired this paper, I would say. So I expected to find something similar, which would be in this case, that politicians potentially underestimate the severity of problems affecting low income people, and that that could help to also explain a potential conservative direction in their views of social policies. I did not find that to be the case. So I think it’s inconsistent with that literature in the sense that that literature shows misperceptions that lead to a potential conservative direction in policymaking. I think that politicians’ perceptions of something different, like the reality in their state in terms of these economic issues, is potentially more accurate. I don’t find large scale misperceptions. And so it basically doesn’t seem to translate to this other part of their perceptions of the world.

Matt Grossmann: But there was one part that seemed to match, which was that there are partisan differences and they track how much the parties want to do about these issues. Is that right?

Adam Thal: Yes. So the part that does match up is, so I think that the other literature on perceptions of public opinion does suggest large partisan differences. So that could help to explain elite polarization. I definitely do find that too. So there’s definite polarization. Democrats in particular tend to really overestimate the scale of these economic problems. Republicans are less likely to overestimate them and in some conditions underestimate them. So there’s definitely this sense where Democrats and Republicans have this different perception of reality.

Matt Grossmann: And part of the literature looks at elite misperceptions as just similar to public misperceptions. We all have cognitive biases and elites would have them as well, and part of it kind of expects them to be different for elites than the public, that they’re in some kind of information environment that might be leading them astray. Anything to say about that? Is this a case where actually the biases kind of match in the public? We think we’re doing better than most people economically, so it might sort of match a basic public misperception?

Adam Thal: So I have looked a little bit at this in the mass public, and generally speaking, I found pretty similar results, which is to say that among the mass public, I also don’t find some major tendency to really underestimate these problems. And there’s the same partisan divide where Democrats think they’re worse, Republicans think they’re better. So I do think that elites are maybe not so different from the general public in this sense. I will say that there’s one important exception to that, which is I did find this one instance where Republican politicians underestimate how many people in their state are financially insecure. That was kind of the one finding I found that aligned with my theory, and I was able to provide some evidence about what might produce those misperceptions, which is that they may be rooted in Republican politicians’ social networks.

So politicians, including Republican politicians, are more affluent than normal people. Politics also creates this environment in which, if they want to win elections, they have to be with donors all the time. They’re incentivized to spend time with rich people too. So they’re affluent people who are very incentivized to spend time with the affluent. And I find that there’s a sort of correlation between what they’re seeing in their social networks and their tendency to underestimate financial insecurity. So it does appear in this one instance that this might be an elite thing where elites are in this bubble and that leads them to underestimate this particular problem.

Matt Grossmann: So we’re speaking at a time when there’s a big divergence between objective economic indicators and people’s subjective feelings about the economy, and there are some people who say maybe actually the subjective feelings are worth giving more weight to. So I know you’ve asked people to give estimates of things that we have real data for, but is it possible that the subject of estimates are also worthwhile and maybe they’re just thinking about general problems that people are having with healthcare or education or economic security that may go beyond the scope of the data that you’re comparing it to?

Adam Thal: Yes. So basically, if I were to have to make a choice, what do I think matters most for these people’s policy views, their subjective perceptions of the problem or the actual reality, I think it’s the subjective perceptions of the problem. If you were to look at this data, you would find that their subjective perceptions are really highly correlated with their policy opinions. It just sort of makes sense. If people have to make these policy decisions in their head, what’s going to shape those decisions is their subjective view of the problem. It just so happens that in this particular circumstance, the subjective perceptions are often not that far off from the reality.

Matt Grossmann: And just to re-ask that one more way. Is it possible that actually, I won’t say the objective data is wrong, but incomplete in such a way that maybe they should perceive it as a little bit more severe than the objective data do? Maybe they’re just thinking about problems that people have in accessing healthcare or higher education that are more severe than the data might [inaudible 00:44:49].

Adam Thal: So if we think about my findings in relation to what’s going on right now, which is this idea that based on what economists are saying, the economy is doing fairly well, that people don’t perceive that, they’re really waiting, sort of the negative news that they see, and perceive things to be potentially much worse than they actually are. It’s interesting that I find something similar here. People have maybe just this negative prism on the economy and tend to just overestimate how bad things are. And it’s interesting that that’s happening both across elites and I think we see it now with voters in the election.

Matt Grossmann: And were there any differences across your three areas that are worth highlighting, or was it pretty much the same consistent findings across areas and the people that you were asking?

Adam Thal: Sure. So there were some interesting differences. So I think the most interesting thing to think about is that of these three problems, which related again to financial insecurity, unaffordable healthcare, and access to higher education, the one that politicians were most likely to overestimate with was access to higher education. I think why that’s potentially interesting is we might think of that one as being the most affluent-

Matt Grossmann: They overestimated the problem or they overestimated the access?

Adam Thal: They overestimated how difficult it was for people in the state to access higher education, and they thought that people were going into more debt to get into college than they actually were. So they overestimated the scale of the problem. I think what’s potentially interesting about that is of the three problems, that is the most upper middle class problem that I measured, because it’s who is most likely to attend college? It’s upper middle class affluent people. And so it kind of aligns with this social network perspective. I can easily imagine that that is the problem they’re most exposed to in their social network. And maybe that’s why they were the most likely to overestimate the scale of that problem.

Matt Grossmann: So you talked a little bit about the media environment as one source of misperceptions, and I know that there’s been some work on that before, but what are the other candidates? Is it plausible that the media environment is just too negative in all three of these areas? Is there a reason why people are better able to update in these areas than others, or is there another source of these misperceptions?

Adam Thal: Sure. So I think one way to think about that is politicians are more affluent than the average person. So a question we can think about is how might they learn about these problems anyway, even if they are not experiencing them? So I think the primary candidates there would be the media environment, but then also potentially hearing from constituents. We might expect that Democrats would be especially likely to hear from constituents who are struggling financially. So that could explain why they are most likely to overestimate these problems. I think those are probably the two main ways that they actually learn about these things, like the media, through their constituents.

I think also it’s useful to consider sort of how psychological biases might play a role in leading them to overestimate the scale of these problems. And it kind of gets back to this idea that maybe if you’re thinking about this and you’re uncertain about what the right answer is, there’s sort of less cost to potentially overestimating the problem than underestimating the problem. So if we underestimate the problem and we don’t do anything about it, that’s potentially a really painful negative thing for society. But if we overestimate the problem and maybe do too much, that seems less painful and potentially less negative for you as a politician. And so there could be some sort of psychological bias that I’d rather overestimate than underestimate.

Matt Grossmann: And is that a reasonable normative take as well? That on the one hand we want people to have true views of the world, but on the other hand, these are problem solvers, so maybe it’s good that they overestimate or at least are very attentive to people having problems?

Adam Thal: Yes. So I think it’s like a two part answer. I think at a baseline, like normatively, I’d rather have politicians overestimate the problem than underestimate the problem. But I think the second part is that I think the consequences of overestimating the problem might be complicated. So we can imagine one world in which politicians overestimate these problems and maybe do too much, but that doesn’t seem to be the actual world we’re living in. We wouldn’t say that politicians are doing too much in the United States to address these problems.

So another possibility is that you could overestimate the problem and then come to the conclusion that it’s unsolvable because it’s so bad in your subjective perception, so you don’t do anything about it. So you could potentially see this link between thinking the problem is worse than it is, and then concluding that it’s hopeless. And so maybe there’s this kind of equilibrium level where they don’t sort of go too high, don’t go too low, and that allows them to think it’s in this sweet spot where, “Oh, we could actually address this,” and maybe overestimating it leads to this conclusion that it’s hopeless.

Matt Grossmann: So let’s talk a little bit about the experiment because you’re able to kind of observe what happens when you inform people more. And it did seem consistent also with the elite perceptions of public opinion research, which also finds that even if you think these are a source of opinions of elites, it doesn’t seem to change their opinions when they are given more information in the course of the survey. Is that right?

Adam Thal: Yes. So what I did in my particular study was after asking them their perceptions of these issues, I essentially gave them accurate information. I said, “Basically, you said that you thought the scale of the problem was this, but the actual scale of the problem is this.” And then I measured a range of policy views that relate to potential solutions to the problem. “So after being told that you misperceived financial insecurity, does that change your views of increasing spending on welfare?” Broadly speaking, I did not find that it changed politicians’ policy views. And so the main conclusion is, they misperceive these problems to some extent, but even when they have the accurate information, it doesn’t appear to change their views, which suggests that to the extent that these misperceptions exist, they’re not a major determinant of their policy views.

Matt Grossmann: So that should cause us to reevaluate the role of this in policymaking. Is it fair to say that we kind of consistently think these things are inputs to people’s policy positions, but often they follow more from the policy positions than actually change them?

Adam Thal: So I think based on this evidence, it sort of speaks to the idea that just like having an accurate understanding of the problem is not going to shift their policy views one way or another. So it suggests that their subjective perceptions of the problem may be less important, for example, than their perceptions of public opinion, their own personal views of is this a good policy or a bad policy. What helps them get reelected essentially might matter more than their subjective perception of reality.

Matt Grossmann: In other studies where people are asked to estimate things, there’s often just basic problems with numeracy. In particular, for example, people overestimate the percentage of people who are X, Y, and Z in small populations in the public. Is it possible that part of it is just we’re expecting too much precision or understanding of what these kinds of data would look like?

Adam Thal: Sure. So I think that’s a very real problem where people just might not be that good at thinking about numbers and sort of forming realistic estimates in their mind. I think two reasons I’m not super concerned about with this particular study is, one, that these are elites, so they’re more knowledgeable about politics than the typical person. They’re more highly educated. And so if anyone’s capable of thinking through this, hopefully it should be these people. And the other thing I would say is that even though I did find that there’s this tendency to overestimate these problems, there’s still quite a few instances where politicians are strikingly accurate on average, where they come within five or 10 points of reality. So they get it wrong sometimes, but they get it right a lot of times too. And so I feel pretty confident in these people’s ability to kind of reason through this.

Matt Grossmann: And you said that there were some differences based on their kind of prior opinions on these issues. Was there any other differentiation among the state legislative candidates, like the ones that were incumbents or the ones that had more prior experience, anything that makes people better at estimating?

Adam Thal: Okay. Yeah. So I think that the most interesting differentiation between different kinds of politicians was that the results of the experiment where I corrected their misperceptions and looked at their policy views were stronger for more experienced office holders. So in the sense that more experienced politicians who potentially actually are already in office, they are more likely to change their policy views in response to being provided with accurate information, which I found interesting and maybe even a little bit encouraging that politicians who are maybe successful and experienced, they take this information a little bit more seriously. The sample size for those politicians is quite small. And so I think it’s hard for me to put a ton of faith in that finding. But if anything, it’s basically that more experienced politicians are more likely to take the information seriously and incorporate it into their policy views.

Matt Grossmann: What would you say the implications are for activists or people who would like policy change in these areas? On the one hand, they spend a lot of time trying to convince politicians that a problem that they’re interested in is large and important and severe. Maybe this is evidence that some of that gets through. On the other hand, you found that it doesn’t really change policy views.

Adam Thal: Sure. So I think that’s a really good question. So I think we might like to imagine, or activists might like to imagine that the solution to the problem is information. If I can just make it clear to you as a policymaker that this problem is really bad, you’ll be motivated to do something about it. The implication of this study is that that will not work or is unlikely to work. Just providing information or making an argument about the scale of the problem isn’t likely to change politicians’ policy views. That could be because they already overestimate the problem to begin with. It could be because that’s just not an important input for them in terms of making policy.

And so if I were to suggest an alternative strategy, it would be instead to think about how to change the political calculus of trying to solve the problem. So the argument you want to make is not look how bad this problem is. It’s look how much you have to gain politically by solving this problem. I don’t have evidence that that would work instead, but it seems like a better direction potentially than just look how bad the problem is.

Matt Grossmann: And how did this make you think about the elite perceptions research area moving forward? On the one hand, you’ve found another category of elite misperceptions that do exist, but on the other you’ve kind of found another reason why they might not really be the source of the kinds of policy outcomes that we get. So how much should we keep studying this and where would we go next?

Adam Thal: Sure. So I absolutely do think we should keep studying this. I think studying elites can be much harder than just studying average voters because it takes a lot more effort, I have found, to survey these people. We can’t just pay some people online to take the survey, like we do typically. And I think there’s just so much to gain from studying these people who are actually in charge of making policy decisions. I think also just because my hypotheses in this particular study didn’t work out, I don’t think that that discounts the value of the study or the value of this approach to research. It’s useful to learn, test a mechanism, and find that it works and helps to explain things. It’s also useful to test a mechanism and find that it doesn’t work and doesn’t really appear to be playing a role. It allows us to have an informed conversation, like we’re having right now, about if it’s not this, what else might it be? What’s the next thing we could look at? And so I really hope that we keep doing this kind of research.

Matt Grossmann: Anything we didn’t get to that you wanted to include or anything you want to tout about what you’re up to next?

Adam Thal: Sure. So I think we got to everything regarding the study. I think what I am continuing to work on is not so much politicians as sort of general problems of living in a political society that’s governed by the affluent, where we just give so much power to affluent people, whether that’s affluent citizens who can make donations or affluent politicians. Something I’m thinking about moving forward is more about how these people who govern our society may be isolated in various ways. So in this paper and in other papers I’ve written, I think about their sort of isolation from these problems that affect other people, how that could lead them to ignore these problems. Something that I’m trying to think about now is sort of how they might potentially feel isolated from a lot of the risks that we see as being a central part of American politics, whether that’s potentially risks to democracy from electing populous candidates, risks from ignoring global warming, for example.

I think that affluent people, maybe including politicians, might really experience a greater sense of safety from the consequences of these problems than everyone else. At a very extreme end of that spectrum, you might think about the billionaire class and how they are building private sanctuaries in islands and hiring militias to kind protect them from the long-term consequences of everything that’s happening in our society. And so I’m continuing to think about how does this class’s ability to protect themselves and isolate themselves from these risks potentially lead them to downplay those risks when they’re making decisions that affect the rest of us.

Matt Grossmann: So part of the trajectory of this overall literature, which you referenced at the beginning, where we find that policy is more consistent with the opinions of the rich than the poor, has been to look at whether there’s some issue with the information environment and the extent to which the richer people are just more active in politics and that information gets through. But there’s always been the background alternative hypothesis that politicians are just people with opinions that are more similar to the elites because they are elites. Seems like this paper might be more consistent with that second view. So I guess think about how the research has evolved since then and where we are in that. Have we not maybe had enough emphasis on just politicians have different views because they’re different than a representative cast of the public rather than because they only hear from that unrepresentative cast?

Adam Thal: So I think that my study is consistent with this idea that it’s maybe more about who they are than about their information environment, who they’re listening to. I think one thing that’s quite hard is that it’s very hard to causally identify the effect of their information environment. And I’m not sure we’ve done a perfect job of that yet. So it could be that we design better studies and we find that the information environment matters more. Like that being said, I think it’s important to take these kinds of [inaudible 01:02:12] findings seriously. And yeah, it brings us to the conclusion that it’s potentially not about the information environment, that it’s just about them being a separate class of people.

Something that brings me back to is the work that Nick Carnes has been doing to think about why are politicians so affluent in the first place? And I think that that work offers a real direction. If we want to think about this as a problem that we want to solve in terms of political inequality, that if you are an activist, it’s not like this question of how can we change the information environment, how can we tell these politicians what they don’t understand? It’s potentially a lot more fruitful to think about how can we think about who becomes a politician in the first place. Because it’s not about changing the information environment, it’s about changing who has power and what their interests are.

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 Grossmann. If you liked this discussion, here are the episodes you should check out next, linked on our website. How to change American’s views of inequality, teaching and TV. Does anyone speak for the poor in Congress? Why rising inequality doesn’t stimulate political action. Why donor opinion distorts American democracy. And When public opinion goes to the ballot box. Thanks to Zander Furnas and Adam Thal for joining me. Please check out The People Think What I Think, and Do Political Elites Have Accurate Perceptions of Social Conditions, and then listen in next time.