Robb Willer studies social and political divides, and maybe more importantly, he tries to find ways to overcome them. In our conversation, he shares his personal background, unpacks persuasion strategies that cut across political lines, and reveals the results of a major new study in his lab that tested a bunch of strategies for reducing political animosity and encouraging people to value democracy over other political attitudes.
Some things that come up in the episode:
- How moral values can be used in persuasion to appeal to audiences across the political spectrum (Feinberg & Willer, 2019; also see this New York Times article)
- The value of correcting “meta-perceptions” as a way to curb political prejudice (Mernyk, Pink, Druckman, & Willer, 2022)
- A “mega-study” testing 25 interventions to address political animosity and democratic attitudes (Voelkel et al., working paper)
Audio for the opening of the show from BBC, CBC, and CNN.
Transcript
Download a PDF version of this episode’s transcript.
Andy Luttrell:
January 6th, 2021 is one of those days that lives vividly in my memory, as I’m sure it does in yours. One of those days when it felt like the ground was just evaporating beneath us. My wife was 8 months pregnant, we were getting ready to bring this kid into the world—you know, a time when there’s so much uncertainty and chaos in your own life that you just need the world to chill out for one second. But one tweet led to another, led to new images, live video footage…of a huge mob descending on the U.S. capitol building, climbing walls, breaking in, shuffling their feet across the foundation that’s kept this country moving forward for what feels like forever.
One of those days like September 11th 2001…or March 23, 2020 when my state issued its stay-at-home order. These are days that, in my lifetime, millions of people collectively and all at the same time realized that they had taken stability for granted.
January 6th was a day when it dawned on me…oh, this is all really fragile. Democracy? That’s something we’re lucky to have, but the universe doesn’t owe it to us. A slow build of years of political tensions, of a public more morally and socially divided, of factions, fictions, and frictions. Are we gonna make it? We’ve got to try.
You’re listening to Opinion Science, the show about our opinions, where they come from, and how they change. I’m Andy Luttrell. And it’s been a minute! Welcome back. The last official Opinion Science episode was four months ago. I hope you stuck around and enjoyed the summer series on science communication, but it’s time to get back to the show’s roots. Conversations with social scientists and professional communicators about having, sharing, and changing opinions.
For our grand return, I’m really excited to share my conversation with Robb Willer. He’s a Professor of Sociology, Psychology, and Organizational Behavior at Stanford University. He’s someone I’ve wanted to have on the show for a long time, and we finally managed to make it happen. As you’ll hear, the research he does covers a lot of ground, but at the center of it is an interest in moral and political divides…and what we can do about them. His lab recently released the results of a monumental effort that aimed to test ways we can restore Americans’ commitment to democratic ideals. And before I go spoiling those findings for you, let’s jump into my conversation with Dr. Robb Willer.
Andy Luttrell:
The first question I had for you is while I was looking at your website it suddenly occurred to me that it’s not RobbWiller.com. It’s RobbWiller.org, which I don’t know that I’ve seen that on academics’ websites. Why a .org?
Robb Willer:
I don’t know. I just… When faced with the choice, I didn’t want to seem like a commercial entity, you know? Since I work for a nonprofit. Although I take a salary, so I don’t know, maybe I’m being hypocritical. Maybe I should just be a .com.
Andy Luttrell:
Oh, I thought for sure RobbWiller.com was taken, so I went to it, and it is fully available, so…
Robb Willer:
Oh. That’s good. Yeah. I’m not surprised.
Andy Luttrell:
So, in terms of your own background, what I most wonder is whether you currently see yourself as a sociologist. On paper, you definitely are a sociologist, but are you? Are you? Is that how you think of yourself?
Robb Willer:
Yeah. It’s a really good question. I think of myself as equal parts a sociologist and a social psychologist, so I regularly attend meetings in sociology, and the SPSB social psychology meetings, and I publish in social psych journals and sociology journals. The way I’ve resolved my multidisciplinary identity is by targeting general science journals more and more, but I’m really lucky to work with grad students from all over, and post docs, as well. Sociology, psychology, organizational behavior, which I’ve also been active in, and then increasingly political science, as well, which is a field that I really like and read probably as much political science these days as just about anything else.
Andy Luttrell:
Was that always the… You had to pick one even though you could have gone any of these directions? Or really was it like, “I want to be a sociologist,” and then eventually you realized oh, there’s these other disciplines out there?
Robb Willer:
Yeah. I think that for me, it is in part because I started off in the subfield of social psychology that is within sociology, which a lot of folks don’t know about in psychology. They don’t realize that there is a sister field called social psychology in sociology but there is. It’s smaller. I’d say it’s something like a tenth the size of psychological social psychology. But it has a rich history. You know, the first books that were titled Social Psychology came out the same year, like 1906 or so, and one was written by a sociologist and one by a psychologist, and they used to share a journal, Sociometry, where some of the early Asch and Sherif stuff came out. That was like co-edited by psychologists and sociologists and was run by the American Sociological Association, and so these two different versions of social psych that were slightly different in orientation, the psychologists were looking down in levels of analysis to cognition. The sociologists were looking up towards society. But they had a lot in common. They overlapped.
And just over time, psychology sort of grabbed a lot more space. Just the social psych in psych proliferated and it didn’t as much in sociology, so I was… I came out of that old tradition in sociology and then was like, “Hey, I want to talk to more folks,” and so I found myself getting more and more active in psychology.
Andy Luttrell:
I remember when I talked to Tom Pettigrew for this, he was talking about how in the ‘50s the Harvard psych department really was also the sociology department, like Gordon Allport was rubbing elbows with psychologists. Just there wasn’t the same kind of distinction at the time, even though now there’s like a very… Feels like there’s a hard line.
Robb Willer:
Yeah. Yeah, I totally agree. Yeah, that was the Harvard social relations department, this effort to merge. I think cultural anthropology was in there, as well, and it was a vision of Talcott Parsons, the sociologist, to try to keep the social sciences knitted together more generally than these disciplines were allowing, and it didn’t work in the long run.
Andy Luttrell:
It didn’t hold, but you’re an example of someone who is sort of straddling all of these islands. Do you think it’s possible? Do you think we ought to do more of that? Or do you think there’s a reason why we have them as separate disciplines?
Robb Willer:
I think we should span across disciplines more, for sure. I do think they’re useful. I think that they’re especially useful for training. I think it’s too much for most early career scholars to take on readings from multiple disciplines. I think that you’re really well advised to read really deeply, frankly within subdisciplines, when you’re starting so that you can get really deep and really know an entire area really well such that you know how to locate your contribution and know that it really is a contribution, know what should come next, so I think that the subdisciplines in disciplines are super useful as you’re training. And then over time, for me anyway, I’ve found that my questions started to fan out from where I started and I just got interested in taking methods, and questions, and background material from different disciplines, and putting them together in new combinations, and doing work that didn’t necessarily fit where I got my PhD from.
Andy Luttrell:
Is there a way that you look at what you’re doing now that you go like, “This is the part that Robb the sociologist contributes to?” Or do you go, “It’s all just blurred together now?”
Robb Willer:
I think that I find my sociology training really helpful in a couple respects. Like on the one hand, I feel like I have good training in macro social dynamics, so I think that I hopefully avoid the pitfalls that many of us social psychologists have of understating the importance of institutions and social structure in constraining these micro level effects in various ways. And then a second thing that I fear I don’t do as well as I should, but I have benefited a lot from the theoretical training in sociology. Sociology has a strong tradition in formal theory building, being explicit about your theoretical claims, trying to make the most general theoretical claims that you can, clarifying the definitions of your terms, being very clear about what kind of relationship you’re proposing exists between variables. There’s a tradition of that in sociology. It’s not hegemonic. It doesn’t dominate sociology. But it’s a tradition that I really benefited from when I was training, and when I’m doing my best work I do that, but I fear that I don’t do it as much as I should.
Andy Luttrell:
So, we’ve talked very generally, like very abstractly at the level of discipline, but what is it you do? Tell people what you do. And I’m curious myself because I look at the things that you’ve produced over your career so far and they all feel like they belong I guess on one CV, but I actually wonder, what do you see as the through line if there is one?
Robb Willer:
Well, so I started off my career really interested in prosocial behavior and cooperation on the one hand, and then hierarchies, especially status and power hierarchies on the other hand, and I was very interested in how those two things come together. Then the field of moral psychology took off in the 2010s and that affected me, as well. I got very wrapped up in that and I had just started doing some political psychology research on the side, as well, and so for most of the 2010s I probably was more of a moral and political psychologist. And now I would say most of my work would be political psychology, political sociology… I mean, some stuff that frankly looks like political science, and so I’m mostly a political researcher all the time.
And the through line for me is just like a deep interest in morality, in right and wrong, and how people think about that, and how that works both at the micro level in the context of everyday interaction, cooperation, prosocial behavior, and then at the macro level. Politics, moral views, moral worldviews. That’s I guess the through line.
Andy Luttrell:
So, I certainly came to know your work through the political persuasion stuff that you’ve done in the 2012, 2015 era, and which makes sense, because that’s often the work that I’m doing these days, is moral rhetoric and that sort of stuff.
Robb Willer:
Sure. Yeah.
Andy Luttrell:
But where did that come from, right? So, it seems like it kind of… I couldn’t find the seed. It sort of appeared. And now we do this thing that I think probably many people would… It’s the work that you’re probably associated with to people, at least to some.
Robb Willer:
Yeah. Well, I think Matt Feinberg and I were… who was my PhD student at UC Berkeley, who’s now a tenured faculty member at Toronto in the Rotman School of Business, he and I were thinking a lot about moral foundations theory and thinking about what parts of it made sense to us, and what parts didn’t, and we were really wrestling with it, and doing a variety of moral psychology projects. It was a time when that area was just opening up and there were so many things to do. You know, there was just so many… There were so many questions that hadn’t been answered yet. And you go back and look at the papers that were published then that are really influential now, and a lot… You can just see there’s a lot of low hanging fruit that people were picking because it hadn’t been done yet, you know?
So, it was this great new field, and one of the things that we saw when we read moral foundations theory is we were like, “Oh, this seems like this could be a tool for political persuasion,” that you could construct more persuasive appeals if you articulated them in terms of the moral values of the people you’re targeting for persuasion as opposed to your own values. And that last part, that people might have different values than the people they’re targeting for persuasion, struck us as particularly meaningful, because it would mean that naturally when people are on autopilot they wouldn’t think to do moral reframing. They would think first to give the arguments that made sense to them, that convinced them. That’s intuitive. That might be part of a larger political and moral project to move people on not just a political issue, but also your deeply held moral values, so it makes sense to approach political persuasion that way.
But also, it could have limitations in terms of the pragmatic result. You might be less persuasive just reiterating your own reasons for your political positions.
Andy Luttrell:
It occurs to me that there’s a question about this that I’ve always had, which is… So, again, just to kind of reframe all this, the notion is that liberals and conservatives prioritize different values. Naturally, they might speak past each other because they’re talking in the way that makes sense to them, but that’s not the way that it makes sense to the other side. But if you could change your strategy so that you’re targeting what the other side actually values, then you could have a more successful communication outcome.
My question is implementation wise, is there a concern that people can authentically speak to the other side’s values, right? I even think this as a research study designer, where I go, “I have to somehow write two versions of the same argument.” One of them naturally falls from my own values. The other, I’m kind of just like pretending I know what these values would conclude about my position. And I worry that that is detectable, right? That there’s a sense of like, “Oh, this technically is couched in the values I hold but it doesn’t work. And it doesn’t feel authentic.” So, I’m curious if you’ve given any thought to that or run into those, in terms of like if we actually want people to do this, are there drawbacks of actually making it work?
Robb Willer:
Yeah. A lot of great ideas there. I think it is possible for people to learn to do this, but I think it helps for there to be some instruction and also for them to be motivated to do it. I don’t think it comes naturally, like our research shows that people don’t do it spontaneously. We do find that if we give people… if we show people a morally reframed persuasive argument and one that is not, that they probably agree with more, and we ask them which of these would be more persuasive, people tend to accurately identify that the reframed one on average is more persuasive. And when we ask them which one do you want to send to be persuasive to this other person, then people will usually, but not always, send the reframed one that they have detected as being more persuasive.
So, there’s kind of like a base level of willingness, but also people aren’t spontaneously doing this, so I’ve found in workshops that you can train people to do this, but it definitely helps to have a plan and really explain the logic of it. And there’s a really nice paper by David Broockman, Josh Kalla, and Adam Seth Levine, where they applied moral reframing in a canvassing context and they had a pretty short tutorial where they taught canvassers how to do moral reframing, and pretty… They were able to do it quite well, eliciting people’s moral values in interactions at the beginning of a canvassing interaction and then fitting an argument to those elicited moral values. For abortion, of all things, which you would think would be really hard to achieve some kind of persuasive effect on, but they found that they were able to get effects and even durable effects on people’s abortion attitudes with canvassers who were trained for a pretty short period of time.
So, that would seem to be… You know, but canvassers are smart, right? They’re above average at interacting with people about politics. Because they’re motivated, they’ve thought about the issue, and then they’ve had this training, and I think it is an open question whether the average person can do this authentically, and how important the authenticity part is. The way I like to talk to people about it is that it’s more honest and I think defensible to use moral reframing where you say you present it as a reason that a person might come to agree with you while having the values they do, rather than presenting it as your own reason. Then it can be authentic and not dishonest, right?
If you say, “I support same-sex marriage because gay Americans are proud, patriotic Americans who serve in the military,” but that doesn’t actually have anything to do with why you support same-sex marriage, that seems to me like you shouldn’t do it that way. Why not say, “Well, I hear that you care a lot about American traditions, and patriotism, and the military, and you know, gay people are proud, patriotic Americans who serve in the military, and it seems to me you might support their rights just even coming from were you’re at with your values.” That’s a way to do it that’s authentic and not deceitful.
Andy Luttrell:
The drawback is you’re sort of announcing your strategy. You go, “I’m now…” It’s like I remember in school, when the math teacher would try to make the problem into a word problem and they’d go, “Well, what are you interested in? Well, imagine five cars drive…” And you go, “Listen, I see right through this. I know that’s not how you think of it. You think that’s gonna work for me.” So, I could imagine that also potentially spurring some reactions where you go like, “Ugh, I see what you’re doing here. You’re trying to talk down to me because I have different values as you,” or something like that.
And an alternative could just be like a, “People are saying,” as opposed to, “I believe this for these reasons,” right? These are just arguments. And these arguments might resonate with you, and I don’t have to say they’re for you, but hey, take them for what they are.
Robb Willer:
I’m not being specific about whether they’re my reasons or not. Yeah. I think that’s also I think a legitimate way you could do it. One thing that is sort of an unstated assumption in a lot of our work is that we try to select arguments which we try to test arguments that would not be offensive to and off putting to the group that already holds the political position we’re trying to be persuasive about. So, for example, we try to make a same-sex marriage argument that is not offensive to liberals that wouldn’t reduce their support for same-sex marriage. That’s what we’re trying to do. And we could probably say that more clearly in our papers, because sometimes it confuses people, like, “Wait. Why don’t people get turned off from this position hearing it in value terms that are not their own?”
But one result of that is that I think in general you look at the persuasive, the morally reframed appeals that we’ve tested, and you wouldn’t think, “Oh, I’m offended by that. I’m unwilling to say that because I disagree with that.” That hasn’t… At least for me, I’m a liberal, I’m a very far left progressive, and the morally reframed appeals that I’ve written haven’t struck me that way where I say, “I don’t agree with this.” They’re mostly just like another good reason for the political position that I hold.
Andy Luttrell:
I’m sure this is a question that’s come up a bunch about this, but are there some issues that don’t lend themselves to this approach? Because it does seem like there are times where you’d go, “I will never find the other side’s way of making this sound right that actually is compelling.”
Robb Willer:
Yeah. Well, I think that the most obvious one would just be political positions that somebody… that a person you’re targeting for persuasion already feels really strongly about, that’s maybe really central to their political ideology to where it’s nested in a set of related cognitions that are also gonna need to move if you’re gonna move this one. And then also there are certain political positions that just can’t pair with other kinds of moral values to make a coherent, persuasive, compelling political appeal. So, what we’ll do is we’ll say, “Okay. Who are we targeting for persuasion?” Okay, let’s say it’s conservatives. Okay, well now there’s three moral foundations, or four if you want to throw in liberty, that are in play for this political position. Can we make an argument that connects national health insurance to respect for authority? Okay, not really. Okay, going down the list. What about patriotism? Oh, okay. Yeah, sure. We could do that. And so, then we’ll make that appeal, so it is often the case that some of the combinations just don’t work.
Andy Luttrell:
Does that constrain, like you talk about workshops. Presumably there are folks who go like, “All right, let’s do this.” Is there anything else like that that when you meet folks like that you go, “Well, hold your horses. If these other parameters are there this may not be the solution.”
Robb Willer:
People are actually in my experience really good at it. You know, they take it up with enthusiasm. They think it’s exciting, the idea that they could have a more positive political conversation, that they might actually convince somebody or at least have a civil exchange. So, yeah, in my experience people are surprisingly good at it, I think because they’re motivated, but I’m talking to a really self-selected group, right? People that for whatever reason ask me to talk to them about this.
Andy Luttrell:
To sort of push things, forward a little bit, so this has all been like there are partisan divisions and there’s an interest in trying to bridge them, account for them, somehow get people to talk to each other in any way that is gonna make any headway, which sort of previews the next kind of phase of things that you’ve done, which also brings us to some of the newest stuff that you have to share, which is other forms of political division or other political concerns. So, in this new manuscript, at some point you mention this notion of America’s democracy crisis, so what is that? What is the big problem that you and your lab are trying to address?
Robb Willer:
So, there are a few different problems related to American democracy and polarization that are concerning to me that motivate some of our research. So, one of the trends that concerns me is the steadily rising levels of affective polarization or partisan animosity between Democrats and Republicans in the U.S. over the last four to five decades. Having grown up in red states in Kansas and South Carolina, it was palpable to me that there were real political animosities that made it uncomfortable for people to live in certain places, to be in families, to work in certain workplaces, be they Democrat or Republican. These political tensions are felt on a day-to-day basis by millions and millions of people and they’re really unpleasant.
In our research, we also find high levels of dehumanization of out partisans, low but still concerning levels of support for partisan violence, so political animosities in the U.S. are just, independent of what role they may play in paralyzing our federal and state governments from addressing social problems, independent of their downstream effects on democratic stability, I still think I would say that polarization itself, toxic polarization in the U.S., is creating just direct harms and problems for people that I take seriously.
But then I do think that it contributes, as well, to democratic stability problems that are very concerning. We see democratic backsliding in many states in the United States. We see concerning levels of Americans reporting that they would still vote for their party’s candidate even if they did some significant transgression of democratic laws or principles. Over 80% of Americans report they would do that. Or I should say American partisans. So, the polarization problems can give rise to democratic stability problems, and that’s something I take really seriously, and I’m really concerned about. Over the last six years, especially.
Andy Luttrell:
And so, some of the examples of the latter point of the democratic stability are things like refusal to accept the results of elections, right? To me, that stands out as the most successful one. Are there other, like if you sort of fill in the gaps of that, is that ultimately what you’re talking about? That specifically? Or is it really like a constellation of things that are all kind of contributing to this overall trend?
Robb Willer:
Well, when we study anti-democratic attitudes, we create a composite measure that includes how much would you support or how likely would you be to vote for a candidate from your party who did X, Y, and Z, and then X would be not acknowledging the results of an election they lost, Y is using government to repress or prosecute journalists who write articles you don’t like, Z could be removing polling stations from areas that benefit the rival party, so there’s a number of ways in which politicians can abuse their power to behave in undemocratic, autocratic ways that we would be concerned about. And we find these different items tend to be highly correlated. So, they are kind of of a piece.
People, in their heads, are weighing the value of partisan gain against what you get from these undemocratic, against democratic principles, and then their reported answer seems to be a combined function of those two things. And I think this is the way in which scholars like Levitsky and Ziplatt show that polarization can give rise to democratic backsliding because the critical check on democratic backsliding that the voting public provides can erode if they look at their choices and they say, “Oh, well, my in party candidate isn’t acknowledging the results of the last election and I hopefully know that’s bad, but I can’t vote for Joe Biden because he’s the devil, and so it’s not really an option. I’m gonna vote Republican or Democrat no matter what my in party candidate does.” And that’s when you can get in a lot of trouble because politicians can do a lot of stuff in those kinds of situations.
Andy Luttrell:
Yeah. That sort of speaks to a question I had, which is how explicit are these antidemocratic views, right? In that you could read this and go, “Well, if I want to measure antidemocratic attitudes I’ll ask people hey, democracy thing. Good thing, bad thing?” But that’s not what you mean, right? It’s sort of like these indirect things. People are trading off that and prioritizing something else, right? So, in some ways you could say it may not just be that people are actively saying, “Let’s reject democracy and move towards these other things.” Maybe they are. But instead, it’s like more important issues have arisen, right? We ought to prioritize making sure that we outlaw abortion. We have to prioritize these other things and getting these other jokers out of office. And the consequence is that that chips away at this system that we have that’s supposed to do all these wonderful things.
So, I guess my question is how explicitly do you think these antidemocratic attitudes are changing?
Robb Willer:
Yeah. Well, I think the reason we study it the way we do, and here we’re really following these political scientists, Graham and Svolik, who had a terrific APSR paper a couple years ago where they studied Americans’ democratic attitudes using a very similar paradigm of would you tolerate this antidemocratic act if it came from an in party candidate, and we were compelled by that paradigm because if you just ask people how do you feel about democracy, you know, like are you pro-democracy? Everybody’s pro-democracy. Virtually everyone thinks that the U.S. should be a democracy. If you ask about principles in the abstract you don’t get very much information, I don’t think, because the way in which these decisions are put to people is more in the form that Graham and Svolik study it in, of like, “Well, okay, you like democracy, but your favorite political candidate did a bad, bad thing. Are you gonna do anything about it? Will you at least stay home?”
And mostly people don’t, but maybe they could, and what could persuade them to is something that we’re really interested in. One thing I do want to highlight on the side here is that maybe the biggest threat to American democracy in 2022 is more of a misinformation problem, which is that something like 60% of Republicans now believe that the 2020 election was not legitimately won by Joe Biden. That is not a situation… I wouldn’t understand that situation as a social psychologist of Republicans trading off democracy for partisan gain. They are in an information environment where they believe, because respected, trusted political and media leaders have told them, that the election was stolen. And so, for those folks there’s no difference between partisan gain and defending democracy. Both involve voting for Donald Trump and whoever he tells you to, and whoever says that the election was stolen, and that’s part of why misinformation is so dangerous, is because then democratic principles are sort of irrelevant, or worse yet are harnessed in favor of autocratic moves.
Now, there’s 40% or so of Republicans who believe or understand correctly that the election was not stolen, and most of them still vote Republican, and for them the stuff that we do that we have emphasized more recently in our research, in for example the strengthening democracy challenge, this big experiment that we wrapped up recently, those people are from great interest there because they do know that Trump and Trumpist candidates are saying something they know is wrong, and they’re facing the choice. Do I vote for these people in the 2022 midterms or not? And that I think is in some ways an easier to treat problem than the misinformation one, which I think is really hard but also worth… We’re also studying that, too, as a community of scholars.
Andy Luttrell:
Yeah. That was my reaction when I saw the items to measure these sorts of things, and it was kind of like if the election results are rejected, would you still support this person? Those kinds of things. It was kind of like, “Yeah, that could mean two different things,” right? Someone saying, “Rah-rah, down with democracy, that’s what I want,” versus, “No, because as far as I’m concerned this is an appropriate way to preserve democracy.” It just sort of raised a question for me like when people agree to this, that like I would still vote for someone who said this, it could actually mean two opposite things about their personal value on democracy. And I don’t know, is there a way to… If that’s all that we’re asking, is there a way to disentangle that?
Robb Willer:
To disentangle the… What are the two things again?
Andy Luttrell:
When someone says I would continue to support someone who says the election ought to not be listened to, that could either mean I believe in democracy, or it could mean I don’t think democracy is important, right?
Robb Willer:
Right, right. Yeah.
Andy Luttrell:
It could mean both of those things. And so, is there any way to read people’s response to that question and figure out where do you actually stand on the value of democracy?
Robb Willer:
Yeah. It’s really interesting. I mean, it’s really interesting. Yeah, so we don’t have a measure. We don’t have a way to really fully disentangle that, because I agree those two things are conflated when you ask the question this way. It’s pragmatically attractive because it’s hard to ask about democracy in the abstract, but then it is social psychologic, it’s experimentally unattractive because it confounds two motivations in your measure.
I do think that we have prioritized a measure that is support for undemocratic practices which doesn’t say, “Will you vote for this person.” It just says, “What’s your judgment of this practice?” It still is coming from an in party politician, but it’s not quite that choice of like, “Oh, and we’re not just asking you what you think of this, but how would you vote on them?” So, the reason we privileged that item as a more central measure like in our preregistration and so on was because it had less of this confound in it. But it’s not devoid of it, because we still are asking, because if we asked about an out party politician doing an undemocratic thing, you’d be like, “Oh, yeah. No, that’s terrible. Everything they do is terrible.”
Andy Luttrell:
So, having an approach to figuring out where people stand on democracy, partisanship, all those sorts of things, tell me about how we have saved democracy, right? You guys did it?
Robb Willer:
Sadly, no. Yeah, we’ve not.
Andy Luttrell:
Yeah, so this big study, so it’s a mega study, which we now have a handful of those in the field. What is it? What is it that makes this a unique approach to tackling this problem?
Robb Willer:
Sure. Well, just as background, in a nutshell what we did was we were interested in ways to successfully treat or effectively treat partisan animosity, antidemocratic attitudes, and support for partisan violence in the American mass public, specifically amongst partisans who in this paradigm, for thinking about this, are the biggest concern, and do show the highest levels of all these things. And so, what we did was usually what we do in our lab, and like in most labs, is we think, “Oh, do we have a good idea for this?” We find that one idea. We go out and test it, try to make a case for it, or abandon it if it’s not good, or if the results are unreliable, and rinse and repeat for years and years and years.
But for this, we were like, “Well, what if we try to take a really different approach and try to find out what are all the best ideas that we can find for treating these outcomes, these problematic outcomes?” And so, what we did was we put out a call on social media and said, “Come one, come all, submit your interventions that can reduce these outcomes amongst American partisans.” And the big constraint is it has to be something we can put in an online survey experiment, which is a pretty big constraint, so it can’t be face-to-face interaction. It can’t be something long term. It has to be something that we specified could be experienced somehow by experimental participants in under eight minutes in a survey experiment.
So, that still left a lot of room for diverse submissions. We got videos, audio, chat bots. There was one guided meditation. There were others that had people write essays on certain topics, and so on, and so on. So, a whole bunch of different approaches to designing interventions that reflected also a great diversity of strategies, from invoking common identities, to correcting misperceptions about rival partisans, to developing cross-partisan social class consciousness. People had a lot of ideas. We got actually too many ideas. We got way more submissions than we had planned for. We got 252 submissions from 400 submitters in 17 countries. We tried really hard to make this project accessible to non-academics and we held workshops, and paired people with academics who were interested in working with non-academics, trying to get those ideas from outside of academia into the study, and then we selected the 25 ideas we found most promising from among the 250 and tested them in a very large experiment with 32,000 participants.
And sorry, it’s a long buildup to-
Andy Luttrell:
Can I ask before we go on?
Robb Willer:
Absolutely.
Andy Luttrell:
How did you decide what those 25 winners were? Like what were you looking… That was a judgment call, right, essentially?
Robb Willer:
Yeah. And a much tougher one than we had bargained for. We expected something like 40 submissions. That was my… We all put in bets on how many we’d get, and I thought we’d get 40. And we got over 250 and I would say well over 100 that were viable, where you’re like, “Yeah, that could work.” And so, it got really hard. One thing is we had a great advisory board which was made up of academics and practitioners. There’s this kind of robust nonprofit and activist space around bridging political divides and defending democracy, and a lot of those folks know a lot of stuff we don’t, and so we wanted to get their perspectives in the process. They are on the ground talking to people about these problematic views all day every day, and so we tried to get them on the advisory board in large numbers. Something like 40% practitioner, 60% academic on the advisory board.
And then we also made our own independent evaluations. And we do a lot of research in this kind of paradigm of like survey experiments with outcomes like these, and so I think we had a decent nose for what were the most promising interventions. Of course, we would think that, but I think we might have been right. And then I think another thing that was really difficult about selecting is how much to weight the diversity of ideas that you select relative to their expected efficacy. So, for example, we found that for misperception correction, like corrections of misperceptions of rival partisans’ views, four of these misperception corrections seemed worthy of being in the top 25. I think that that decision was validated since two of them probably were in the top three or four interventions overall, so I think we were right about that, but a critic could say, “You’re not valuing diversity enough. You just need one or two of those and that’s good enough.”
But you know, I think the results suggest that we were right to let four in because two of them performed really, really well, and the other two performed only pretty well, and I think it was if we hadn’t done that, then we would have potentially overlooked our very best intervention. So, that was just a very difficult thing.
For example, the guided meditation got in. In part because it had a track record of having shown good effects on partisan animosity in a previously published paper, but also because it was very different. We thought it would be illuminating if it turned out to be in the top five or something. But sadly, my favorite intervention that did not make the cut was called the pale blue dot intervention, and it had this photograph that was taken from I believe the Voyager satellite of the Earth from many, many, many miles away, and had this audio track of Carl Sagan talking about the smallness, just the triviality of human conflicts that are just sitting on this blue dot in the middle of the galaxy, and I thought it was terrific, and I thought it’s effects might have been really impressive, and it basically… It finished 26th, you know? It was just barely out of it and there were extended debates about whether it should get in.
Andy Luttrell:
I’m sure you could have exercised your power and made it cross the finish line.
Robb Willer:
Yes. Well, I tried to have a very egalitarian relationship with the selection committee.
Andy Luttrell:
Democratic.
Robb Willer:
Exactly. Yeah. Yeah. Don’t want to be autocratic about the strengthening democracy challenge.
Andy Luttrell:
So, before we get to who won the race, so pale blue dot is one that you would have put your bet on. I’m curious to know which kinds of interventions you felt like pretty confident, like, “These are the ones,” and the ones you went like, “Come on. We’re gonna have a guided meditation and that’s really gonna do anything?” What were your own biases before the results came in?
Robb Willer:
Well, one of my biases was that I was very confident in the meta perception correction interventions, and I think that confidence was mostly borne out. So, meta perceptions are perceptions of what other people… how other people perceive you or your group. And sometimes there’s a more general use of meta perceptions that’s just perceptions of other people’s perceptions.
And we had done research on meta perception corrections around support for partisan violence where we had found that Democrats and Republicans overestimate how much the other political group tends to support partisan violence by 300, 400%, and when you correct those misperceptions people then say, “Oh, well, I’m gonna ratchet down my own support for partisan violence.” And the effect there was something like 40% and we found it persisted weeks later in a follow-up survey. And so, I guess in our lab we had had this experience of even these small interventions where you’re just correcting a few, just giving people a single page in a survey with a few numbers on it, and you’re getting an effect that lasts for weeks. That led us to have a lot of faith that that paradigm could work.
And you know, it had been shown in a lot of different settings. It had been kind of co-discovered in political science and psychology as an effective way to intervene. So, I had a lot of faith in that. I also had faith in common identity interventions because I’d seen research from Matt Levendusky showing that they could… You know, creating a cross-partisan American common identity in an effective way could move the needle, and of course the rich social identity theory tradition in social psychology made me confident in that, as well.
I think stuff that I was less confident in included… I mean, there was a lot of stuff that involved just playing inspired speeches from history for people, and I felt like those speeches would not necessarily become relevant to contemporary partisan divides. So, people might listen to a speech and say, “Oh, wow. That is an inspiring speech by JFK. It isn’t really relevant to why I hate my Republican uncle or why the Republicans are destroying the country.” So, I was suspicious of those.
There was also like lower quality submissions just in terms of how engaging they were and so on. I was more skeptical that those would hold people’s attention and be really engaging. And then there was stuff that I think I was wrong about. Probably underestimated the power of elite cues even though in our lab we do research on elite cues, but cues from trusted political leaders in your group, and even bipartisan endorsements of things can do a lot more than I initially expected.
Andy Luttrell:
So, what do you find? With all of these data that you could probably talk forever about what you found, and it probably is the longest supplement I’ve ever seen.
Robb Willer:
Yeah. We got a little crazy on the supplemental material.
Andy Luttrell:
There’s a lot of extra tables. So, if you are to give sort of the overview of like to you, what ultimately do you take away from this enormous effort?
Robb Willer:
Wow. Yeah, there’s so much, and that’s been really the trouble for us is that it is very hard to distill the insights down to just a small number of insights, so I’m gonna try to be as parsimonious as I can here. So, first of all, for democratic attitudes, I would say like antidemocratic attitudes, like improving these, including support for partisan violence, I guess I would highlight three strategies. I don’t know how parsimonious this is. Correcting meta perceptions turns out to be really effective. Second, elite cues turns out to be really effective, so presenting people with leaders from their party that are saying, “Do this, do that, like you should not be violent towards rival partisans. You should defend democratic principles.” Et cetera, et cetera. And then a third would be common identities, you know, like building common identities that connect people across party lines.
Now, having said that, I kind of want to add a fourth, which is one of the most effective interventions that we tested was something we called democratic fear, which was a short video that was created by some political scientists here at Stanford, Katie Clayton and Mike Toms, and it showed scenes of democratic collapse in settings like Turkey, and Venezuela, civic unrest, police repression in the streets. Really underscored how bad things can get if democracy really, really falls apart. And then it culminated with scenes of the January 6th Capitol riot or whatever we call it, and really made it tangible that it’s like, “Well, no. This could actually happen here. Don’t take for granted the democratic stability we’ve enjoyed for a century and a half. Things can fall apart and when they do it’s really bad.”
And so, that was also a powerful intervention that had a lot of positive effects, and in fact, if you were just specifically looking at the support for undemocratic candidates variable, which may be the most important one because that’s the one where the public plays a role, where public opinion really matters for democratic backsliding. It’s like, “Well, are you gonna elect the people that are backsliding the most?” The two interventions that made the biggest difference were correcting misperceptions of the democratic views of rival partisans and then this video that underscored the risks associated with democratic collapse. So, it’s basically like the other side is not dying to destroy democracy, so you don’t necessarily need to yourself, and then also if both sides defect here and you wind up with democratic collapse it could get really, really bad.
Andy Luttrell:
So, those are impacts on antidemocratic attitudes, and in some ways I know that those effects weren’t… I mean, not enormous, right? And you’d almost think that democratic fear video should have been like, “Oh, unilateral shift in these views.” And even still, it was more of a nudge than anything. But it seemed like the really potent stuff was on the other outcomes, so what did those look like?
Robb Willer:
Yeah. You’re exactly right that for the antidemocratic attitude, support for partisan violence, effect sizes were smaller than they were for partisan animosity, where we got the biggest effects, and also the partisan animosity effects were big enough to be durable. I think the best predictor of durability in a two-week follow up was just how big was the effect in the first place, so the bigger effects endured. The smaller effects didn’t. For partisan animosity, the two strategies that stuck out… I mean, for one, lots of interventions were effective. Because we were able to select the 25 most promising from 250 submissions, we were able to wind up with mostly winners. 23 out of the 25 reduced partisan animosity significantly. We had a really big sample, so we were able to detect even small effects, but it was noteworthy that everything was significant almost, you know?
We also added an active control where it wasn’t just nothing, like people saw some information about structure of government and so on, and I think we had 17 interventions that outperformed that, so it was even using the active control, just a lot of winners. But the two strategies that were reflected in some of the best or the most effective interventions for that outcome would be again, common identity, so building some sort of overarching American identity, and then secondarily presenting people with relatable, sympathetic exemplars of the rival party. There were multiple videos that did that, that just showed you like, “Oh, here’s somebody from the other side and they’re likable, and they have a perspective that’s understandable, and they’re being respectful towards people on your side.”
Andy Luttrell:
I was wondering in looking at the actual interventions that you tested, I wondered if there was a way to identify core themes. Because it kind of seemed like there were a few things that yes, they were 25 different interventions, but they kind of relied on a smaller handful of strategies. I’m curious if you’ve done anything or thought at all about like is this ultimately evidence for if you can accomplish these three things through whatever specific kind of intervention you want to do, that we could think of this more in terms of like core principles as opposed to, “Hey, do the biggest one.”
Robb Willer:
Yeah. Oh, I definitely think that it is wise to take from our results certain general strategies that are effective that could be implemented in somewhat different ways than the ways they were implemented in our study. You know, for example, one of the top performing interventions was this elite cues intervention that showed a short video, like a one minute long video, an ad spot that was made by the two gubernatorial candidates in Utah in the 2020 election, or during the 2020 campaign, and they got together and filmed this spot where they said, “Hey, we’re competing against each other but we also both agree to recognize the results of the 2020 election, recognize whoever wins it, we’re gonna recognize the presidential outcome.” They just got together and just sort of reaffirmed the basic democratic principles of elections.
It’s actually a quite sweet and touching video in a way, because you realize you don’t see those very much, and it’s a reminder that when people are running against each other in a campaign, agreeing to honor the results, that they’re competing but also at some level they’re cooperating, as well, on some basic rules of the game. And so, they brought that to the surface.
But the best way to implement that would not be… You know, it’s interesting that that had effects, even though most of the participants in the study were not residents of the state of Utah, didn’t know who these dudes were, and it was an election that was already over, and it was just a 60-second spot. This is not the ideal implementation of this strategy. You would want to do something like get Barack Obama and George W. Bush to sit down and do a 90-second or 120-second spot that’s about the 2022 midterms and how they will be legitimate, and explains the process in more detail, and then gets promoted by Meta, and Twitter, and TikTok, or whatever in various forms to get as many exposures as possible. Maybe there are slightly different versions so that people remain engaged when they see it for the third or fourth time in their feed. The ideal way to implement that would be more than a little bit different than that implementation.
So, I think that if you were trying to take ideas from this, it would probably be take the elite cues, the misperception corrections, and the fear of democratic collapse principles, at least for attacking antidemocratic attitudes, and then try to make the interventions have a stronger dose than what we tested, and then repeatedly expose people to that content or content that moves those levers, and that that would be the most likely to make a bigger effect, to have a bigger effect that would endure.
Andy Luttrell:
I even wonder if some of these individual interventions cluster together as ultimately doing the same thing, right? So, like the video you describe kind of serves to correct a misperception I have about the other side. You go, “Ah, the other side doesn’t universally believe in undermining democracy and violence.” And some of the other interventions do that. I even noticed like you mentioned common identity. A handful of them, there’s like one called common identity, but several of them have elements that are about, “Hey, ultimately we have more similarities than we do differences.” And so, I just wonder like are these actual 25 unique interventions or are they 25 instantiations of three strategies that are viable?
Robb Willer:
Yeah. I think the latter. I think your characterization is spot on. I think it would probably be maybe more like seven or so strategies that we would detect as probably separable, most of which I’ve listed here. I’m trying to think if there’s any I haven’t. Yeah, it’s like elite cues, sympathetic exemplars from the outgroup, misperception corrections. Oh, another one would be invoking descriptive norms, so there was a lot of… which is similar to misperception corrections, but I think different, because instead it’s saying, “You’re in this group. Most people in this group think this, so you should too.” And tried to leverage cases where there is consensus on certain issues or greater overlapping public opinion than you would expect.
Yeah. That would probably be the only one I haven’t mentioned, is like leveraging general public opinion more to either reduce the salience of partisan differences or induce some kind of conformity response.
Andy Luttrell:
Before we wrap up this study, I want to highlight one thing that I think is a really… maybe it’s not the headline finding, but it’s maybe to me the most intriguing, which is that sometimes in these mega studies it just is like, “We ran a race and this horse won.” And everyone goes, “Oh, hooray for that horse.” But you actually have a really interesting finding, which is that by trying to affect two different things, you get to comment on whether those two things are the same or not, right? So, there’s this feeling that if only we could get Democrats and Republicans to like each other more, we’d fix democracy. We would finally convince people to uphold these principles of government and democracy. But you kind of actually are able to say something about the viability of that in the long run, so I think you know what I’m hinting at.
Robb Willer:
I do.
Andy Luttrell:
So, what am I talking about?
Robb Willer:
Yeah. Yeah. But I think you put it quite well. So, there’ve been a couple papers that have come out recently. David Broockman, and Sam Westwood, and Josh Kalla have a nice paper provocatively suggesting partisan animosity, affective polarization, whatever you want to call it, dislike between partisans isn’t particularly related when you start experimentally manipulating it to democratic attitudes measured any number of ways. It’s a very compelling paper. And then we also have a paper that basically came out of the pilot tests we did for the strengthening democracy challenge making a really similar point, that interventions that robustly reduce partisan animosity did not tend to reduce democratic attitudes as we measured it a variety of different ways.
But you know, we realized that the challenge dataset really provided probably the best dataset yet, best single dataset yet for analyzing this problem, because it is treating partisan animosity, or at least attempting to, 25 different ways, and so you can get a pretty robust sense of whether experimentally manipulating partisan animosity, does that do anything to these other variables? But partisan animosity, we kind of confirm these recent papers. It’s not particularly related to support for undemocratic practices, support for partisan violence. However, intriguingly we found that it was… We found good evidence for the first time that partisan animosity is related to a democratic attitude, and it was specifically support for undemocratic candidates.
And so, there’s something different about supporting undemocratic candidates as opposed to just supporting undemocratic practices that starts to pull your partisan animosity into the equation. And what’s more, what we found was that the treatments that reduced people’s willingness to support undemocratic candidates, they tended to follow one of two causal paths. They either were treatments that were effective in reducing your tolerance of undemocratic practices, or they were treatments that reduced your dislike of rival partisans. And either one of those was a credible way you could wind up being less likely to support undemocratic candidates. And the way I like to think about this is you’re thinking about whether or not to vote for this candidate who’s done something undemocratic that’s from your party. It’s kind of a joint function in your mind of how much you care about the norm or principle that they’ve violated or broken here and how much you hate the other people that you’re gonna be benefitting if you don’t vote for this person, you know?
And this is part of the way people would explain, Republicans would explain voting for Trump in 2016. Many Republicans would say when interviewed, “I didn’t like the tweets, and I didn’t like a lot of the stuff he said, and he’s a bully in the debates, but I can’t vote for Hillary Clinton because she’s the devil and I’ve heard all this stuff about her.” And so, this would be a lot like that, of like you need to both care about the principles that are being violated and not just hate the other side so much that you’re completely unwilling to stay home or vote for them in order to defect away from an in party candidate that does something undemocratic.
Andy Luttrell:
And it kind of goes back to what we were saying however many minutes ago, which is how much of this is really the erosion of people’s value on democracy versus this tradeoff that feels like there’s no way around it, right? And so, in some ways it’s sort of saying that part of it, at least we could say, is the tradeoff part, right? And if I could make that less important to you, what is ultimately not changing is this adherence to the principles. But now I don’t have this other thing that’s pulling me away, like I can’t hold… We just can’t hand the keys over to them. But now you go, “Well, would it be so bad if for a couple years we do hand over the keys if we get to hang onto these principles that we think are important?” Sort of seems like that’s a way that you could wrap that up.
Robb Willer:
Yeah. And you know, and I should say that I think the normative implications for this at the individual and societal level are super complicated and difficult to take positions on with any certainty. I mean, I can say that I think I would like to be in a society where people do not tolerate undemocratic moves from politicians, right? And that the critical behavior is will in partisans tolerate that or not? Or will they defect? I mean, I guess you could say also swing voters have a heck of a lot to say about it, as well, but we don’t have a lot of those, and so it really… If you’re gonna try to stop the democratic backslides in politicians via the ballot box… There’s other ways to do it, like the courts. It’s really gonna help if some in partisans at some point are like, “No, this is too much. We’re not gonna stick with our person.”
At the same time, if I filled out that battery, I don’t know that I would look like the most democratic person in the world. I’m pretty sure I wouldn’t. Because I have really strong political views, and I weight them heavily, and I think that the rival party will do a lot of harm to the country and will not help me see my views turn into policy and take action, and so the moral calculus for me is really complicated, and I’m sympathetic to the people who score low on these democratic measures. I get it. While at the same time, again, wanting to be in a society where we won’t let this happen.
Andy Luttrell:
So, have we saved democracy? Did you guys do it?
Robb Willer:
Definitely not yet. Definitely not.
Andy Luttrell:
Assuming not. How hopeful does your program of research make you? Are you the kind of person who looks back at what you do and goes, “Okay, there’s hope here.” Or do you go, “We’re basically just documenting our demise?”
Robb Willer:
Yeah. I mean, this is a tough question because I’m not the most optimistic person by nature, but I care a lot, so I keep trying to… I just try to do good work and then hope that it makes a contribution and try to do that last step of driving it towards application, putting it in the hands of people who make decisions that are really impactful. And I think… You know, I think we have advanced the science of the causes of tolerance of democratic backsliding in the mass public. I think we advanced the science there. I think we have sort of a tool kit of these are the levers that you have to pull if you want to reduce people’s tolerance for democratic backsliding, or support for partisan violence. These are the causes that the research, as of 2022, says are most efficacious and most impactful. The effect sizes are small. I think there’s a lot to be said about how to apply the knowledge, how best to apply it, but I think it’s good.
I think it’s good for us to know these three causes are bigger and more impactful than these five over here. That I think is helpful. And if the right people can find a way to use it, then it could make a small difference and elections get decided by razor thin margins in the U.S. in the 21st century, so small differences can actually aggregate to make big differences in the end. That’s what I hope.
Andy Luttrell:
Well, great. I will keep my fingers tightly crossed for the future. And in the meantime, thanks so much for coming on to talk about this stuff.
Robb Willer:
Yeah. Thanks a lot for having me, Andy. It’s such a great podcast. I’m such a fan. Yeah. It’s an honor to be on. Truly.
Andy Luttrell:
Oh, great. Yeah. Thanks.
Andy Luttrell:
Alright, that’ll do it for another episode of Opinion Science. Thank you so much to Robb Willer for sharing his work with us. To learn more, you can go to the webpage for this episode for a link to Robb’s website…or maybe you remember it—robbwiller.org. That’s Rob with the two Bs. Or not 2 Bs? I’m so sorry. For details about the mega-study testing strategies to reduce partisan animosity and promote commitment to democracy, you’ll also find a link to their report in the show notes.
If you want to know more about this show, go to OpinionSciencePodcast.com. And if you’re new to the podcast, check out the Episodes page, which has a link to some of my favorite episodes to get you started. Follow the show @OpinionSciPod on Twitter and Facebook, and subscribe wherever you like to get your podcasts.
Okay, I’m super excited to be back. I’ve already got a bunch of great stuff waiting to get released, so I’ll see you in a couple weeks for more Opinion Science. Bye bye…
