Home

Donate

Challenge Yields Experimental Interventions to Strengthen U.S. Democracy

Justin Hendrix / Aug 24, 2022

Audio of this conversation is available via your favorite podcast service.

It is well understood that the United States is in a divisive period. And yet, not all polarization is undesirable, and in some cases referring to the animosity that exists between groups as "polarization" can distract from what's really at stake, such as in conflicts over civil and human rights. Having negative feelings about another group is entirely natural if, for instance, that group seeks to deny your right to vote, or simply to exist. A high degree of animosity against fascists is a good thing. Nevertheless, efforts to reduce partisan animosity, support for undemocratic practices, and partisan violence may indeed help create the space and time for deeper fissures to be addressed.

A little more than a year ago, a coalition of multidisciplinary researchers at Stanford, MIT, Northwestern, the University of Pennsylvania and Columbia set out to crowd source ideas to address the political divide in what was dubbed the Strengthening Democracy Challenge. “Anti-democratic attitudes and support for political violence are at alarming levels in the US," said Robb Willer, director of the Polarization and Social Change Lab and Professor of Sociology at Stanford, at the time of the announcement. "We view this project as a chance to identify efficacious interventions, and also to deepen our understanding of the forces shaping these political sentiments.”

After reviewing more than 250 submissions from researchers, activists and others, the research coalition selected 25 interventions it deemed most promising to test against one another in an "experimental tournament" utilizing a sample of 31,000 U.S. adults. To learn more about the challenge, some of the promising projects that emerged from it, and whether tech platforms may play a role in efforts to address polarization, I spoke to Willer and his colleague, Jan Gerrit Voelkel, a Ph.D. student in the Department of Sociology at Stanford University and also a member of the Polarization and Social Change Lab.

What follows is a lightly edited transcript of our discussion.

Justin Hendrix:

So what does a ‘Polarization and Social Change Lab’ do?

Rob Willer:

We work on a couple different central political dilemmas in the U.S. So one concerns conflict between the left and right, which we see as impeding efforts to solve social problems, especially via any kind of legislation. And then we also work on other kinds of political dilemmas, including divisions on the American left that may get in the way of affecting social change through social movements and policy advocacy.

Justin Hendrix:

So a year ago you set out to do this challenge, the ‘strengthening democracy challenge.’ Tell me a little bit about it and what you hope to do?

Rob Willer:

So we were motivated by concerning public opinion data around levels of support for democratic principles, the concerningly high levels of support for anti-democratic practices that we saw in the general public, also concerned about ever increasing partisan animosity that's been expressed among American partisans. That's been increasing for about four or five decades now. We were also concerned about data that you'll see on support for partisan violence, which is not high, but I would say any non-zero levels of that are concerning.

So what we wanted to do was try to do the study that we could think of that could contribute the most practical knowledge for addressing these problems. And in contrast to the typical studies we run in the lab where we get an idea, we go and test it, and then we do a paper on it. We thought what would be better would be to try to find all of the good ideas that we could find for addressing these problems.

And so we sent out a call on social media for anybody to submit ideas for interventions, things that somebody could experience in an online survey that would improve their partisan animosity, their support for anti-democratic practices, their support for partisan violence. And we were blown away by the response. We had over 250 submissions from more than 400 researchers in 17 countries, just way more enthusiasm than we expected. And then we whittled that down to 25 interventions that we tested in a massive megastudy that we conducted earlier this year.

Justin Hendrix:

So I want to talk about some of the specific interventions, but before we do that, I want to talk a little bit about the overall methodology. So you use this term ‘depolarization’, depolarization as a term is not as studied.

Jan Gerrit Voelkel:

In this space, they were a lot of researchers and also practitioner groups like civil society or organizations who are trying to reduce partisan animosity, which is one form of polarization, oftentimes refer to as affective polarization, which you may contrast with other aspects such as attitudinal polarization, whereas attitudinal polarization refers to attitudes often on policies that are far away from each other, affective polarization refers to the decreasing dislike that partisans have for the other side.

And so there has been this research area building up that is trying to reduce this partisan animosity. And that is what we mean when we refer to or when we are using the term depolarization. But you can also extend this further and introduce other problematic attitudes that Rob has referred to. You could see anti-democratic attitudes. Some people have talked of that as being a potential consequence of polarization. So when we started out the project, we also did a deep dive into what actually are the outcomes that are worth studying the most. And then we decided to have a tournament with multiple potential targets, including partisan animosity, but also support for undemocratic practices and support for violence.

Justin Hendrix:

You've got some real heavyweights in political science and other disciplines that are on the advisory for this folks. Lilliana Mason, I see Jamie Settle. Who else was advising this project?

Jan Gerrit Voelkel:

So we really had experts from across the board. We have great scholars from sociology, for example, Chris Bail. We also have a lot of practitioners who are serving on the board. For example, Kristen Hansen, from Civic Health Project, who has been really a great source of support for the project. We have Eszter Hargittai.. We have Matt Gentzkow who is from economics. So we tried to represent leading scholars from across the social sciences to reach out to all of these communities and get the best ideas in this space.

Justin Hendrix:

So you ended up with 25 projects that you took forward. And maybe you can explain a little bit about the general methodology, because each of the projects essentially had, I guess, a similar kind of architecture or underlying methodology, in that they had access to this massive sample of survey participants that you recruited.

Rob Willer:

Yeah, that's exactly right. So we took 25 of these ideas that were submitted by participants in this challenge and administered them in a massive survey experiment. And what that meant was that we took 32,000 participants, a nationally representative sample of American partisans, and they came up to the study and got randomly assigned to either be in one of these ... To either experience one of these 25 interventions or to be assigned to a control condition, which is sort of like a placebo condition. And what we were… the reason we use this design is so that we could compare the effects that all of these 25 interventions had relative to this control condition on the participants' attitudes and their views and preferences.

We also were able in this design to compare them to one another. And this is a unique advantage of this design is that this is better than 25 research teams running 25 independent studies. I don't know if that's what would've happened in the absence of this, but something like that might have happened over a period of years, but when you run them all together, you're able to compare them against one another because they share the same sample. We use the same measures to see what the effects of the interventions were. So it's a huge advantage to do a study in this way.

And then on the back end, after people have experienced one of these interventions or the control condition, they then are administered a set of survey questions that assess things like partisan animosity, support for undemocratic practices, support for partisan violence, and a variety of other outcome measures that are related to democracy and polarization in the U.S. And so in this way, we can assess which of these interventions made a difference on these problematic, concerning, potentially problematic attitudes, and which ones didn't move the needle.

Justin Hendrix:

So I want to go through and just maybe talk about a few of them. We won't be able to go through all 25 of course, but just to give the listeners some sense of what some of the interventions looked like. One that stood out to me perhaps because I've had Jay Van Bavel from NYU on this show before– Common Identity, a project from NYU and the University of Cambridge. What was this project setting out to do?

Jan Gerrit Voelkel:

Yeah. So this intervention try to make the basic argument that Democrats and partisan share a common identity that, and then included a variety of information in order to make this argument. So that included providing information about more similarities between Democrats and Republicans than typically assumed, both in terms of policy, attitudes, with regard to any democratic attitudes or violence, also in terms of what their leadership supports, having quotes from Democratic and Republican elites, and then making the overarching argument that there is a shared interest and a shared sense of being American. So this really built on various different research programs to try to design and an intervention that was as effective as possible. And it turned out to be quite successful.

Justin Hendrix:

Can you speak to the results? How far was it possible to move people?

Jan Gerrit Voelkel:

So I would divide the results in two groups. So for example, the common identity intervention reduced partisan animosity by about 9 or 10 points, which is on a scale from zero to 100. So if you assume that common partisan animosity is pretty high, it might be around 70 or so. So then reducing partisan animosity by 10 points is actually quite a substantial decrease. And we also found in a durability test that the decrease was quite sustainable, even two weeks later. So as strong anymore, but it was still clearly reliable and detectable. On some of the other outcomes that we were interested in the decrease was what we would call statistically significant, but it was smaller in size. So for example, support for undemocratic attitudes was reduced by around 1.8, I think on the same scale from zero to 100, but the baseline level for support for undemocratic practices is also lower than, for example, partisan animosity.

Justin Hendrix:

I'm just kind of imagining that maybe that type of reduction might allow people to stay in the same room with one another for a little bit longer.

Rob Willer:

Yeah, I think certainly the partisan animosity effects speak most directly to that, in part because partisan animosity as it turns out is highly correlated with preferences for social distance from rival partisans. So people who have more benign attitudes towards rival partisans, less dislike and even hate, tend also to be more inclined to affiliate with, interact with, be willing to live near and so on rival partisans. And we found that both in the measures and how correlated they are. And then also in terms of the treatments, those interventions that successfully treated partisan animosity also tended to treat social distance preferences effectively as well.

Justin Hendrix:

So let's look at another one that I found interesting, the Road Not Taken, a reflection on ‘counterfactual selves’ as a means to reduce animosity and violence. This one comes from Fordham and folks at the University of California at Irvine. What's this reflection on counterfactual selves?

Rob Willer:

Counterfactual selves is a fascinating intervention and that it encourages participants to reflect on how their attitudes might be different if they grew up at a different part of the country, or just had a different path through life. And this is to me, a very intuitive sort of intervention that if you can reflect on the fact that if you were born under very, very different circumstances statistically, that the odds that you would have very different political attitudes are quite high. Assuming the different circumstances were living in a different region or living in a rural, growing up rural rather than urban or urban rather than rural or so on. So if you grew up differently, one would intuitively think as a sociologist and social scientist, that you would have very different views.

Interestingly, this intervention did not have large effects. Although I had expected it would. It had a small effect on partisan animosity. So views towards rival partisans, but really quite small. And I think this is one of the values of this project, is that it tells us not only which interventions work, which ones make a difference, and which ones make a big difference, but also those interventions that don't do as much as you might intuitively think, because this is one that I would've bet a lot of money on ahead of time, and I would've lost some of that money depending on how you arranged the gamble.

Justin Hendrix:

I assume too, that it may just point to the limitations of the method, that maybe there is empathy to be gathered from considering a counterfactual self, but maybe not in a survey?

Rob Willer:

Yeah, I think that's right. I think that all these interventions are susceptible to that concern that if they were implemented differently, they might have a different effect. And we can't be sure that they wouldn't have a substantially different effect if they were executed optimally, for example. I mean, counterfactual selves seems like something that, well this isn't a scalable way to implement, it would actually implement really well in a VR intervention, like a virtual reality intervention where you can really, really imagine you're in a completely different cultural background. You're growing up in a different setting that would make it much easier to visualize. People may find it quite hard to visualize that if they just haven't been in those circumstances.

Justin Hendrix:

And I know there are scholars I'm thinking of Courtney Cogburn and colleagues at Columbia who've actually done that, particularly with race and VR and trying to create that counterfactual.

So another one, which I found interesting, A Common Economic Plight and a Common Economic Enemy, from the University of British Columbia. This one sort of reminded me of the message of the Poor People's Campaign– I don't know if you're familiar with that– which seems to try to draw people together across the political divide by reminding people that there's an economic architecture in place, which they may share as an adversary.

Rob Willer:

Yeah, absolutely. I'm a huge fan of the poor people's campaign, which I think is phenomenal. And actually a lot of our research in the lab that focuses on moral bases of political persuasion really resonates with the Poor People's Campaign and how they approach things. Here, this was another intervention that I would've bet a good amount of money on doing very well in the challenge, the idea that you could create cross party class based, social class based solidarity is to me a really attractive method obviously in history, we see examples of it working very well and even mobilizing revolutions.

Here, in our study anyway, it was not a game changer. It didn't move the needle very much for the outcomes that we were studying. It didn't affect support for democratic practices or support for partisan violence. It did very, very slightly reduce partisan animosity, but really just a tiny and barely significant effect. And this was actually well executed as well. I thought this one was a quite polished implementation of its idea. And this is helpful knowledge to know that a well executed version of this idea, it might affect a lot of things. It might be effective for certain issue oriented advocacy. In fact, I bet that it would be, but it wasn't particularly effective at reducing polarization related attitudes.

Jan Gerrit Voelkel:

And it also speaks to your earlier question about if you would try to do the same idea in different ways, right? Because this is also similar to the other common identity intervention, trying to build up similarities between Democrats and Republicans. And we see that one of those common identity interventions did quite a bit better than this economic interest-based intervention. So that also speaks to that even if you follow similar abstract ideas, that the effects can be quite different.

Rob Willer:

And there's a long line of, oh, sorry. And there's also a long line of research in sociology and political science on how the U.S. features unusually low levels of class consciousness. And this challenge relative to Western Europe, Northern European countries, especially, or Latin American countries, and this intervention is trying to harness that to create cross party solidarity. And it may be going up against one of the big cultural public opinion findings about America, which is there's not as much class consciousness as you'd expect especially given levels of inequality.

Justin Hendrix:

I want to bring up one that I struggled with a little bit. So this one was called Correcting Overestimates of Opposing Partisans, Willingness to Break Democratic Norms. This one came from a team at the University of California at Berkeley, different parts of MIT, and Toronto's Metropolitan University. And the abstract says participants are told that most people do not know much about the other party. They are then asked to guess what people from the other party believe when it comes to actions that undermine how democracy works. For example, using violence block laws, reducing the number of polling stations to help the other party, or not accepting the results of elections if they lose. So, of course, this is a timely intervention to test in the United States. And so basically the participants got eight questions, you chose whether you were taking the quiz from the Republican or the Democrat kind of point of view.

And I suppose based on this survey of 30-odd thousand people, essentially the intervention hopes to show that for the most part, the alternate party may not embrace the kinds of anti-democratic behaviors that you might think. So why did I struggle with this one? So the idea seems to suggest that the average Republican is less interested in pursuing anti-democratic means to secure a victory, but of course the context we're in, of course makes it kind of difficult to swallow that one. I wonder, given where we're at the durability of the Big Lie, what we're seeing in terms of secretaries of state and other candidates in charge of elections that are embracing false election claims, that sort of thing. How does that one square? What am I missing about this one?

Rob Willer:

So I think you put your finger on something that's really, really tricky in this space, which is what's the role of partisan interest vis-à-vis Democratic principles. And then what's the role of misinformation in the Democratic problems that we're facing in the U.S. And just to recap the study, sorry, excuse me, the intervention that you're referencing. So what it did is it gave people corrective information about how much Democrats or whatever, excuse me, let me start that over.

So to recap this intervention, what it did was it gave corrective information to Democrats and Republicans about the levels of reported support for undemocratic practices in public opinion polling of their political rivals. So Democrats find out the actual levels of support for undemocratic practices that Republicans report in national representative surveys, and Republicans get the same about Democrats. And while support for undemocratic practices like these is I think concerningly high, it is nonetheless lower than partisans think their rivals report.

And so when you give people this corrective information, they then ratchet down their own support for undemocratic practices. Because I think to a great extent, people are supporting undemocratic practices because they think the other side is doing it at very high levels. We see a similar pattern with support for partisan violence. We have paper out of our lab earlier this year, showing that people overestimate how much their rival partisans report high levels of support for partisan violence. They overestimate 300-400%, like just massive overestimates. The levels of support for partisan violence are concerningly high, like 10 on a 100-point scale. We'd like to see that at zero, but they're not 40, which is what people perceive them to be. You give people that corrective information, then they rash it down their own support for partisan violence.

And that treatment tends to persist over time. So this is pointing to the fact that we have very exaggerated partisan stereotypes that likely come from our animosity towards rival partisans, but also our segmented media environments and social media environments.

So how do you square this with the fact that we see majorities of self-identified Republicans endorsing what we would call the Big Lie? And then we see, I think right now, a very small minority of major Republican nominees in the 2022 midterm election who endorsed the results of the 2020 election, according to one 538 study, it's under 20%. So at the elite level, we see a great deal of anti-democratic activity in the Republican party right now. That wasn't the focus of these interventions. They were correcting people's views of public opinion amongst rank and file Democrats or Republicans voters. But then your question would be, but don't we see majorities of Republicans denying the results of the 2020 election?

And we do. That is the case in public opinion data. However, if you asked those folks, is it okay to deny the results of elections that your party has lost? Most of them would probably say no, they would just say that isn't what 2020 was. Then according to what I'm hearing, 2020 was a rigged election. I'm hearing that from the sources that I trust in the media and the political leaders I trust. So I would not support that at as high of levels as Democrats think that I would, but I don't think that's the situation we're in. I support democratic principles. That's why I'm so worried about this stolen election.

So just to put a bow on this excessively long answer, I apologize, but it's a very nuanced thing. There's a whole suite of democratic problems we're dealing with that are much more better understood as misinformation problems.

And I think especially strong Trump supporting Republicans are denying the results of the 2020 election because they're putting faith in Trump and Trump affiliated Republican leaders. And that our research, especially on the democracy variable, might apply better to moderate Republicans who maybe don't believe the Big Lie, but are still voting Republican and are tolerating anti-democratic actions for the most part. And our research would speak to people who see this trade off for what it is, that they're trading off democratic principles for partisan, for gains for their preferred party, how can you intervene with those folks and change the calculus for them? So that's how I would think of it, as two related, but distinct problems.

Justin Hendrix:

I want to come back to the question about the polarization frame, but before I do that, I want to ask maybe both of you briefly, is there another project that we haven't discussed so far that you admired?

Rob Willer:

I thought another really, really interesting intervention was one we call Democratic Fear, which the way that it intervened was it showed a video of scenes of civic unrest following democratic collapse in a number of countries, and then culminated by drawing a connection to the capital riot that happened on January 6th, the last year. And this intervention was effective at increasing Democrats and Republicans support for democratic practices or reducing their support for undemocratic practices. It also improved partisan animosity.

Interesting, which I think I can I would say is because it sort of helped Democrats and Republicans see what's at stake. We sort of take for granted democratic stability in a country that's had relative democratic stability for 160 years, however, that can go away surprisingly quickly. And when it does, it looks really, really bad. And what's more, we have some credible evidence from just a year and a half ago of what it would look like if it started to go away. And it looks really scary.

Interestingly, that intervention increased support for partisan violence amongst Republicans. And my read on that would be that they saw this and either there was sort of a normative influence effect of seeing a bunch of Republicans being violent on January 6th; or possibly some anger against the intervention, perhaps, because they didn't equate the January 6th Capitol riot with these other scenes of democratic unrest. We don't know exactly why, but it's an important caveat for applying that intervention is you wouldn't want to come out of it and say, haha, we just need to show people these scenes of what it looks like when democracy collapses and that's going to help people attach some stakes to all this. You probably would want to test some different implementations that might be more effective or effective among Republicans without creating the support for violence backfire.

Jan Gerrit Voelkel:

For me, another intervention that I thought was a very smart idea was a short video about, or showing the gubernatorial candidates in Utah for the 2020 election that was submitted by Ben Lyons who's a Professor at the University of Utah. And in that video, the two speak about their commitment to accepting the results of the elections independent of which of the two candidates would win. So it was a great example of a very common strategy of having political elite endorse practices, setting a norm that then voters could follow. And here we find consistent effects for partisan animosity for support, for violence and for support for undemocratic practices, which were all reduced by this intervention. And I think the reason that this is so interesting is because it applies a principle and shows how actual political candidates in the real world can really do this. They can really have an impact if they want to.

And that influences voters who see these ads, so I think this could be really inspiring for political candidates across because the country either, if they want to do that themselves, or if they can find an opponent in both primaries or general elections to say, ‘hey, together we are going to accept that results of the election, or we support these democratic principles,’ and they can set the tone for a fair fight for whoever's going to win the election.

Justin Hendrix:

So I've got two questions left. They're not small questions, but hopefully we'll get through them reasonably quickly. As I'm sure well aware, there's a broader debate in scholarly communities about polarization as a concept, as to whether it's a useful frame in the political context we're in the United States. So you've got some referring to the animosity that exists between groups, whether polarization can distract us from what's really at stake, whether it applies in conflicts over civil and human rights where one group may be denying the rights or even the existence of another. Do any of the submitters grapple with these questions? And do you feel like the challenge grappled with these questions?

Rob Willer:

Yeah, I think it's an excellent question. I mean, I think that you're right, that for certain political problems we're facing, we might think of them as not being equivalent across Democrats and Republicans. I certainly think that way. And our data set allows us some purchase on that. So you can look specifically among Republicans at levels of belief in the 2020 election, for example. So you could analyze how these interventions affected that specifically. Or you could look specifically at Republican support for some specific democratic reforms, such as automatic voter registration and see, well, which of these interventions help to increase support amongst Republicans or just Democrats and Republicans in general on making it easier for people to achieve democratic access. So some of these questions that come up I think that our data set actually offers us a lot of purchase on and can be really helpful for.

So that'd be one thought I have. I think another thought is that when you look at data like the data we collected, it tells you a lot about the nature of the democratic problems we're facing in the U.S. I'll just focus on that aspect of political problems. So the levels of support for anti-democratic practices, undemocratic candidates, partisan violence, the ones that we observed were a little bit higher amongst Republicans, but I would think of this as more of a difference of degree than kind, because they're really only a little bit higher for the most part. The levels are really, really similar. So you might say, well, why do we see Republican extremists doing the Capitol riot? And why do we see Republicans denying elections, misinformation aside? Or why do we see Republicans who don't deny the results of the 2020 election still presumably voting in large numbers for Republican candidates who do in the 2020 midterms?

Why do we see that? And this research can help to speak to that. But one of the findings that you would get is that the levels of support in the general public are really for those kinds of things are really similar amongst Republicans and Democrats, but it's Republican leaders that are moving to capitalize on those opportunities and democratic leaders, thankfully are not. And I think that one thing scholars of democracy across cultures and time periods highlight is that when you have high polarization, you see a real risk for democratic backsliding because partisans stop serving as an effective check against undemocratic moves. And because they perceive too big of a trade off from switching to vote for a rival party or from staying home. And our results are very consistent with that.

So Democrats or Republicans are both in this highly polarized environment. You ask them ‘what if your in party candidate did this? What if they did that?’ And for the most part, they say, ‘you know what, I would still vote for them because the other party is just a non-option for me’. And that's true amongst Democrats and Republicans. And right now Republicans are clearly sleazy on that more.

Jan Gerrit Voelkel:

To add to that, I think that our study is unique in the sense that it can really speak to the problem of which problematic attitudes in the U.S. track with affective polarization. So that has been quite a debate in the literature for some time. When we studied effect, or at least when I studied effect polarization a couple of years ago, I had a lot of assumptions about what would be the consequences for democracies with high levels of affective polarization. And in our study, we try to actually measure if you move positive animosity with your intervention, which of these other potentially problematic attitudes are you also moving?

And we find that if you can reduce positive animosity to a strong extent, then as Rob said, that trickles down to reduce support for undemocratic candidates, that is a really important finding I think. However, we also find that interventions that substantially reduce partisan animosity do not really reduce support for undemocratic practices or support for violence reliably, suggesting that between partisan animosity and those outcomes, there might be not so much of a causal relationship.

So I think when we have these debates, it is really important to actually get empirical evidence on the question, to what extent these different constructs are actually related to each other. And one main takeaway of the challenge is, if you want to reduce anti-democratic attitudes, design an intervention that really focuses on reducing anti-democratic attitudes instead of trying to reduce polarization, and then assuming that's going to trickle down to anti-democratic attitudes.

Justin Hendrix:

How did you deal with this sort of possibility of false equivalence?

Rob Willer:

Well, one thing that we thought about when we were thinking about how to design the challenge was we thought about this question of, would we be creating a sense of false equivalence? And one thought we had was that if it's the case that democratic support from democratic practices and partisan violence are much, much lower than Republicans, then submitters would do better in the challenge by focusing on Republicans. If there's a false equivalence that a lot of people are falling for, that is truly false, totally false. In the sense of there aren't problematic democratic motivations and problematic levels of support for partisan violence that also are found amongst some Democrats, then a submitter that took that into account and just focused on Republicans would be a competitive advantage.

And so we thought let's not put our thumbs on the scale here. Let's let this play out and see if submitters will capitalize on some natural imbalance. And as it turned out there wasn't a big imbalance. Republicans were higher on these things that they weren't dramatically higher and interventions tended to target both sides pretty equivalently. They didn't make the bargain to just try to do really, really well with Republicans. I can't be sure that was the best strategy. It's possible that targeting Republicans specifically would've been better. But anyway, we set up a competitive environment wherein not falling for a false equivalence that wasn't indeed false would've been incentivized, if that makes sense. And we got the results that we did.

Justin Hendrix:

Underlying so many of these animosities– as Lilliana Mason and others have rightly pointed out– is the issue of race. Do you think that any of these interventions took that squarely on?

Jan Gerrit Voelkel:

I don't think that we have seen as many interventions taking that route as we had hoped that we would ge. It was another potential route. We wish there would've been more gender focused interventions as well. So I definitely think that there was still a lot of room for testing interventions that were very much focused on that. There's been a great work by Sean Westwood about the relationship between partisan animosity and racial animosity. So it is certainly possible that there is a lot of promise for interventions I feel to focus on that.

Rob Willer:

I'll just add to that. A couple of the interventions did feature warm relations between individuals of different racial or ethnic backgrounds and given how connected that is to party identification, that’s possible that contributed. Seeing positive relations across racial and ethnic groups may queue improved partisan relations as well.

Justin Hendrix:

And that may be something that the mega study framework could be applied to in future.

Rob Willer:

Yeah. I mean, I would love to see a mega study designed to reduce racial resentment amongst Americans. And there's been a project sort of like that was run by Calvin Lai, who tried to solicit interventions that would reduce implicit biases and has also, I think, taken this mega study framework to this space of workplace discrimination. But I think that work is still in progress. So there's folks that are thinking, how can we use this methodology to work on race related problems in the U.S.

Justin Hendrix:

So for my last question onto, I guess, another topic that is often muddied and difficult to discern in the literature of cours. This is the Tech Policy Press podcast, we talk about technology, social media, we've talked about, of course the role of social media in driving division and animosity polarization, we've talked about whether social media could be used to depolarize. Do you think any of these interventions could be scaled or the insights could be adopted in some way by technology firms to counter some of the ill effects of partisan animosity that you've detailed here? Can you imagine that happening?

Rob Willer:

Yeah. I think the results are applicable to social media platform design, and we really hope that they are, and we're eager to support that. I think that there's two, generally speaking, two ways you can take these results to applications. So one is operating on general principles that we've identified as more or less efficacious, more or less effective ways to move these problematic attitudes in the general public. So we come out of this research with a real clarified sense of what are the levers that you would want to press if you wanted to improve people's democratic attitudes. And that includes reducing their exaggerated out party stereotypes. That includes building a sense of common identity. That includes giving them a sense of what's at stake and if there is some sort of democratic collapse in the future that isn't a low stakes game. And these are some of the levers that you would want to know if you were doing platform design and you would also want to know that some other ones don't do so much. the counterfactual selves, for example.

So clarifying what does and doesn't matter for affecting these outcomes, I think is really helpful. And I know the colleagues of mine that work in platform design, they think all the time about, well what are the things on these platforms that make these problems worse and what are the ways you can make them better? They intervene in a wide variety of ways to try to improve these problems. And it's very helpful for them to have an improved toolkit I think. A second way you can use this knowledge that we've built here to intervene is more specifically.

So Jan talked, I think quite well, about the Utah intervention that we tested, which showed people a bipartisan endorsement of the 2020 election that came from the two Utah gubernatorial candidates that intervention could be just copied for in other electoral contexts and Facebook or Meta, whatever we're calling it, they do stuff to support democracy and voter turnout. They sometimes commission their own content to be created for those ends. They did for vaccine outreach as well. They could do something like that. That's bipartisan, it's not putting your thumb on one side of the scale and it's improving American’s sense of democratic integrity. They could recruit George W. Bush and Barack Obama to film an endorsement of the 2022 midterm elections and that their integrity should be trusted for example.

And I would say, regardless of whether they produce the content, they could promote it. If some other group created content like that, Facebook could help promote it with free ad credits or what have you. So there's a lot of opportunities for media platforms, social media platforms to make a difference in this space.

Justin Hendrix:

Will there be a ‘Strengthening Democracy Challenge Two’?

Rob Willer:

Oh, that's such a good question, Justin.

Jan Gerrit Voelkel:

Oh God.

Rob Willer:

Yeah, right. Yeah. Jan did so much work on this, I did too. It's exhausting to consider. I mean, a couple possibilities that have come to mind would be replicating some of the most effective interventions in another geopolitical context for example, a setting of either high levels of political animosity between partisan groups or a setting with some other kind of political strife. So that would be one thing we could do, or just use this structure of research study on a new question, like how to make Americans care more about economic inequality for example, or take action on the levels of concern that they already have. There's a lot of ways that you could take a design like this and try to wrestle with new questions and we'd be excited to do that next.

Jan Gerrit Voelkel:

It’s already been great to see other people using the mega study framework and trying to push that in different directions, so we can also learn from them. A lot of respect for Katie Milkman has done this very, very successfully over the last few years already. We've seen now an NYU project on moving climate change. We have seen that in many labs, projects across many different disciplines, but now also I think our lab is going to work on this, but we'll also be learning from all of these other scholars ... We're coming up with ideas about how to run, how to make our studies in a way that generates the most knowledge. And I'm excited about that.

Justin Hendrix:

Jan, Rob, thank you very much.

Jan Gerrit Voelkel:

Thank you.

Rob Willer:

Thanks Justin.

Authors

Justin Hendrix
Justin Hendrix is CEO and Editor of Tech Policy Press, a nonprofit media venture concerned with the intersection of technology and democracy. Previously, he was Executive Director of NYC Media Lab. He spent over a decade at The Economist in roles including Vice President, Business Development & Inno...

Topics