In social media discussions and on college campuses, fears of “cancel culture” are driving threats to freedom of expression. Are the fears justified? And why does online moral outrage seem to spin out of control? Nicholas Dias finds that “canceling” behaviors, such as calling out speech as offensive and retaliating against the speaker, are rare, even when stimulated by offensive speech. But Americans widely believe others will “cancel”, especially those from the other party. William Brady finds that people learn about the online norms of outrage expression from their social networks and conform to them, but they overperceive moral outrage online. Perhaps our behavior is not as bad as it seems but we react to what we see others doing. 

Guests: Nicholas Dias, Penn; William Brady, Northwestern 

Studies: “Speech Norms in Contemporary America” and “How social learning amplifies moral outrage expression in online social networks.”


Matt Grossmann: How misperceptions and online norms drive cancel culture, this week on The Science of Politics. For the Niskanen Center, I’m Matt Grossman. In social media discussions and on college campuses, fears of cancel culture are driving real threats to freedom of expression. What are people afraid of, and why does online moral outrage seem to spin out of control? Perhaps our behavior is not as bad as it seems, but like in other areas of social life, we react to what we see others doing.

This week I talk to Nicholas Dias of Penn about his new paper with James Druckman and Matthew Levendusky: Speech Norms in Contemporary America: The Realities and Misperceptions of Cancel Culture. They find that canceling behaviors such as calling out speech as offensive and retaliating against the speaker are rare, even when stimulated to think about offensive speech. But Americans widely believe that others will cancel frequently, especially those from the other party. I also talked to William Brady of Northwestern about his paper with Killian McLoughlin, Tuan Doan, and Molly Crockett: How social learning amplifies moral outrage expression in online social networks. They find that people learn about the online norms of outrage expression from their social networks and conform to them. He also finds that people overperceive moral outrage online. So what looks like a censorious culture might be a few people egging on the rest. Let’s start with the Dias project, which sought to look broadly at any behavior that people call canceling.

Nicholas Dias: Our study really has, I think, three key takeaways. The first is that Americans vastly overestimate the prevalence of canceling behavior. Americans themselves rarely engage in canceling, but they perceive canceling to be widespread among others, and particularly members of the other party. Second is that Americans misperceive the motives behind canceling as nefarious. Again, particularly when the other party is doing the canceling. Americans, when they do cancel, will describe their own behavior in positive terms and canceling behavior by members of their own party in positive terms, so as fair and empathetic. But they’ll describe identical behavior by members of the other party in negative terms, so biased, oversensitive. And then third, when Americans do cancel, they appear to be motivated by what was said and not who said it. So a person’s race, party, or status as a public figure has very little effect on these individual decisions on whether or not to cancel somebody. Instead, when Americans do cancel, they appear to do so in response to comments they consider disagreeable or offensive.

We think that this matters because some political commentators have begun to question whether America’s commitment to free speech is eroding. They suggest that the rise of cancel culture perhaps signals that Americans are now willing to constrain speech socially amongst themselves if not through government regulation. I think our study cast out on this idea that Americans widely embrace cancel culture or that this popular concern that cancel culture signals some fundamental shift in America’s free speech attitudes. I think our findings also suggest that these misperceptions themselves, beyond the canceling behavior, may be harmful to American political culture. So if people misperceive canceling as widespread and nefarious, it may lead to self-censorship and make productive political discourse more difficult. The fact that these misperceptions tend to surround the out party suggest that these misperceptions could exacerbate partisan animus and limit contact between people who disagree, which is obviously normatively not a good thing. I think also if elected officials believe that their constituents desire cancel culture in one form or another, they may enact laws based on those flawed beliefs, like say the don’t say gay bills that we’ve seen in several states.

I think ultimately, having an accurate picture of what cancel culture is is essential for addressing it if we want to address it at all, and for understanding American speech norms.

Matt Grossmann: You don’t shy away from using this term cancel culture or calling these things canceling behaviors. So talk a little bit about that popular motivation for the piece and maybe the story behind your collaboration and how you went about connecting that to research.

Nicholas Dias: Sure. Yeah. So there’s no agreed-upon definition of cancel culture, but based on our review of news coverage on the topic, we think it’s pretty safe to say that most definitions refer to a practice of silencing or ostracizing somebody that is canceling them for making offensive comments as opposed to, say, engaging those individuals in debate and dialogue. What we might refer to as counter speech, which is sort of traditionally in the political tolerance literature or an American legal literature have been held up as the right way to deal with offensive speech.

As for whether cancel culture represents like a real trend, certainly the news media are increasingly covering cancel culture in name. So in doing background research for this study, we found that just six outlets, so the New York Times, the Washington Post, Wall Street Journal, CNN, MSNBC, and Fox News mentioned cancel culture in over 2300 articles or TV programs between January 2019 and October 2021, so a lot of coverage on this topic that is overwhelmingly negative. But it’s really not clear to us at least whether this crisis about cancel culture signals any real shift in Americans’ behavior or attitudes towards free speech. Our findings strongly suggest that canceling behaviors are not widespread in the American public. Many of the things that people consider cancel culture, so criticizing someone’s speech, boycotting or protesting events, complaining to a person’s employer, have been with us for a very long time, since the beginning of America. But perhaps this behavior is more visible now that our lives are documented on social media.

Matt Grossmann: Let’s get concrete here first about the things that you were calling canceling behaviors and then about some of the statements that you gave people to determine whether they would engage in those behaviors.

Nicholas Dias: We had an experimental component to our survey where we presented participants with scenarios wherein different speakers made different statements that many on the political left or right might consider offensive. These were distillations of comments for which people were supposedly canceled according to the news coverage that we reviewed. Our left-leaning statements included things like, “America is a racist nation, and all police are bad,” whereas our right-leaning statements included things like, “Confederate statues are about America’s heritage and are not racist.” or “There’s no such thing as transgender, only male and female.” And that’s based on the comments that we saw in our news coverage and who was saying them.

In terms of the particular canceling behaviors we looked at, we asked the gamut, so we included everything from criticizing a person on social media to complaining to a person’s employer to boycotting a person’s event or merchandise to doxing. We really tried to deliberately cast a broad net in terms of the behaviors that we looked at so that we could capture the full range of perceptions. But of course, I should note that the status of any of these behaviors is constituting cancel culture as contentious. So respecting that, we analyzed our data by behavior and ran several robustness checks. The patterns we see in terms of misperceptions and what motivates canceling behavior are similar across behaviors, though Americans engage in some canceling behaviors more often than others.

We ran a survey with a sample of around 1,800 American adults that was representative of the American population with regard to age, gender, education, region, race, the usual suspects. And so, our survey had a descriptive component and an experimental component. In the descriptive component we asked people what they had themselves done and what they had personally seen others do in response to offensive speech. This let us get a sense of how much people actually engage in canceling behavior relative to their perceptions of how much canceling happens. And then we had an experimental component. We gave participants scenarios with different sorts of speakers making different sorts of statements, again, statements that many on the political left or right might consider offensive, and then we asked participants which of several canceling behaviors they might be likely to engage in in response to that scenario. That let us get some purchase on what actually drives canceling behavior insofar as it happens. We also asked participants how the typical member of their own party and the typical member of the other party might react to identical scenarios.

Matt Grossmann: You say people don’t engage in this all that much. Give us a sense of how often people report engaging these behaviors and how often they do so in response to an offensive comment and I guess maybe the hierarchy a little better that I assume that there are some things that people do much more readily than others.

Nicholas Dias: Sure. Yeah. A minority of Americans engage in canceling behaviors, but the frequency varies considerably depending on the behavior. At the bottom of the range, as little as 3% of participants reported engaging in doxing. So that’s posting a person’s personal information online. In the middle of the range, around 8% report complaining to an employer about something offensive that an employee said. And then at the top of the range, around 26% report having criticized someone on social media. But of course, social media criticism can encompass anything from the harshest of vitriol to pretty lukewarm comments.

Nicholas Dias: In the experimental component of our survey, we find that insofar as people do engage in these sorts of canceling behaviors, they aren’t really motivated by who makes the statement. Again, in these experimental scenarios we’re manipulating the party of the person making a comment, the race of the person making a comment, whether or not they’re a celebrity, an elected official, college professor. Those differences in terms of who’s making a statement really don’t seem to matter much to people. What really seems to matter is the statement itself, the content of the statement itself. So people are motivated by what someone says and in particular whether or not the statement is disagreeable, which they tend to find offensive.

Matt Grossmann: You also report that people say that they see cancellation by others or have seen it. What do you think that they are reporting here? Is this a situation where just one instance can just be reported by a lot of people or it’s particularly memorable? Are people just reporting things they saw on TV or things that they saw on social media here rather than directly experienced?

Nicholas Dias: I’ll start by summarizing. Despite the fact that few people actually engaged in canceling behavior according to our data, many of our participants reported having seen others engage in these behaviors. Those perceptual gaps are very, very large. So for each behavior that we looked at, participants reported others were more likely to engage in canceling than themselves by at least a factor of two. For example, whereas 7% of participants had boycotted or protested at an event, 37% had seen others do so. In a really striking example, participants were 10 times more likely to have seen someone else engage in doxing, again, sharing somebody’s personal information online, than to have done it themselves.

We can’t say definitively where people saw these canceling behaviors or why these instances stuck in their minds. We do suspect this perception that canceling is widespread reflects the media’s voluminous and overwhelmingly negative coverage on the topic. With media coverage in high profile cases, people might develop these inflated perceptions, and people have a tendency to report the frequency of events on the basis of the ease with which they can of draw examples to mind. We think that’s probably at least some of what that represents.

We can also say that perceptions about the out party are particularly extreme, so that is people thought members of the out party were especially likely to engage in canceling behavior. Respondents overestimate the likelihood that those from their own party will cancel others by a factor of just one or two, but when it goes to the other party, it’s a factor of two or four. I think that also suggests that party identification and partisan biases are also feeding into these perceptions. People perceive out-group members more than in-group members to be sensitive to negative rhetoric. That’s a very old finding in communication, third person effect. And so far as those perceptions are particularly extreme about the out party, that probably helps Americans to maintain this perception that their political party is more virtuous than the other political party.

Matt Grossmann: Let me push back a little bit on, I guess, whether this is a misperception. It seems like some of the more prominent examples like someone being supposedly fired from their job or reported on by their employer wouldn’t require a whole lot of people engaging directly in the doxing or the contact the employer. And even the criticism, of those people who are responding to the person online, it would mostly be critical. I guess, how much is this really misperception rather than just a phenomenon where most people are bystanders but the overall effect is to look like a cancellation?

Nicholas Dias: I think some of our behavior is consistent with this interpretation that it’s not that people are misperceiving the prevalence of canceling behavior, but they’re accurately reporting on these really prominent examples of canceling behavior and they’re basing their judgments on those really salient examples. I will say though that we have really strong consistency between our descriptive findings and our experimental findings. These gaps and perceptions extend to judgements of individual scenarios where people have to judge how likely it is that a typical member of the in party or the out party would respond to a particular situation. I think it’s possible. I think probably the more likely explanation, at least from our perspective, is that people are building inflated perceptions on the basis of this widespread and overwhelmingly negative media coverage.

Matt Grossmann: There was a partisan difference. Republicans were slightly less likely to cancel themselves and much more likely to see it as a problem and to complain specifically about wokeness and political correctness, the terms that are used with this in the popular media. On the other hand, I guess you say you saw fewer differences than you were expecting in the actual behavior. So reconcile that for me. To what extent is this consistent with the idea that there is just a partisan difference on this behavior versus this is something that everyone does in the right circumstance?

Nicholas Dias: So while we see really large gaps in terms of partisan perception about who engages in canceling behavior, we see very, very modest differences in terms of actual behavior by party. Republicans and Democrats are perceiving canceling behavior in really similar ways, so seeing themselves as the least likely to cancel, in partisans as somewhat more likely to cancel, and out partisans as the most likely to cancel. Members of both parties attribute positive motives to the canceling behavior of in partisans and their own canceling behavior versus negative motives to the canceling behavior of out partisans. Again, both Republicans and Democrats seem to be motivated to engage in canceling behavior by the same things. Again, not motivated by who makes a statement but what the statement is and whether the statement is disagreeable or offensive.

The main party difference that we do see is that Republicans overwhelmingly perceive cancel culture to have a net negative effect on society, whereas Democrats are more divided, although they also skew negative in terms of their cancel culture perceptions. We do find that Democrats are slightly more likely to cancel than Republicans, but those are pretty modest differences and are likely due to the fact that they perceive cancel culture less negatively. So despite this, Americans tend to perceive members of the out party to engage in the lion share of canceling, and I think in particular the political right and even former President Trump have made this argument that cancel culture is the domain of the political left, it’s something that Democrats do, not Republicans, and we definitely don’t see any evidence of that, that big asymmetry.

Matt Grossmann: So you do find that it’s about the statements themselves rather than the people that give the statements, but that does seem a little consistent with, I guess, critiques from the left end and the right, so I’ll do both. It seems like from the right that is what people are saying, that we’ve redefined something that is just something I disagree with as something that is offensive. From the left, they would say, “Well, your statements that are made by people on the right are offensive to particular minority groups.” Whereas, your statements that are more people on the left are not of that type, They’re not calling out a specific minority group and making an offensive comment about them. So I guess is there anything in your study that would help us figure out maybe between the different kinds of statements you asked about or anything like that that would help us figure out if either of those critiques is valid?

Nicholas Dias: Yeah, so it’s difficult to say with our data whether people are motivated by strictly offensiveness or whether, as you’re saying, we’ve redefined what is disagreeable to be offensive, which shouldn’t necessarily be the same thing. In our data, ideological disagreeableness, offensiveness, and people’s willingness to cancel our pretty correlated. There’s only so much we can say from our data. But that said, I think other recent work, particularly work from Jesper Rasmussen who’s looking at support for government regulation of hate speech suggests that statement offensiveness or severity is really what drives people’s perceptions of what speech should be curtailed.

Matt Grossmann: Now let’s turn to the second project which looked more specifically at social media, including all outrage expressions, not just the behaviors called canceling. Here’s how Brady summarized their project.

William Brady: Broadly I’m interested in studying the ways in which new social technologies impact the ways that we communicate about morality and politics. In this study, we were interested in testing whether our expressions of moral outrage are affected by the ways in which social media platforms deliver social feedback to us and also the ways in which social media platforms display other people’s messages to us in our newsfeed. Just to be clear, by moral outrage I mean, for example, the feelings of anger and discuss that tend to be invoked when we feel that someone or something has transgressed against our sense of right and wrong. Moral outrage is really important to the study because it guides norms by giving people a sense of they might be punished if they’re misbehaving, in other words, if they’re going against our understanding of right and wrong. That’s a general overview.

And then the key findings of the paper where that people’s outrage expressions on Twitter can be explained at least partially as a function of the amount of positive feedback that they receive from the platforms every time that they express outrage. In other words, people who receive more likes and shares in the past when they’re expressing outrage, they start to express more outrage over time. This squares with our general understanding of reinforcement learning. So in simplistic terms, social media platforms can act almost a little bit as a skinner box where we get rewarded for certain types of expression through social feedback, and then that behavior increases.

In our case, we find that outrage acts in that way, and it might even be particularly susceptible to this type of learning. But the one thing I do want to mention is it wasn’t as simple as reinforcement, so that’s not the whole story. We also found that people are also influenced by social norms of their network, and that’s determined by the types of expressions that they’re seeing when they log in and view their newsfeed. So people who saw more outrage in their feeds, they tended to express their own outrage in response more frequently than people who saw less outrage in their news feeds.

Matt Grossmann: Give us some concrete examples of what an outrage expression might look like and some of the real-world events that motivated this paper.

William Brady: One of the contentious US political topics that we studied in this paper was the Brett Kavanaugh confirmation hearings in 2018. Most listeners will remember this. On the political left, there was outrage in response to Kavanaugh being considered as a Supreme Court justice because there are, of course, the allegations that he sexually assaulted Christine Blasey Ford among other women. But then on the political right, there was all this counter outrage, you could call it, because the political right thought, “Well, these allegations are largely just a political tool to limit the ability for right wing Supreme Court justice to be confirmed.” So it was a case in which there was outrage on both sides, sometimes for different reasons, but generally about the same topic.

And then we also studied lots of other events that elicited outrage in American politics. Some of the other examples were, this was I believe 2019, the Trump administration’s ban of transgender individuals serving in the military. If I recall it, at that time it had reversed a previous decision by the Obama administration. There were several examples like this. What we were looking for were these cases where there tended to be swells of outrage on social media from both the political right and political left. And that allowed us to look at the dynamics not just on one side, but on both sides of the political spectrum.

Matt Grossmann: You talk about the social feedback encouraging further outrage expression. Obviously, we know in a Twitter context that this is about likes and retweets. How specific is that to this being about Twitter specifically or social media at least in thinking about how people are behaving?

William Brady: Yeah, that’s a really important question. I think any platform that delivers some form of social feedback in response to people’s posting behavior, that’s going to be prone to the effects of social reinforcement that we document in the context of Twitter. So that includes Facebook, Reddit, TikTok, Instagram, YouTube. They all have their own flavor of likes or thumbs up, things like that.

Now, any time you’re getting that social reinforcement, it can encourage your behavior or it can regulate your behavior if you get less than you expected. That’s another form of social reinforcement. The extent to which different types, different forms of feedback across the platforms translates to higher or lower amounts of social influence, that’s not something that, as far as I know, has been empirically tested. But thinking about platforms including online versus offline, we’re generally responsive to social feedback everywhere. Social learning is a fundamental aspect of nearly every human’s experience, including non-human animals. But the idea is that new social media platforms, they’re actually amplifying our natural tendency to learn from others’ feedback because the feedback that’s delivered on these platforms, I think, arguably it’s unique. When I say unique, it’s delivered in ways that’s more quantifiable than you would get in a face-to-face interaction. So exactly how much likes and shares you’ve got compared to another time. It’s giving you literally in a number. And that’s highly salient because as soon as you get on and get notifications, you see the numbers pop up. It’s also sometimes even delivered to us in intervals that we know from reinforcement learning literature makes us more likely to repeat behaviors that are rewarded. There’s several reasons to think that the social media platforms are amplifying this natural tendency we have to learn from social feedback in any context.

Matt Grossmann: You specifically looked at people who were basically fighting culture war debates online, which might be a very particular subset of the population where most people are not really engaging in. You also had an experiment that tried to see if similar mechanisms were apparent for everyone. But I guess to what extent is this just a phenomenon about a small subset of people who want to argue about these things online?

William Brady: I’ll talk about this very generally. On one hand, we know that on Twitter and other social media platforms, especially Twitter and Facebook, a minority of highly active political users are typically accounting for the majority of outrage and extreme political content. So that’s definitely true. I think there are exceptions though, for example, major cultural events that cause even the less politically active person to express outrage. One example that comes to mind would be Trump’s electoral victory in 2016. I think it made a lot of people express opinions or express outrage just because it seemed to be very important on both sides of the political spectrum.

But an important point that I want to make is that, on social media, the majority of people who are surveyed are still reporting that they’re seen in their newsfeed or otherwise political content even if they’re not themselves producing it. This is partially at least because content algorithms of the platform that determine what people see, those outbreak political content because other people tend to be more likely to engage with that. And that often can be, for example, outrage evoking content. There is this issue of representativeness, I like to think of it, on the platforms. Even if you and several people on your social network wouldn’t necessarily produce the political content yourself, you might still see it in the form of more extreme people expressing opinions, and it pops up in your feed. You might also just see clickbait news coverage of things that are deemed as cultural wars. And so, even if a minority of people are producing it, if that content is spreading around, it can still have some influence. And that’s a big problem that we’ve been thinking about in our research.

Matt Grossmann: You also experimentally manipulated feedback and asked people to respond to that. So talk about the setup of that experiment and what you learned from that.

William Brady: Yeah, I mean, broadly speaking, the purpose of experimentally manipulating feedback in this experiment was really to replicate the findings that we found in the field on Twitter. We were using observational data in that case. As you mentioned, there are some selection effects because we tend to be studying people who are already expressing outrage, although we do go look back at their history of tweeting behavior. So really, the idea is just to, A, replicate that finding, and B, show that there is actually a causal effect of social feedback and norm learning influencing our outrage expressions. There’s a little bit of an issue of generalizability and then an issue of being able to make that causal inference specifically. So really I think that was the main point of that experiment. When we look at the field studies and the studies done in the lab, you get this general sense that this is an important mechanism for explaining our outrage behaviors online. The mechanisms I would be referring to, of course, would be the social reinforcement and the norm learning.

Matt Grossmann: I don’t think we’ve explained the finding about the interaction between those two, so do you want to do that, just that they somewhat trade off with each other? If you have more totality of the behavior that you’re hearing, it doesn’t matter as much on the reinforcement.

William Brady: Exactly. One of the findings that you’re referring to was that people in more extreme networks, and these are the people who tend to learn from the norms displayed in their environment to express more outrage because there’s more outrage in these extreme networks, they’re actually less sensitive to the effects of reinforcement learning. So the social feedback they’re getting does not predict their outrage expression to the same extent. This could be explained by the fact that if you can directly absorb… sorry, observe the norm, you really don’t need to use the feedback as much to guide your behavior. Because basically as humans, we’re constantly trying to make inferences about what is acceptable in our social network. If we have one source of information that leads us in the right direction, we think we might discount the other information or it might not impact us as much.

Now, the other explanation could be that some users actually come to more habitually express outrage, even beyond it’s social value. Our paper certainly didn’t disentangle these two explanations, but there is some evidence coming out now from other labs that some people’s behavior on social media can be predicted in terms of habit-like behavior, but it really seems to depend on their history of the platforms. This tends to be only people who have long histories of using the platform, and they’ve been engaging in certain ways for several years.

Matt Grossmann: Brady sees the two projects as offering complimentary explanations.

William Brady: What my research is suggesting is social media can create situations where there is outrage or this understanding of extreme political attitudes or even views about cancel culture, they get overrepresented in people’s news feeds. Or it’s just an issue where the media chooses to focus on that or political pundits, and that narrative is still spreading around even if it’s not necessarily representative of everyone. I think that is generally a case in which you are biasing social information in ways that impact our social learning processes. And so, again, whether that’s conforming to it in a false way, this is generally referred to as a situation of pluralistic ignorance, or whether you’re retaliating to something that you think your out group is doing, even though not everyone is believing that, both of these to me are complimentary processes of the more general issue that we’re responding to social information that is not necessarily representative of the base rates that are represented in our social network.

Matt Grossmann: Dias sees more differences between the two mechanisms, but is open to both.

Nicholas Dias: I don’t think that we were too surprised to find that people perceive cancel culture to be something that the other party does rather than something they themselves do or members of their own party do. I think what that suggests is that cancel culture has become yet another domain of partisan conflict. We know that Americans love people on the other side of the political spectrum more than ever previously measured, at least in surveys. Republicans and Democrats will describe one another as hypocritical or close-minded and are uncomfortable having one another as romantic partners, children, in-laws, friends, neighbors, et cetera.

I mean, given that Americans widely perceive cancel culture as having a net negative effect on society, certainly Republicans perceptions are much more negative, but Democrats perceptions are also negative, but they’re somewhat more divided on the issue, given that cancel culture is generally understood in these negative terms, I think it’s pretty natural that participants would think members of the other party are the ones doing the canceling. Again, because these partisan biases are so widespread, because this helps citizens maintain a positive perception of their party vis a vie the out party.

We are looking at individual decisions to cancel somebody or not. That’s true. We weren’t specifically geared to look at what role social norms are playing in driving those behaviors. I think that’s a really interesting subject, and I think something that we’re looking to explore in future studies to understand the extent to which people perceive that they are expected by, say, members of their own party, for example, to engage in these behaviors or perceive that by engaging in these canceling behaviors they can signal their devotion to the party. I think that’s very possible. Not something that we looked at.

I will say, though, again, our study wasn’t geared to look at that specifically, but I can say that we looked at the correlation between people’s reports of having seen canceling behavior in the wide world with their behavior in our experimental scenarios, and we don’t see any correlation there.

Matt Grossmann: I also asked Dias about whether cancel culture is mostly the kind of Twitter behavior studied by Brady.

I wanted to reflect a little bit on the extent to which this is a specific phenomenon about a specific social network or a specific set of social networks versus a wider phenomenon about interpersonal behavior. In other words, when people are talking about cancel culture, is the quintessential example this social media and especially Twitter behavior?

Nicholas Dias: Yeah, so I certainly think the most salient examples of cancer culture to have taken place on Twitter. I think that’s fair to say. In terms of to what extent perceptions of canceling behavior feed in to the actual behavior that people are themselves willing to engage in, Republicans and Democrats alike, as I’ve mentioned, perceive cancel culture in negative terms. So I’m pretty skeptical of the idea that social norms are playing into canceling behaviors. Again, we didn’t design the study to understand how norms affect canceling behavior specifically, so we can’t make any really strong statements. The best evidence we can offer is that the correlation between people’s reports of how much they see canceling behavior and their own willingness to engage in canceling behavior in response to the experimental scenarios that we gave them, there was no correlation there. It was a pretty precise no.

Nicholas Dias: That said, it’s really easy to imagine how someone who wouldn’t necessarily be willing to be the first person to criticize someone on social media, for example, but might be willing to pile on after others have criticized someone. Again, I think this is an interesting possibility that we’re hoping to explore in future studies.

Matt Grossmann: Are these problems in need of a solution or cooked-up controversies? Brady is cautious in offering solutions when some moral outrage is justified?

Should we be trying to reduce these expressions of moral outrage especially if they get misperceived, they might lead to self-reinforcing cycles? There do seem to be some reasons to be concerned about them, so should we be trying to reduce them? And have you identified part of the problem that we don’t seem to be able to so far?

William Brady: Yeah, so just to be clear, I definitely don’t want to… Well, what I do want to say is that anytime you have inaccurate social perceptions that leads to outrage, that is a bad thing. I think we can all agree on that. You don’t want your outrage to be in response to something that is not necessarily an accurate understanding of the issue or of the social attitudes underlying some issue. I think we can agree on that. Any attempts to correct that social information I think is very important.

What I was arguing earlier is just more when we think about how do we actually implement that on a social media platform. My point is just that I think understanding the interaction of our social learning processes and the affordances that are created or constrained by having to interact on these platforms, that’s something that’s really important, and it’s not something that people can think about. We have some data suggesting that people understand that algorithms are influencing what they see, but what they don’t make the connection about is that that means that some of the social information they see is not necessarily representative because the algorithms are [inaudible 00:39:21] content do. Not manipulating, but they’re influencing the content for goals that are different than just what’s the most accurate thing in your social network. It could be profits. It could be engagement. It could be whatever is going on. So I think that’s the key that it definitely will help people to have a better understanding, but anytime you have the ability or you have a chance to correct social information, the accuracy of that, you should definitely do that.

Matt Grossmann: I guess so on the other side, what if I’m a practitioner trying to create some outrage? It seems like I have a pretty clear model here. I need to be creating a sense that everybody else is outraged about this and constantly making sure I’m providing feedback to everybody who goes along with it. Is that what political actors are doing, and is that what they should be learning?

William Brady: Well, let me answer that by talking about what I see them doing rather than suggesting what they should do. Surely it’s the case that one strategy on arguably both sides of the political spectrum… I don’t research ideological differences in this, but I’ve seen it on the right wing media for example. If you spread around or point out a minority viewpoint, even if it is a minority viewpoint, it begins to gain legitimacy the more it spreads and the more that people are forced to engage with it. That’s actually a very dangerous way to let a minority view, for example an extreme view, have prominence and take up space in political discussions. That’s actually one thing, speaking about practitioners, rather than thinking about, “Well, what can I do to provoke outrage?” “Well, what can I do to reduce it?

I mean, you have to be aware that anytime you want to comment or critique a minority view that you see, you might not even know it’s a minority view, you might just say, “Look at what crazy thing the out group is saying now.” Well, again, you’re still spreading that view around anytime you share it. What I worry about is that view gaining legitimacy even if it only had 1% of the population believing that. Once you start gaining people to engage with, it increases the chance that more people will actually, ironically, come to view it as legitimate. And so, that’s something that I would tell people make sure you’re paying attention to. Think twice before you share that view that you just want to say, “Oh, that’s so wild.” You might just talk about that offline where you’re not spreading it around as much.

Matt Grossmann: Dias also does not think that they have necessarily uncovered a problem to be solved.

Nicholas Dias: Our findings, I think it’s safe to say, are more consistent with the view that this isn’t a widespread problem. And to the extent that people perceive that it is a problem, it’s probably a function of elite communication about this topic in particular, media coverage and this really focused political rhetoric on the right. I take the point that we are taking this rhetoric from the right, that this is a serious problem. We’re taking it seriously. But I think that’s important in terms of scoping the problem. It makes a difference whether or not canceling is actually a culture, that is to say this is a set of behaviors that’s widespread in the American public versus something that’s engaged in by a really slim but particularly engaged minority. I wouldn’t say that we’re buying into that frame, but I do think it’s important to take it seriously. Again, we hope that the hard empirics that we provide with this study can help inform that conversation.

Matt Grossmann: But Brady is working on further research to figure out how to affect people’s polarizing behavior.

William Brady: My latest line of work in this area is thinking of ways to help people learn about social information in the context and reality and politics in the most accurate way possible, especially when they’re learning about it through the influence of social media platforms. I mentioned the educational intervention. The other thing that I’m interested right now is thinking about ways to get people to understand the people from opposing political point of view by really getting them to understand the hierarchy of moral values that people have and to learn that actually they’re more flexible than people think. Because I think the more that we tend to think about other people’s moral views as extremely hard and stubborn, the less we think it’s actually possible that we have any chance to have a productive conversation with them.

Matt Grossmann: MBS will also be following up on Brady’s work.

William Brady: I think our next step will be to examine to what extent canceling behavior is driven by, again, this perception that once political party expects somebody to engage in canceling behavior or the belief that individuals can signal their devotion to the party by engaging in canceling behavior. I think that’s a really interesting point, and I think what we’ll be doing next.

Matt Grossmann: There’s a lot more to learn. The Science of Politics is available biweekly from the Niskanen Center. I’m your host, Matt Grossman. If you liked this discussion, I recommend checking out these episodes, which all focus on misperceptions and how they drive behavior. Do Americans implicitly trust government despite our public anger? When liberals and conservatives use genetics to explain human differences. Did Facebook really polarize and misinformed the 2016 electorate? How news and social media shape American voters. And how online media polarizes and encourages voters. Thanks to Nicholas Dias and William Brady for joining me. Please check out Speech Norms in Contemporary America and how social learning amplifies moral outrage expression in online social networks, and then listen in next time.

Photo Credit: iStock