Not Another Politics Podcast – Episode 11

The killings of George Floyd, Breonna Taylor, Rayshard Brooks, and many other black people at the hands of police have driven nationwide protests. To be true to our mission, we want to look at this complex moment through the lens of research.

No paper is getting more attention than Princeton Asst. Professor Omar Wasow’s “Agenda Seeding: How 1960s Black Protests Moved Elites, Public Opinion and Voting”. On this episode, we discuss the substance of the paper and more.

Listen on Apple Podcasts or wherever you enjoy podcasts.

Transcript:

William Howell:           

I'm Will Howell and welcome to Not Another Politics Podcast.

So for today's show, we're going to do something a little bit different. There is widespread unrest and outrage and anger in the aftermath of the police killings of African-Americans around the country. There's a lot of talk right now, as we observe these protests, about not just what the protests mean in their own right, but what kind of politics they're setting in motion. In this moment, we have the opportunity to interview Omar Wasow from Princeton University. He's an assistant professor in the politics department there, and he's written a paper that speaks to and is actually caught up in this particular political moment that we inhabit. It's entitled “Agenda Seeding, How 1960s Black Protest Moved Elites Public Opinion and Voting.” It's forthcoming in the American Political Science Review. So my interview with Omar ran a good deal longer than most of the interviews that we conduct on this podcast.

And we thought it important to give it a full airing on today's show. And I'll say that this is a paper that has received a tremendous amount of media attention. And for the most part, the kind of attention that's been given to it is pretty superficial. And it tends to some of the findings, but without trying to dive deeply into where the findings come from, or without trying to be sensitive to the extent to which the findings that he calls from the 1960s do in fact, speak to this present moment. And so we thought it important to air this interview more fully so that you have a better sense about what this paper says, what the underlying claims are and the nature of the evidence that support those claims.

So you've written this paper at quite an extraordinary time. There have been protests and uprisings all across the country. And I think there's a lot of interest in trying to make sense of what's at stake in this moment. And we'll come back to this moment later, but I'm hoping we can delve deeply into this piece of research that you've written. And it's a part of a larger project that you're involved in, Omar. Could you just say a little bit about how you came into this topic, what your interest is in it?

Omar Wasow:              

So I went to graduate school in significant part because I was interested in what caused a turn in our political culture from a focus on civil rights, to a more punitive set of criminal justice policies that undergirded the rise of mass incarceration. So how did we go from the Civil Rights Act and Voting Rights Act to law and order and the war on drugs. And as I dug into it, I began to see that the tumult of the sixties both contributes to shifts in public opinion in favor of civil rights, but also shifts in public opinion on this taste for law and order. That led me into a more deep dive specifically on protest movements and some of the different ways those movements can have political consequences. I think it's worth stepping back, even one kind of question before and saying, do protests have effects and in political science, the overwhelming consensus is that elites dominate politics.

There isn't a way for marginal groups to have much of an impact. And if you begin with that presumption, then we should expect that protests do not move politics, that marginal groups do not have voice in an elite dominated system. And to be clear, there are folks take Lee, Erica Chenoweth and Maria Stephen, Dan Gilleon who have found protests influenced politics. So I'm not the first to do this, but we are a kind of statistical minority in political science, arguing that there are these kind of bottom up effects in politics. Then we get to a sort of question of what are the multiple ways that protests might influence kind of larger political outcomes we care about. And I think it's a really helpful question because my paper focuses pretty narrowly on how protests influence media.

But for example, protests help you generate money. If you're doing to do fundraising, having a high profile event helps build organizational capacity, people are drawn to your movement, and that's an important part of kind of building a social movement. There are ways in which low profile organization can have long term effects that don't show up as a protest. So there are all of these ways that a protest might be working, partly about building infrastructure and organizational capacity, fundraising, but also other methods that may leverage different kinds of power, like the power of boycott, or as I focus on in my research, the power to generate influential media.

William Howell:           

Right. And the generation of influential media may in turn affect voting behavior, which you document. Just to situate where the kind of effects that you are documenting, it has to do with the effects of protests on what the media talks about. And in turn, how changes in what the media has to say, how that affects willingness of people to vote for the democratic party. Do I have this right?

Omar Wasow:              

Yeah. I mean, I think that the simplest version of the story is newspapers don't tell people what to think. They tell people what to think about. And that is sort of the roots of a large literature in political science called agenda setting. The kind of core story of the paper is, before the media go about making their editorial decisions about what to prioritize, there are marginal groups, not just marginal groups, everyone from the President down to marginal groups is trying to tell the media what to focus on. And so agenda seeding, I'm suggesting is a way in which marginal groups are able to basically get their agenda into the newspaper. And once it's in the newspaper or the television in media, then that influences public opinion and public opinion in turn influences everything from voting to congressional behavior. And at the core of that, particularly in the context of the civil rights era, is, are you growing a kind of a civil rights coalition, which in that period was the democratic party, or are you potentially growing the competing coalition, which at that time was very focused on law and order.

I ended up collecting 275,000 headlines, front page headlines from seven different newspapers. So, things like the Wall Street Journal, the LA Times, the Chicago Tribune, the Washington Post. And in that 1960 to 1972 period, and then using a set of statistical tests that basically just said, if a protest happened today, does that predict a front page headline tomorrow? And what I found was that in fact, yes, if there was a particularly a protest in which the tactics were nonviolent, you had a headline that mentioned civil rights. And if you had a protest, there was protester initiated violence, the headline would mention riots, right. So that was showing in two ways protests are moving the media and the type of tactics employed are influencing the type of coverage.

William Howell:           

Yeah. So what you see is that these protests are getting coverage and that the nature of the protest influences the type of coverage. And then you go on and you document this relationship between protests and actual voting behavior and support for the Democratic party. What do those tests look like and what do you find?

Omar Wasow:              

There the core of the analysis is, we've got about 3000 counties in the continental US, and we have county level voting in this 1960 to 1972 period. And so I'm sort of saying, okay, well, let's imagine we knew what percentage of that county would vote democratic given prior history and prior things we know about the traits of that county and then kind of analogizing to a medical experiment or some other kind of randomized controlled trial say, well, let's imagine some counties are treated by protest and some counties are not. And what I find is that counties that were "treated" by a protest do vote differently. And in particular, the tactics employed by protesters are associated with different voting behavior. So counties that were proximate to nonviolent protest, and I should be clear, this is our civil rights protest, is African American led protest vote more liberally slightly, but statistically significantly, and counties that are exposed, "treated" by protest where there's protester initiated violence vote more conservatively. That is to say vote more for the Republican party.

William Howell:           

The bottom line here is that protests, depending upon whether or not they are violent or nonviolent, have very different effects on the willingness of white voters to back the democratic party, such that when they are nonviolent, you see increasing levels of support among whites for the democratic party. And when they are violent then voters, white voters, who are proximate to those protests are less likely to support the democratic party. And then you analogize too, a randomized field trial. But as you struggle within the paper and recognize protest, though, unlike pills and placebos are not randomly administered. But there are additional kinds of concerns, because what you want to say is that protest causally affected changes in media coverage and causally affected changes in public opinion and by extension voting behavior. In addition to doing a bunch of robustness checks, what are some of the moves that you make in order to back out or exploit variation in protest activity that might be plausibly thought of as being random?

Omar Wasow:              

Right. So the main kind of claim for causal inference here is to use rainfall as an approximation of random assignment, particularly in April 1968, after Martin Luther King, Jr. is assassinated. And this is true in a bunch of literature that basically large gatherings are sensitive to weather. And so you can think of the likelihood of an event happening is in part a function of whether it was raining that day or not. And an amazingly, as I've gone through all of this old newspaper for the content analysis and other kinds of things, I've come across multiple articles where it's like, march dampened by heavy rainfall. And so now when I present this work, I always showed that because somebody doesn't believe rainfall matters and I'm like, "No, look. Here it is documented in the sixties."

And the other kind of key detail here is that King's assassination leads to or contributes to 137 protests that escalate to protester initiated violence. So there's this dramatic spike in protest activity immediately following his assassination on April 4th. That is useful because one counter-argument moment might be that, oh, what you're picking up is some kind of crime effect or something, so crime is going up in the summer and violent protests are going up in the summer and you're conflating the two or something. But there's no historical spike in crime in April. So just to kind of tie it up, you can use a method to say, well, if rainfall influences protests and we think protests are influencing voting, let's use rainfall to predict protest activity.

I show where I run a bunch of tests. Rainfall is a very good predictor of this violent protester activity. And then we can use a kind of predicted protest activity to see if that predicts voting in November. And if we think rainfall in April is predictive of voting in November, the only obvious channel is through this violent protest activity. And so that's the heart of it. And so that's, to my mind, strongly suggestive that the only reasonable channel is through that week of heavy protest activity when 95% of violent protests follow King's assassination.

William Howell:           

And this is a period in which those protests that occurred in April of ‘68 tended overwhelmingly to be violent protests. Right?

Omar Wasow:              

Yeah, that's right.

William Howell:           

So, right. So the instrument predicts the incidence of protest, that violent protest per se, but it happens to be done at a time when the overwhelming majority of protests that occur were violent in orientation. And a core part of then your paper is to draw this distinction and to underscore the differential effects of nonviolent versus violent. And so here by looking at this one period, it's not speaking to that difference per se. It's just saying if there was going to be a protest in ‘68, in April of ‘68, it was likely to be violent in orientation and then when those that occurred, it tended to depress support for democratic candidates later in that year.

Omar Wasow:              

There's one subtle thing I would add, which is that actually all of the 137 used in that design are ones that were violent, that had protester initiated violence. There's something, which I'm actually curious to dig into a little more, but the data, there's very few nonviolent protests recorded in that period so much so that I can't even run the analysis. I think in the dynamics of collective action data, it was something like 15 nonviolent protests recorded in that period. So that's a little hard to reconcile with what we might think following his assassination and what we see in the current moment, but they're almost a hundred percent violent protests. So it's not like a muddy signal about both nonviolent and violent. Here, it's rainfall predicting violent protest, almost exclusively.

I think one of the things I'm most fond of in this paper is the kind of non-model analyses, where it's plots of reasonably raw data. If we just look at protest activity and public opinion, they co-vary very strikingly. And in particular, the period in which nonviolent protest is active, which is in, let's call it 1960 to 1966. What we see is that there are, when the March on Washington happens, there's this ... 250,000 people. The New York Times reports the greatest call for a redress of grievances the nation has ever seen. When they're these major events like that, the civil rights becomes the most important problem in America according to public opinion polls. Following Selma, Bloody Sunday, this brutal crackdown by white supremacist police officers and vigilantes, we see this just incredible spike in concern for public opinion around civil rights, Voting Rights Act passes in five months. So you can ignore all the models, look at that raw data.

And it's very hard to not say, okay, there's a Selma effect in ‘65. And then when you look in the later period, roughly ‘65 to, call it ‘69 or ‘70, there's a similar pattern, which is as protest activity that includes violence is spiking, we see changes in public opinion. And I think it's really important to underscore at least two things. So in one model of what's going on, well, America is a deeply racist country and there's this backlash to the civil rights and that is contributing to the rising taste for law and order. And I don't broadly disagree with that. I think that is broadly true, but it doesn't explain all sorts of important variation that we observed in the data.

So for example, what we see in the public opinion poll data is that it's, what I thought was noisy data, is that it's going up in the summer and down in the winter and up in the summer and down in the winter. People are more concerned about crime and riots in the summer, and they're not concerned about them in the winter. It has a seasonal structure. And if you are a story about concern about law and order is just a reaction to civil rights, you've got to explain why it's going down in the winter. That, to my mind, suggests that there's some other factor at play. And at least, and a very reasonable one, is the movement we're seeing on the ground where protests escalate to violence, they get covered in the media and that drives concern about law and order.

Again, this is me saying, do we observe associations between attitudes about violent protest and a shift out of the democratic coalition to the republican coalition. Given that in all of these different kinds of relatively model free assessments, we see this pattern of a real sensitivity to on the ground disorder that suggests to me, even if you have some questions about the specification of these models, you've got to explain this other variation and that's not obviously explained by any of the kind of slow or moving trends like racism or the economy or something else.

William Howell:           

Okay. So the best evidence that you have for identifying the relationship between protests and subsequent voting behavior and allowing us to interpret it as a causal effect comes from the protest that you observed in ‘68, wherein you leverage, as you've discussed, the random incidents of rainfall in order to protect subsequent protest activity. Nearly all of the protest activity that we observed in the aftermath of King's assassination in ‘68 is protest activity that turned violent, and that you set aside the handful of cases that turned nonviolent.

A concern that I have is that while this allows us to kind of, or gives us some leverage to pin down the relationship between protest activity and voting behavior, it doesn't speak as clearly or as directly to the distinction between violent and nonviolent, precisely because all of the behavior that we observed in 68 is violent in orientation, and it's not too much of a stretch of the imagination, it seems to me, to think that well, had those protests that we observed in ‘68 turned out to be nonviolent, that nonetheless, because the politics of ‘68, the mood of ‘68, the cultural moment of ‘68 was very different from what we observed between ‘62 and ‘66, that nonetheless, it too might have evoked a kind of law and order response.

The data don't allow us to speak directly to this distinction that counterfactual I have in mind is not the King is not assassinated. It's that in the aftermath of King's assassination, we are widespread nonviolent protests. And that might, nonetheless, induce a very different kinds of effects than we observed on the effects of protest that you have in the earlier period, primarily from ‘62 to ‘66, what do you think?

Omar Wasow:              

Understood. And I actually, given another a hundred words in the paper, I mean, there was a version of the paper that said there are two plausible counterfactuals here. The one you're articulating there, King is assassinated, but the protests remained peaceful, or King is not assassinated and the protests don't exist. It is possible that mass nonviolent protests could have produced some other outcome. But I think that's where it's also useful to look at the other evidence in the paper. So when we see a mass nonviolent protest effect on public opinion between 1960 and 19 sort of roughly 66, the big shifts in increases in public concern for civil rights. So the story you're proposing would have to not just assume that there's some backlash to nonviolent protest, but that somehow there's a very different dynamic around King's assassination, which is itself a pretty sympathetic moment, that that's producing some kind of very different dynamic than we see you say around the March on Washington, or we see say around Selma.

And so it seems deeply implausible to me that King's assassination would produce ... I mean, what we observed repeatedly in this period is that state or vigilante violence against peaceful protests produces increasing sympathy for the cause of civil rights. And so the claim or that counterfactual you're proposing is that vigilante violence against this kind of icon of peaceful protest would somehow produce more sympathy for white supremacy. And it is a plausible case, but it's not what we see in the data in the protest on public opinion data. And so to my mind, the plausible counterfactual of, if the protest were nonviolent, is that actually, it's exactly the reverse, that if I had the data, my expectation is that, or if we could somehow reduce that test, had these protests not been violent. My expectation is not that we would have seen this increasing support for the republican party and the law and order coalition, but rather that would have been consistent with ‘64, would have increased support for the civil rights coalition. So you're right, we can't fully disambiguate it with these data, but I don't think there's other evidence that supports the underlying hypothesis. And so that, to me, is the kind of best case against what you're suggesting.

William Howell:           

Yeah. So in the paper, and over the course of this interview, you've been careful to think about how the relevance of nonviolence versus violent tactics are contextually defined. It's not for the ages, all people, no matter what it's guaranteed, if you behave one way versus another way, it necessarily follows that it elicits a certain kind of response. And so, in a rough way, the lingering concern I have is that the context of ‘68 looks very different than the context of ‘62, ‘63, ‘64. And in that sense, the greater incidents of nonviolent protests that we observed in ‘62, ‘63, ‘64, had they persist it to 68, wouldn't necessarily have guaranteed the same kinds of effects.

Now, what you're saying is that in the aftermath of King's assassination, your suggestion is that they may well have, that that could have persisted. It was a, as you put it, a sympathetic moment, but it could also have been that a set of people say, well, I'm just tired of this general unrest. I'm tired of all the attention paid to this issue and I'm fed up. And even in the face of non-violent protests, the ways that whites who live near these protests or amidst these protests, respond may have looked very different in ‘68 than it had in ‘62 to ‘64.

Omar Wasow:              

So now there are two assumptions that you're making. So one is, there's some significant shift in white attitudes towards nonviolent protest and nonviolent protest and a shift in white attitudes might have produced some other outcome. And so I think I'm willing now to kind of really dig in and say, no, you're making a useful point about thinking about context, but it's like, Kennedy's assassinated before ‘64. I mean, there other kinds of profound disorder going on that are not associated with a counter mobilization to civil rights. In fact, what we observe in ‘64 is a landslide for civil rights. And so I just think that the kind of the fundamental dynamic here is the human capacity for sympathy or fear. And I think those dynamics are actually not so time specific and that those underlying dynamics are actually much more fundamental. Even if I can see the point there, there is the possibility that nonviolent protest could have produced not only a different dynamic and whites could have ... They're a bunch of potential counterfactuals that I can't rule out, but they seem inconsistent with a broad body of evidence, both in my paper and in other work.

William Howell:           

Great. One point that I think is simply worth articulating, which has to do with changes over the sixties. It's something that we've been talking about at some length here, that there are lots of stuff happening in ‘62 and ‘64 and ‘66 and ‘68. And that may bear upon white public opinion and white voting behavior and on media coverage, that's outside of the incidents of nonviolent versus violent protests. When you add to the models, a host of year fixed effects and the way that you have throughout, you have county fixed effects because they're whole bunch of characteristics of counting that you can't account for exhaustively. When you add in the year fixed effects, the point estimates on protest activity, they remain discernible and they retain their signs as a function of whether or not they are violent or nonviolent, but they shrink markedly in magnitude. So when we think about the magnitude of these effects, that then if we think that there ought to be year fixed effects in there, it would suggest that they're somewhat smaller.

Omar Wasow:              

I think my core question is valence. Do we observe the sign being different? And the sign is different in all of the estimates for nonviolent versus violent. And then the question you asked was about just kind of a magnitude, for example, the ‘68 election. Even if I say, okay, let's take the most conservative estimate and let's say, is that meaningful to move an election? It's meaningful to move an election. And so I think the actual magnitude only matters in as much as there's like a substantive oomph to it. And so if it moved from something substantively meaningful to something substantively not meaningful, then I think that becomes a much more important critique. Let me say this one other way, which is that, the core contribution this paper makes relative to work in American political science is that there's very little work in American politics that says, what is the differential effect of nonviolence versus violence, and so if that's the heart of our question, the magnitude of the coefficient is not the core of it. The core of it is, do the signs change and if the signs change that tells us that there may be differential effects.

William Howell:           

This piece of research that you've put out has, well, it's attracted a lot of attention. What kind of reception are you getting? How would you characterize the reception you're getting? And what would you say to those who offer one critique or another?

Omar Wasow:              

I think there's kind of three or four different ways in which people have responded. One is that there's just, people are grateful to have some model for how to think about what's going on. And so there's some folks who were like, Oh, somebody tweeted this figure that was like, here's the protester activity. Here's the state repression response. And you know, that like giving people a kind of mental model for the world is like a deeply satisfying thing, if it holds. Second kind of response has been a kind of right of center response, which tends to focus on the, oh look, there are better tactics and worst tactics. And the kind of right of center response tends to engage in critiquing protesters for not using good protest tactics.

And what's hard about that is that that's not my project in the sense that I'm trying to ask a question, which is, how can marginalized groups, whether you're Act Up, Adapted, a disability rights group, or groups operating against incredible odds in other parts of the world, how do you overcome those odds, and let's look at the sixties as a way to understand that. And this kind of right of center critique draws on something that King said at one point in Letter From a Birmingham Jail, which is that he complains that pastors are critiquing the protest tactics. But he says, I regret to say, you seem more concerned about our protest tactics, then the underlying injustice that motivates these protests.

And that's something I'm trying to like sit with also. I don't want my work to be just a, hey you protesters, just do it right, if we're not thinking about the underlying injustice as much, if not more. He actually has a line, the following sentence in King's letter is, we would not want to engage in the kind of superficial social analysis that would focus on effects without causes. And so it's like, okay, right. I don't want to be a superficial social analyst, like, what is the underlying cause here, the underlying cause here has profound injustice. And so—

William Howell:           

But before we get to the third, which I want to hear. On this one, are people who are right of center misinterpreting your work in so far as you're drawing this clear distinction between nonviolent and violent and that one has salutary effects whereas the other one elicits a political response, which is damaging the short term and quite possibly in the longer term as well. And in so far as protestors have agency in shaping whether or not a protest is going to be violent or nonviolent in orientation, then your work is offering counsel to how they ought to respond to these deep injustices, no?

Omar Wasow:              

So let me be fair to someone like Ross Douthat who summarized my work in the New York Times. I don't think he misrepresented the work. I think he did a reasonable job of doing justice to the work. I think the question is, how often is he writing about state violence against black people. How often is this at the top of his list of items to write about? And if the only time he writes a prominent piece about black protests is when he gets to sort of tell black protesters to do it right, that suggests a kind of superficial social analysis. That critique could be leveled at me, but I've published two papers and this took 15 years. And so it's like, I don't, there's not, in the book, can I try to do more on the left of center injustice, absolutely.

And in my own kind of media, kind of engagements with the media, can I do that? Absolutely. And so I think it's a critique I take for myself as well. And I'm learning as this goes, like I didn't fully anticipate all the ways in which the work would have a public life. I was on a radio show in San Francisco with two very prominent black academics and we were talking about the, this was still early in the wave of protests. So there was a mix of big nonviolent, some violent, and of the three calls that came into this show, two were about tactics. And I sort of found myself thinking, look, I care a lot about tactics, but if we're talking about tactics, we're not talking about injustice. And that to me is a way which I don't want the work to kind of push out of the spotlight, that the kind of core problems that motivate our attention to protest. So even as somebody who's paying a lot of attention to tactics, I don't think tactics are the most important thing.

Then on the sort of maybe more kind of left of center, there's been a critique, which, echoing this point about, you're not focusing enough on injustice. That sort of says it's even bigger than that. It's sort of saying, it's not just that there's like an injustice against George Floyd, but to talk about consequences of black activists choosing different strategies is to take white supremacy out of the core of our kind of analytical lens and that it is really important to focus our attention on the media that had engaged in kind of biased reporting or the white moderates, who rather than responding with empathy, responded with fear. And that they have agency and that they are part of a backlash. And so, where are they in the story and in a sort of more general form, there's just kind of this totalizing or immovable prejudice that is like, of course there was going to be a Nixon election or of course there was going to be a law and order victory, because that is the history of America of like white domination of African Americans.

And there I take the critique of it's important, particularly in the contemporary moment to hold the media and the larger white public to account. Like, you can look at a burning police station and say, I want repression. You can also look at that and say, wow, something's really wrong in our society and what would it look like to think more deeply about that. So I think it's useful for me to sit with that critique, but on this question of an immovable prejudice, a sort of white domination that constrains black activists profoundly, I concede a lot of that, but of course, the whole point of what folks like Bayard Rustin and Angela Davis and Stokely Carmichael, and Ella Baker we're trying to figure out is, given an intransient white public, given a racist society, what can we do?

And in that context to say that there is this sort of totalizing domination eliminates their agency. And so very much at the heart of this project is to say, even under terrible conditions, whether you're somebody who's a political prisoner in Ireland, you can engage in a hunger strike. And that changes the political dynamics or a hunger strike in Guantanamo in the US. People under exceedingly oppressive conditions are able to assert agency. And then the question is, well, in that context, what's going to work. And there, I think my evidence one, again, big shifts in public opinion, suggest that even in the context of a fairly bigoted society and a fairly systematically biased media nonviolent protests were able to get coverage that said, this is like a legitimate claim for rights.

And conversely protests that escalated to violence produced media that tended to emphasize crime and riots, and that the public opinion moved on both of those. And the fact that there's so much variation cannot be explained by a story that has a kind of static racism or immovable prejudice. So that variation suggests to me, there's both the opportunity for black activists to make a difference, and that there's a swingier white media and public that are responsive to those cues. And so I think that's a story I need to tell better, but I think that's also very much at the heart of the paper.

William Howell:           

Thanks for listening to Not Another Politics Podcast.

Wioletta Dziuda:          

Our show is a podcast from the Harris School of Public Policy, and it's produced by Matt Holdup. Thanks for listening.