Day after day, protests have arisen in cities across America. The outrage was sparked by video of a white police officer kneeling on the neck of George Floyd, even as the 46-year-old black man begged for breath. Floyd was arrested May 25 for allegedly trying to buy cigarettes with a counterfeit $20 bill and died after being pinned to the ground for eight minutes and 46 seconds by the Minneapolis officer’s knee.
That spark easily found both fresh and long-simmering fuel. Among recent events, white men killed a black jogger, a white woman called the police on a black birder in New York’s Central Park (SN: 6/4/20) and the pandemic has taken a disproportionate toll on black people (SN: 4/10/20). Those events underscore centuries of racism that has limited black people’s access to housing, health services, education and jobs.
The anger, anguish and calls for racial justice that first boiled over in Minneapolis quickly spread coast to coast. While many protests have been peaceful, some have turned violent — instigated sometimes by looters, sometimes by individuals among the protesters and sometimes by law enforcement using force to disperse crowds.
Whether these protests will help dismantle systemic racial inequities in the United States remains to be seen. But some lessons and parallels can be drawn from the civil rights protests in the 1960s, says Princeton University political scientist Omar Wasow. His research shows that the media covered civil rights protests in the ’60s in different ways depending on whether protestors were peaceful or violent. And that coverage shaped public opinion and behavior at the ballot box.
When protestors remained peaceful, particularly in the face of aggression and violence, the resulting images shocked a complacent nation into action. But when the protestors themselves turned violent, even in self-defense, the media message shifted from a framing around civil rights to one around the need for control, Wasow finds. For example, Martin Luther King Jr.’s assassination on April 4, 1968, triggered a week of violent protests around the country. Those protests helped Republican candidate Richard Nixon, campaigning on law and order, prevail over Democratic candidate Hubert Humphrey, lead author of the 1964 Civil Rights Act, in the November presidential election, Wasow reports May 21 in the American Political Science Review.
“An ‘eye for an eye’ in response to violent repression may be moral, but this research suggests it may not be strategic,” he writes.
Science News talked with Wasow about his findings and how they apply to the current protests. The interview has been edited for length and clarity.
SN: What was your big question about the protests of the 1960s?
Wasow: A large body of political science finds marginal groups have no influence. I wanted to see if protesters actually influence politics. I found that protests could be very influential through their effect on media. If you think about it, almost nobody directly observes a protest. The way a protest reaches us is through the news. That coverage varied if the protest was violent or nonviolent. A nonviolent protest [that made the news] predicted a front-page headline the next day that mentioned civil rights. When protests escalated to violence, that predicted a front-page headline with a word like “riot.”
So a protest influenced media coverage and that coverage influenced public opinion, or how people responded to survey questions such as: What is the most important problem in America today?
As protest activity mobilizes, the percentage of Americans who say civil rights is the most important problem in America increases. When protests turn violent, public opinion shifts to concerns about crime and riots.
SN: How did you evaluate the link between violent versus nonviolent protest and later voting?
Wasow: In the early part of the 1960s, most civil rights protests used nonviolent tactics even when met with police violence. Those events were followed [later in the decade] by a wave of protests that often escalated to protester-initiated violence, peaking when Martin Luther King Jr. was assassinated in April 1968.[That coincides with] a big shift in public opinion. In the early part of the 1960s, survey respondents say the most important problem in America is civil rights. But in the late 1960s, we see a spike in concern for crime and riots. That’s a puzzle. Are those shifts at all associated with protest activities on the ground?
In political science, voting is a really important outcome, so that’s what I looked at. The basic framework is we have a county and it is “treated” or “not treated” by a protest. A county is [considered] treated if there is a protest within 100 miles and within two years of an election. I looked at two conditions. Under one, I compared counties treated with nonviolent protests to counties that experienced no nonviolent protests. In the other, I compared counties treated to violent protests to those with no violent protests. I wanted to know: Do the treated counties vote differently than the not treated counties?
In the primary models, I estimate the effect of protests on voting across the 1964, 1968 and 1972 presidential elections. These models compare each county to itself over time. In addition, to try and make better “apples to apples” comparisons, I also used a method called matching that only compares counties with and without protests that are very similar on variables such as percent black or percent foreign born. Another thing I looked at are counties that are 90 percent white. I find that counties close to nonviolent protest between 1960 and 1972 see increased Democratic vote share. Conversely, counties close to violent protest vote more for the Republican Party. That’s likely because, following the 1964 Civil Rights Act, Democrats tend to be seen as the party of civil rights and Republicans as the party of law and order.
SN: Could something besides protests have influenced those election results?
Wasow: If we had godlike powers, we could randomly assign which counties to treat with violent or nonviolent protests and then see what happens in the November vote. Obviously we can’t do that, but we can look for possible natural experiments. Martin Luther King Jr.’s assassination in April 1968 sparked many violent protests across the country, so I could compare what happens when counties did or did not experience violent protests.
What I do here, which builds on work by some economists, is use rainfall as something that might predict protest activity. There’s a lot of work that shows protest activity is sensitive to weather. More rainfall equals less likelihood of protest activity. Less rainfall equals more likelihood of protest activity.
Counties that experienced less rainfall were much more likely than those with more rainfall to experience protests following Martin Luther King Jr.’s assassination. So there’s two steps in the process: rainfall’s effect on protest and protests’ effect on voting. If rainfall can predict voting, the only plausible path is through protest activity. There’s not another plausible explanation.
I also conducted a placebo test. That’s because rainfall is associated with geography, and geography is also associated with voting behavior. So I asked: Does rainfall before Martin Luther King Jr. is assassinated influence voting in November? That’s a placebo, like a sugar pill, because we shouldn’t expect it to have any effect. I find that rainfall before King’s assassination does not predict voting in November. I also look in the last two weeks of that April after most violent protests ended. And again, the rainfall in that period doesn’t predict voting in November. It’s not just rainfall in April. It’s rainfall in one week of April that predicts voting in November. That allows me to say it’s not just geography, it’s not just the South is rainier. I make the case that experiencing a violent protest caused people to vote for order in November.
SN: What role do journalists play in shaping the narrative around protests?
Wasow: I scanned thousands of newspaper pages and created a corpus of articles about protests and then asked: If a protest had been categorized as nonviolent, what sorts of words are commonly used? And similarly, if a protest had been categorized as violent, what are some of the most common words?
Nonviolent protests seem to be covered as if they were traditional attempts to redress grievances, seek rights. So the words we see are “civil rights,” “demonstration,” “march,” words that suggest this is a legitimate claim for rights. When there were violent protests, the words that were commonly used were things like “riot” and police.”
The key idea here is people are protesting because they’re angry about some injustice, but the kinds of tactics employed will focus the media’s attention on that injustice or, in some cases, shift the focus away from that injustice. That’s why tactics matter so much. The approach tells the media what to pay attention to and by telling the media what to pay attention to, protestors are telling the country what to pay attention to. This finding was revelatory for me. I didn’t start out thinking this was a study of media.
SN: Do those findings apply to today, given the media has changed a lot since the ’60s?
Wasow: Media are much more fragmented than they were in the 1960s. Everybody has their own unique media feed. That’s going to mean that following the news may be a more siloed experience, where some people are very focused on activist violence while other people will be much more focused on police violence. Depending on which channel you watch, depending on who your friends are on social media, you may be getting very different narratives.
To be clear, that’s not entirely different from the ’60s. Most of the Southern media was pro-segregation, and media outside of the South tolerated Jim Crow and was not interested in the concerns of black people. The idea that there might be two different visions of what’s happening was not so unlike a black press that covered the concerns of black people and a white press that was indifferent or even hostile to those concerns.
SN: What kind of impact could violent protests have on the 2020 election?
Wasow: It’s hard for people to appreciate that there’s a set of voters who are open to concerns about racial equality, but it’s not their top priority. They also are very concerned about order. Think of somebody who might be an Obama-turned-Trump voter. In the ’60s, there were people who supported the Democratic candidate after the passage of the Civil Rights Act [in July 1964]. But they joined the law and order coalition after the period between 1964 and 1968 when there was a lot more violent protest. These voters are influential because they move between parties and because they are in swing states.
On the one hand, some whites today have become much more concerned about racial equality and center their conversation on the underlying injustice against George Floyd. But it might also be possible that more whites move toward the law and order coalition and support Trump. I think it’s too early to tell.
SN: Recently, newspapers have run images of the police taking a knee in solidarity with protestors. How do you think such imagery affects media attention?
Wasow: My model suggests that peaceful events don’t usually get as much press because they are less dramatic. But I had to simplify the model [for this study]. A slightly more complicated version of the model is that violence is just one way of creating drama. Seeing police behave in a counter-stereotypic way is dramatic. And consistent with my theory, nonviolent protest can be effective if it’s able to do something that captures the attention of the media. Violence is one way of creating spectacle, but it’s not the only way.
SN: Critics have said your study puts too much responsibility on protestors. What do you think?
Wasow: What’s important is the causal story I’m trying to tell. A story that says “this is all about white moderates” deprives the protesters of their agency. I want to begin the story with, despite overwhelming odds, this subordinate group at the margins of society has power. And the question is: How can they use that power to advance their interests most effectively?
SN: What is your advice to today’s protesters?
Wasow: There are two kinds of deep narratives in which we talk about protests: a rights-, or justice-, framed story or a crime story. That was true in the ’60s and that’s true now.
In the 1960s, leaders of the civil rights movement were incredibly focused on how to get their message out to the whole country and used protest as a means to gain influence. What they found was that large, peaceful demonstrations without conflict didn’t interest the press. A New York Times reporter covering a march in Mississippi said something along the lines of “no blood, no guts, no glory.” The key idea is that nonviolence was often not enough to generate the kind of attention that was necessary to create a national crisis.
To create a sense of crisis, it became necessary to engage in this very strategic kind of nonviolent protest, which was for protestors to become the object of state violence. Activists picked places like Birmingham and Selma because there were these police chiefs with a hair trigger for violence who would engage in brutal repression in front of cameras. That would shock the conscience of these otherwise indifferent or even hostile actors outside the South. It changed public opinion.
If you’re an activist on the ground thinking about and angry about this injustice against George Floyd and a long history of police violence in this country against African Americans, if you want to put that at the center of the national conversation, it’s important to be thinking about this: Is what we are doing on the ground elevating the justice frame or elevating the riots frame?