This article is a collaboration with the Monkey Cage, an independent political science blog hosted by The Washington Post.

While the brutality and injustice of George Floyd’s death in police custody have been condemned by liberal and conservative politicians alike, two very different narratives emerged as the protests grew. As of this week, a majority of Americans — 74 percent, according to a Washington Post-Schar School poll — generally supported the protesters’ efforts to draw attention to police violence against African Americans. Still, as night fell in the early days of the protests, and stores were ransacked and buildings went up in flames, a second narrative took hold: These protesters were less interested in conveying a political message than in destroying property and breaking the law.

After the violent uprising, some commentators — notably Ross Douthat in the New York Times — drew on research I had just published, in the American Political Science Review, to draw comparisons to 1968. In the presidential election that year, after protests and violence erupted in dozens of American cities, Richard M. Nixon, who campaigned on a “law and order” platform, beat Hubert Humphrey, lead author of the 1964 Civil Rights Act. And indeed, my study found that protester-initiated violence in 1968 probably tipped the election to Nixon, costing Humphrey some 763,000 votes nationally.

Over the past week, however, as the protests grew larger and more organized — as they remained largely peaceful and as hundreds of videos of police violence against demonstrators circulated — it became clearer that 1964 was probably the stronger historical analogue. In that year, nonviolent civil rights protests led not only to the passage of the Civil Rights Act but contributed to the landslide reelection of President Lyndon B. Johnson.

My research project looked at black-led demonstrations between 1960 and 1972 in an effort to understand the political consequences of protest. For much of the 1950s and early 1960s, the Southern press was largely pro-segregation and the national mainstream press ignored the concerns of African Americans. Why and how did that change? Using a data set of thousands of demonstrations in which activists and police used nonviolent and violent tactics, I found that protests — as filtered through coverage in mainstream newspapers — powerfully shaped the issues voters thought were most important; as a consequence, they shaped how the electorate voted.

The effects of protesters’ actions can be seen in the issues voters considered most salient during that era. Civil rights was cited as the most important issue by only 5 percent of the electorate in December 1962, for example, but that figure spiked to 48 percent in mid-1963, as the March on Washington captivated television audiences around the country. Though concern about civil rights as the most important problem faded almost as quickly, it rose sharply again after the “Bloody Sunday” march in Selma, Ala., in March 1965 — during which state troopers bludgeoned and tear-gassed protesters. Images of the clash contributed to another nationwide surge in concern about “civil rights,” and the Voting Rights Act was passed five months later.

On the other hand, the view that social control was the most important issue languished in the single digits through the mid-1960s but rose to 41 percent in August 1967 — during the “long, hot summer” when a five-day riot in Detroit led Michigan Gov. George W. Romney to deploy the state National Guard. (Interest in that issue subsided quickly, too.)

To shed light on the causal link between protests and public opinion, I looked not only at the data set on protests and at polls but also at some 275,000 protest- or riot-related headlines from seven major newspapers (including The Post, the Los Angeles Times and the Wall Street Journal), from 1960 to 1972.

First, I found a protest on one day was highly predictive of a front-page headline the next day: Protests were indisputably making the news. And the types of tactics employed by activists and police shaped the type of coverage received. In the early half of the 1960s, when protesters relied primarily on nonviolent tactics — even when the police met them with violence — the media generally focused on concerns of the civil rights movement, analysis of headline content showed. Words like “church,” “march” and “demonstration” outweighed words like “fire” and “shot.” Civil rights leaders in the ’60s were keenly aware of the media’s role in amplifying the narratives that would help or hurt their cause and routinely staged protests with the intent of making themselves objects of violence.

When the police cracked down with dogs and clubs, those actions ultimately served the protesters’ aims by visually “dramatizing injustice” — just as images of the police brutally used to clear the area near Lafayette Square in Washington, D.C., last week appeared to have done. By putting their bodies and lives on the line, activists were able to “shock the conscience of the nation,” an oft-used phrase. Whites outside the South who were otherwise indifferent to the persecution of African Americans started taking notice of the viciousness of Jim Crow. The rise in the belief that civil rights was the national issue tracked that news coverage.

In counties near nonviolent protests, I found that support for the Democratic candidate — which, after passage of the 1964 Civil Rights Act, represented the civil rights coalition — increased in the 1964, 1968 and 1972 presidential elections.

From the mid- to late 1960s, however, protesters engaged in more aggressive resistance to racial inequality and police violence. In cases with significant protester-initiated violence, I found that news coverage tended to frame these protests in terms of crime and riots — whether the state used violence against protesters or not. From 1964 to 1972, I found that counties that were near protester-initiated violence tended to shift toward Republican presidential candidates. Depending on the year, this shift ranged from roughly two to eight percentage points. Drawing on this data, I ran simulations of the 1968 presidential election under the counterfactual scenario that Martin Luther King Jr. had not been assassinated and 137 demonstrations with protester violence had not occurred. I found that Humphrey won about two-thirds of the time, taking five additional states in the Midwest and Mid-Atlantic to beat Nixon.

This past week — in terms of protests and violence — we’ve seen echoes of the 1964 and 1968 election cycles. The initial wave of protest activity looked more like 1968, grief and anger spilling over into rage that filled TV screens with images of vandalism and a police station in Minneapolis going up in flames. In the days since, however, the protests have been massive, global and overwhelmingly peaceful. As in 1964, the violence on screens was coming from the police. “In protests against police brutality, videos capture more alleged police brutality,” as one apt headline put it. Greg Doucette, a lawyer who has collected video clips of more than 350 incidents of police violence, was blunter: “Cops are rioting.”

In the 1960s, nonviolent civil disobedience required organization, training and leadership. It asked civil rights activists to abnegate some very basic human instincts of self-preservation and self-defense. The consequence could be serious injury, even death. Many protesters last week displayed similar discipline, and the resulting news coverage appears to be winning over the public. The Washington Post-Schar School poll found that 69 percent of Americans say the Floyd killing represents a broader societal problem, not an isolated incident.

One challenge for social movements today is that the media in the United States are a lot more fragmented. But news bubbles are hardly new. As journalists Gene Roberts and Hank Klibanoff noted in the Pulitzer Prize-winning book “The Race Beat,” before the civil rights movement, the mainstream media largely ignored black Americans and discrimination; to learn about those subjects, one had to turn to black newspapers. Contemporary cleavages of polarized media fall more along lines of partisanship than segregation, but the challenge for protesters is the same: how to get issues of police violence and discrimination onto the national political agenda, when they were previously covered sporadically, if at all. The rise of the Web, social media and partisan cable channels complicates that challenge of creating a common understanding of the underlying problem. Yet the recent polling suggests the protesters broke through.

For a growing international movement trying to draw attention to the long history of racist and brutal policing, nonviolence in the face of police repression is an exceedingly difficult strategy to sustain. Evidence from the 1960s, however — and perhaps this month, too — suggests using such tactics to generate media coverage of a pressing social problem can be a powerful tool for building a coalition for social change.