What’s Driving Political Violence in America?

Hate crimes are rising, and so is support for political violence. New research explores why—and what we can do to stop it.

By Zaid Jilani, Jeremy Adam Smith



Last month, Americans saw a bloody spike in politically motivated violence. An anti-Semite killed eleven people at the Tree of Life synagogue in Pittsburgh. A vocal racist shot down two random African Americans at a Kentucky supermarket, after failing to break into a black church. A man in Florida was caught mailing pipe bombs to politicians, journalists, and celebrity critics of the president.

While political terrorism is still quite rare in the United States, there’s no question that these seemingly isolated acts are part of a larger pattern. Against a backdrop of rising political and social polarization, the Anti-Defamation League has reported a 57 percent increase in anti-Semitic incidents since the 2016 election. Studies of FBI data show that hate crimes—violence against people or property because of group membership—have risen during the past few years. In California alone, hate crimes have jumped 44 percent since 2015.

How many Americans endorse these acts? Earlier this year, political scientists Lilliana Mason and Nathan Kalmoe presented a paper at the American Political Science Association’s annual meeting, titled “Lethal Mass Partisanship.” With data from two different national surveys, they found that 24 percent of Republicans and 17 percent of Democrats believe that it is occasionally acceptable to send threatening messages to public officials. Fifteen percent of Republicans and 20 percent of Democrats agree that the country would be better if large numbers of opposing partisans in the public today “just died,” which the authors call a “shockingly brutal sentiment.” Nine percent of both Democrats and Republicans agree that violence would be acceptable if their opponents won the 2020 presidential election.

While only small percentages in both parties outright endorse violence, these still represent tens of millions of people. So, what drives some people to go beyond partisan disagreement and actually embrace political violence? Do our violent words lead to violent actions? What can we do to prevent more violence? While the association between political rhetoric and violence is not perfectly understood, researchers are starting to map the social and psychological forces that seem to be driving pugnaciousness between groups of Americans. We know that political violence can spread, like an illness. Here are five factors that make infection more likely.

1. Aggression
Mason and Kalmoe identified three degrees of violent attitudes: Some people hold views that rationalize harm towards political opponents, others express happiness toward their deaths, and still others endorse outright violence against them. Put together, these are the components of lethal partisanship.

Mason and Kalmoe then looked into various factors to see what might be contributing to these three attitudes. Demographic categories such as gender, age, and education didn’t matter. Contrary to what some liberals might expect, in these surveys positive feelings towards President Trump didn’t predict more violent attitudes.

So, what did? By far the biggest predictor of lethal partisanship across the board was having aggressiveness as a personality trait. This isn’t surprising, of course—aggression and violence go hand in hand. But a deeper look at aggression reveals how it fits together with other traits and shapes human behavior. Aggression all by itself is not good or bad; any of us can become aggressive when we face a direct threat. But aggression can go too far when inner and outer restraints are absent.

In neurological studies, more aggressive people tend to show less activation of the default mode brain network, which is associated with empathy and emotion regulation, which in turn helps suppress aggressive impulses. As psychologist Scott Barry Kaufman notes in Scientific American, aggressive people are more likely to retaliate when treated unfairly by others, which is not necessarily a bad thing (“although they tend to care much less about whether others are treated unfairly”).

However, aggression also shapes political outcomes. “Politicians who are more antagonistic get more media attention and are more often elected than more agreeable politicians,” he writes. “In the general population, antagonistic people are more likely to distrust politics in general, to believe in conspiracy theories, and to support secessionist movements.” In a series of experiments published in 2014, Kalmoe found “that exposure to mildly violent political metaphors such as ‘fighting for our future’ increased general support for political violence among people with aggressive personalities.”

Aggression might be more a part of some personalities than others, but it’s also the case that some social situations are more likely to trigger aggression than others. So, what might foster aggression in public life? To answer that question, we must move beyond personality traits to look at group membership and what happens in our groups.

2. Intense partisan identity
After aggressiveness, Mason and Kalmoe found that “partisan identity strength”—how much being Democrat or Republican is part of who they are—is the most important factor in endorsing violence.

There are many studies—mostly from political science and sociology—showing that more Americans are using their political party affiliation as a source of meaning and social identity, with these identities linked to differences in “leisure activities, consumption, aesthetic taste, and personal morality,” as Daniel DellaPosta and colleagues write in their 2015 paper, “Why Do Liberals Drinks Lattes?

Worse, the Republican Party has become whiter in recent decades, while the Democratic Party has become more racially diverse—which could be intensifying party antagonism. A recent study of survey data by political scientist Diana Mutz found that nothing predicted support for Donald Trump more than a feeling of threatened status among white Christians—an insight ratified by several studies from Robb Willer at Stanford University and the Public Religion Research Institute.

    “The parties are significantly more racially distinct than they were a few decades ago”
―Dr. Lilliana Mason

“Indeed, the parties are significantly more racially distinct than they were a few decades ago,” says Mason. The implication is that Americans aren’t just disagreeing about the issues when they talk politics—they may feel a sense of tribal, existential threat when someone disagrees with the positions of their political party. “The relationship between ethnocentrism and violence is abundantly clear cross-nationally and historically,” write Mason and Kalmoe in their paper. It could be that the problem isn’t partisanship, exactly—instead, the truly dangerous ingredient could be the racialization of party affiliation.

“All of the research to date was pointing in this direction,” adds Mason in an interview. “But we have a long tradition of treating partisanship like a largely benevolent force. It makes sense that as an identity grows stronger, and conflict intensifies, people will begin to approve of violence.”

3. Anger, contempt, and disgust
While Mason and Kalmoe’s study gives us some sense of how common the tendency to accept political violence is—and some of the personality traits and belief structures that may be associated with it—a 2015 study points us in the direction of the emotions involved. In “The Role of Intergroup Emotions in Political Violence,” San Francisco State University researchers David Matsumoto and Hyisung C. Hwang and the University at Buffalo’s Mark G. Frank tried to figure out which emotions can drive violence by a group against an outgroup.

They examined the emotional tone of major political speeches that occurred prior to political events throughout history, looking at the emotions expressed in words, the judgments underlying the emotions, and the nonverbal expression of the emotions that could be seen in video form.

They also examined speeches made by “ideologically driven” leaders who despised opponent outgroups that resulted in violence, such as Hitler’s speeches; they also studied those that did not, like Gandhi’s Salt March and pro-Tibet protests at the Beijing Olympics in 2008.

They found that speeches which preceded violent events tended to express more anger, contempt, and disgust (ANCODI)—but not fear, happiness, sadness, or surprise. These negative emotions tended to target specific “outgroups”—Jews, in the case of Hitler’s speeches.

The “amount of anger, contempt, and disgust expressed nonverbally by violent-group leaders correlated significantly only when they referenced the opponent outgroup; they did not correlate when the violent-group leaders referenced something other than their opponent outgroup,” note the researchers. In the months before major non-violent events, anger, contempt, and disgust actually declined.

The researchers suggest that the ANCODI model may be a way of monitoring the expression of emotions by group leaders that could provide an early warning of looming violence. The researchers also suggest that interventions to prevent violence could be measured by how effective they are in reducing ANCODI emotions. One important caveat of the study the researchers concede is that they looked at group behavior. “We do not know if this phenomenon applies to individual acts of violence,” they write.

4. Moralization and moral convergence
In 2015, Baltimore witnessed mass protests after a man named Freddie Gray died in police custody. A small number of these protests turned violent.

Earlier this year, a team of five researchers searched the popular social media platform Twitter for tweets about the Baltimore protests. They wanted to investigate “moralizing” tweets—that is, tweets that viewed the protests as a moral issue rather than as a political disagreement. A moralizing tweet might, for example, refer to people as “disgusting” or “evil” or “traitorous.”

In fact, they did find a positive association between the number of moral tweets and the occurrence of violent protests (gauged with arrest data). “The days in which there were violent protests, we saw that there was a lot more moral language being used,” says study co-author and University of Southern California Ph.D. student Joe Hoover. “Which was consistent with the idea that morality and violence in these contexts might be linked.”

The team also ran an experiment using another prominent protest marred by violence: the far-right rally in Charlottesville, Virginia, in 2017. Respondents were asked to what extent they thought protesting against the far-right demonstrators was a moral issue; they were then asked how acceptable it was to use violence against these far-right activists. What they found is that people were more likely to embrace violence the more they saw it as a moral issue.

In an additional experiment, the participants were given the same prompts, but they were told either that the majority of Americans agreed with their view of the protest, or that few Americans agreed with their view. They found that “moralization predicted violence only when participants perceived that they shared their moralized attitudes with others.”

In other words, when it comes to violence, there’s validation and safety in numbers. The researchers dubbed this phenomenon “moral convergence,” when many people come together around a strong idea of what’s right and what’s wrong. The “risk of violent protest, in other words, may not be simply a function of moralization, but also the perception that others agree with one’s moral position, which can strongly be influenced by social media dynamics,” they write. 


Of course, our groups often do good, as well. Hoover concedes that moralization, moral convergence, and violence don’t always have to be linked. Some of the most disciplined nonviolent movements in American history have been highly moralistic. For instance, the Civil Rights Era Southern Nonviolent Coordinating Committee in the American South trained its activists to not respond violently even if attacked.

“We are thinking about these factors as risk factors,” says Hoover. “Certainly, there have been many protests where participants held a strong moral position and surely perceived moral convergence and there was no violence.” Moralization and moral convergence are like kindling, but “you still need a spark; you need something to actually light the fires.”

Hoover suggested that police could play a positive role in preventing protest violence by practicing de-escalation that makes clear they respect the protesters’ ability to be heard. As the research by Matsumoto, Hwang, and Frank suggests, leaders also have a responsibility to not fuel anger, contempt, and disgust.

5. Group leadership
When it comes to violence, leadership matters. There are many, many studies—starting with Stanley Milgram’s classic electric-shock experiments—which show that people are much more likely to inflict pain on others when an authority figure tells them to. When leaders engage in violent rhetoric, so do their followers; when they urge calm, people do calm down. Research has documented that words do have an impact on both beliefs and behaviors.

For example, a 2017 Polish study found “frequent and repetitive exposure to hate speech leads to desensitization to this form of verbal violence and subsequently to lower evaluations of the victims and greater distancing, thus increasing outgroup prejudice.” As part of the study, researchers surveyed participants on how frequently they encountered hate speech against refugees; they found that those who were more exposed to the speech were more prejudiced against the group and more accepting of restrictive immigration policy.

Taken together, these studies suggest that our political leadership—everyone from pundits on cable news to the President of the United States—would do well to avoid promoting the political tribalism that leads people to strongly identify with one group and demonize the other. They could also reduce the use of angry, contemptuous, and disgusted rhetoric to refer to political outgroups. And Americans everywhere could learn to rely less on the echo chambers of social media, where moral convergence and affirmation can fuel violence against people and property. Most importantly, we have to be able to stress civility and humanity towards the other side, even when it’s difficult.

Recently, Florida Republican Congressman Carlos Curbelo received a threatening tweet. “I will kill Carlos Curbelo,” wrote a 19-year-old constituent named Pierre Alejandro Verges-Castro.

“What Pierre did is very serious,” Curbelo told the Miami Herald. “We have two young daughters and like any family, we worry about our safety and security, especially in light of all the acts of violence we are seeing throughout our country.”

 Despite these fears, Curbelo chose to reach out to Verges-Castro and held a joint press conference with him where he publicly forgave him. Although the teenager has been arrested by the FBI and may face some form of penalty, the congressman stressed that he doesn’t want the incident to tar him for the rest of his life. As Curbelo said:

Let us be respectful and sober in our conversations, our debates, and especially in our social media interactions. And as for Pierre, I wish him the best. He made a mistake, and his life shouldn’t be ruined because of it. I truly believe in second chances. I am hopeful Pierre will become a model citizen who is now in a unique position to be a part of the force for healing that our community and our country so desperately need.