In recent weeks, Americans have witnessed several acts of apparently political violence: Black Lives Matter activists were shot in Minneapolis; three people were killed at a Planned Parenthood clinic shooting in Colorado Springs, and 14 were killed in an attack in San Bernardino.

It is common, and perhaps understandable, to assume that these attacks stem from particular groups that endorse political violence and are therefore more prone to violent acts.

In fact, support for violent political action is more common than many might think.

In 2010, I fielded two national surveys that asked respondents their opinions about the following statements:

"When politicians are damaging the country, citizens should send threats to scare them straight."

"The worst politicians should get a brick through the window to make them stop hurting the country."

"Sometimes the only way to stop bad government is with physical force."

"Some of the problems citizens have with government could be fixed with a few well-aimed bullets."

"Citizens upset by government should never use violence to express their feelings."

Although most people opposed violence, a significant minority (ranging from 5 percent to 14 percent) agreed with each violent option, and 10 percent to 18 percent expressed indifference about violence in politics. This implies that millions of ordinary Americans endorse the general idea of violence in politics.

Interestingly, these violent attitudes did not depend on standard political and demographic characteristics. For example, Republicans and Democrats were indistinguishable in their support for political violence, and liberals and conservatives were, too.

The same was true for religious identification: Across all the religious groups represented in the survey, there were no clear differences in violent attitudes. Admittedly, the modestly sized samples included few respondents from smaller religions in the U.S., which makes reliable statistical comparisons difficult. But the averages within religious groups were within a few percentage points.

So what does relate to violent political attitudes?

By far the strongest factor is an aggressive personality. People who behave more aggressively in everyday life are significantly more likely to support political violence.

The personality questions asked respondents how well statements like "given enough provocation, I may hit a person" described them. Compared with someone having the least aggressive personality, a person having the most aggressive personality was 32 points more likely to support violence.

Another important factor was political disaffection. People who doubted that elections get government to pay attention to citizens were 12 points more supportive of political violence compared to those with the most confidence in elections. These results mirror a 1998 Pew survey in which citizens who distrusted government were more supportive of violence against government.

Of course, these questions focused on violence against leaders and government, and not violence against citizens, as in the recent attacks in San Bernardino and elsewhere. Nonetheless, this snapshot provides one of the clearest views we have on support for political violence.

But support for political violence is not just about personality or disaffection. It also has to do with the language of our politics. Many worry that political rhetoric is fueling the fire. My findings suggest this concern is valid.

In these surveys, I included an experiment in which people were exposed to nonpartisan political vignettes either with or without violent metaphors. Whereas the nonviolent vignettes said things like "I will work hard to get our economy back on track," the more violent vignettes said things like "I will fight hard to get our economy back on track." This rhetoric of 'fighting' for a cause, declaring 'war' on problems, and suffering 'attacks' from opponents is how political leaders, journalists, and citizens often talk about politics.

Reading these "fighting words" also increased support for political violence, mainly among those whose personalities already predisposed them to aggression. Violent metaphors made some people feel more positively about political violence even if they wouldn't necessarily act violently themselves.

Other rhetoric may influence violent attitudes as well. Political leaders, pundits, and citizens regularly demonize opponents and emphasize the righteousness of their own goals. Language like that may facilitate moral disengagement, which allows people rationalize the harm they do to others. Violent acts — including terrorism — are aided by these mental gymnastics.

Ultimately, the challenge in a democratic society is that leaders and citizens must be able to express their political differences freely, but at the same time we must try to ensure that their words do not provoke unintended consequences. Although politics will always be contentious, my research suggests that combative and even violent political rhetoric can make some Americans see violence as an appropriate means to an end.

Nathan Kalmoe is an Assistant Professor of Political Science at Monmouth College. He wrote this article for the Monkey Cage, an independent blog anchored by a group of political scientists from universities around the country and hosted by the Washington Post. See www.washingtonpost.com/blogs/monkey-cage.