A chain of new occasions has sparked alarm about emerging ranges of political violence within the U.S. Those episodes come with the assassination of political activist Charlie Kirk on Sept. 10, 2025; the homicide of a Democratic Minnesota state legislator and her husband in June 2025; and two makes an attempt to kill Donald Trump all through the 2024 presidential marketing campaign.
Some surveys have reported that numerous American citizens are keen to fortify using pressure for political ends, or they imagine that political violence would possibly every now and then be justified.
My analysis is in political science and information analytics. I’ve performed surveys for nearly 25 years. For the previous 3 years, I’ve studied new ways that leverage synthetic intelligence to behavior and analyze interviews.
My very own fresh surveys, which use AI to invite other people about why they provide their solutions, display that the strangely prime stage of fortify according to those questions is most probably the results of confusion about what those questions are asking, no longer exact fortify for political violence.
Police officers lead a procession as pallbearers lift caskets after a funeral rite for Minnesota state Rep. Melissa Hortman and her husband, Mark Hortman, on June 28, 2025, in Minneapolis.
Stephen Maturen/Getty Photographs
A failure to keep up a correspondence
Why would a couple of surveys get the solutions to this necessary query fallacious? I imagine the purpose is a matter known as reaction error. It implies that respondents don’t interpret a query in the way in which the researcher thinks they’re going to.
Because of this, the solutions other people supply don’t in point of fact replicate what the researcher thinks the solutions display.
As an example, asking whether or not any person would fortify using pressure to succeed in a political function raises the query of what the respondent thinks “use of force” method on this context. It might be interpreted as violence, nevertheless it is also interpreted as the usage of criminal method to “force” any person to do one thing.
Such reaction mistakes had been a priority for pollsters ever since survey analysis started. They are able to impact even apparently simple questions.
What did you imply by way of that?
To steer clear of this downside, I used an AI interviewing machine evolved by way of CloudResearch, a well known survey analysis corporate, to invite respondents probably the most similar questions on political violence from earlier surveys. Then I used it to invite what they had been considering after they spoke back the ones questions. This procedure is named cognitive interviewing.
I then used AI to head thru those interviews and categorize them. Two brief reviews that summarize this procedure as carried out to each polls are to be had on-line. Those analyses have no longer been peer-reviewed, and the consequences must be thought to be very initial.
Nevertheless, the consequences obviously reveal that respondents interpret those questions in very alternative ways.
Nuance issues
As an example, in my survey, about 33% of Democrats agreed with the remark that “use of force is justified to remove President Trump from office.” Alternatively, when requested why they agreed, greater than 57% gave responses like this: “I was not thinking physically but more in the sense that he – the president – might need to be ‘fired’ or forced out of office due to rules or laws.” Nonetheless others had been envisioning long term eventualities the place a president illegally seizes energy in a coup.
If you account for those other interpretations of the query, the AI most effective coded about 8% of Democrats as supporting use of pressure in violent phrases underneath present prerequisites.
Even right here, there used to be considerable ambiguity – as an example, this sort of reaction used to be no longer atypical: “The language ‘use of force’ was a bit too broad for me. I could not justify killing Trump, for example, but less extreme uses of force were valid in my eyes.”
In a similar fashion, 29% of Republicans agreed that “use of the military is justified to stop protests against President Trump’s agenda.” Alternatively, virtually the entire respondents who agreed with this remark envisioned the Nationwide Guard interceding nonviolently to prevent violent protests and riots. Handiest about 2.6% of Republicans gave feedback supporting use of the army towards nonviolent protests.
Virtually all those that agreed that use of the army used to be justified expressed ideas like this: “I see the military coming and acting as a police force to stop or prevent the demonstrations that become violent. Peaceful protesters must be allowed to exercise their right to free speech.”
Other people get ready to march in a ‘No Kings’ protest towards Trump management insurance policies in New York Town on June 14, 2025.
Michael M. Santiago/Getty Photographs
When is political violence justified?
Even questions that explicitly ask about political violence are open to huge interpretation. Take, as an example, this query: “Do you think it is ever justified for citizens to resort to violence in order to achieve political goals?”
The loss of a selected situation or location on this query invitations respondents to have interaction in a wide variety of philosophical and ancient hypothesis.
In my survey, virtually 15% of respondents stated violence may every now and then be justified. When requested concerning the examples they had been considering of, respondents cited the American Revolution, the anti-Nazi French Resistance and lots of different incidents as a reason why for his or her responses. Handiest about 3% of respondents stated they had been eager about movements within the U.S. on the present time.
Additionally, virtually all respondents said that violence must be a final hotel when all different non violent and criminal strategies fail.
One respondent illustrated each issues of one sentence: “The (American) colonists tried petitions and negotiations first, but, when those efforts failed, they resorted to armed conflict to gain independence.”
A decision for working out
Even those numbers most probably overestimate American citizens’ fortify for political violence. I learn the interviews, checking the AI machine’s labeling, and concluded that, if anything else, it used to be overestimating fortify for violence.
Different elements can also be distorting reviews of public fortify for political violence. Many surveys are performed basically on-line. One find out about estimated that any place from 4% to 7% of respondents in on-line surveys are “bogus respondents” who’re settling on arbitrary responses. Every other find out about reported that such respondents dramatically build up sure responses on questions on political violence.
Respondents can also be keen to espouse attitudes anonymously on-line that they’d by no means say or do in actual lifestyles. Research have instructed that “online disinhibition effects” or “survey trolling” can have an effect on survey effects.
In sum, my initial analysis means that reaction error is a considerable downside in surveys about political violence.
American citizens virtually universally condemn the new political violence they’ve witnessed. The hot ballot effects appearing differently much more likely stem from confusion about what the questions are asking than exact fortify for political violence.
