In the aftermath of Charlie Kirk’s assassination—and the assassination of Minnesota state legislator Melissa Hortman only three months ago—questions inevitably arise about how to reduce the likelihood of similar heinous actions.
Results from arguably the most important study focused on the U.S. context, the Strengthening Democracy Challenge run by Stanford University, point to one straightforward answer: show people that very few in the other party support political violence. This approach has been shown to reduce support for political violence.
Showing that few Americans support political violence is clearly not the only approach (see various resources like these), but it represents a strong starting point. This method often corrects misperceptions about out-party political violence support. The work of social psychologist Kurt Gray and others shows that people think more like prey rather than predators; people usually try to reduce harm and, generally, only possibly respond if attacked first. When Americans think that those in the other party are less threatening to me / us, there is less need for me / us to be threatening to them.
The Strengthening Democracy Challenge found that five of 25 tested interventions significantly reduced support for political violence. This article unpacks these five interventions, especially the top two performers, which explicitly showed that very few in the other party support political violence.
Even with a need for more research, it is unfortunate that, aside from a few leaders, organizations have not taken this conclusion from the Strengthening Democracy Challenge and have not implemented programming or messaging around it.
Yet this can change, and organizations (including my own, More Like US) can take steps to share that few Americans support political violence.
Such an approach is entirely compatible with still being concerned about violence. It can simultaneously be true that a small sliver of the public can cause terrible damage in terms of violence, and the American public should know that the vast majority of their fellow citizens oppose it.
Looking at online spaces may lead to an incorrect perception of support for political violence. However, More in Common shows that those who spend more time on social media indicate more support for political violence, and the Financial Times / Co-operative Election study finds that content from the political extremes on social media are vastly overrepresented.
A useful resource is the Polarization Research Lab, which tracks attitudes and perceptions toward political violence with weekly surveys. Americans vastly overestimate the share of those in the other party willing to condone political violence, including murder, where perceptions of support are at least 15 times greater than in reality. Results like these should be publicized much more widely.
Background on the Strengthening Democracy Challenge and questions about political violence
The Strengthening Democracy Challenge tested the effectiveness of interventions of eight minutes or less focused on the U.S. political environment. All interventions were online and asynchronous, so they did not test face-to-face conversations or interpersonal activities.
While the researchers covered various goals, they were most focused on significantly reducing support for political violence, anti-democratic attitudes, and partisan animosity.
Over 250 interventions were submitted, and the organizers chose 25 that seemed most promising. About 1,000 Americans tested each of these most-promising interventions, and roughly another 5,000 served as a control group. The organizers called this setup a “megastudy.”
Regarding political violence, participants were asked questions about the extent to which it was 1) OK to send threatening and intimidating messages to leaders in the other party, 2)
OK for an ordinary member of one’s own party in the public to harass an ordinary member of the other party on the internet in a way that makes the target feel frightened, 3) justified for members of one’s own party to use violence in advancing their political goals these days, and 4) justified for one’s own party to use violence if the other party wins more races in the next election.
Ultimately, the researchers published their findings in the prestigious journal Science last year.
Results reveal the benefits of showing that very few in the other party support political violence
Of the five interventions significantly reducing support for political violence, the two that performed best clearly showed that very few in the other party support political violence. In many cases, this finding will help correct misperceptions of widespread political violence support.
In terms of the mechanism to why this works, it can help to think of an example. One may think to themselves that if those in the other party do not support violence against us, there is little reason for me to support violence against them.
Going deeper, psychologist Gray and others show that people have evolved as prey rather than predators, so moral actions are usually based on defensiveness and harm reduction, not proactive viciousness.
The two interventions that explicitly showed those in the other party overwhelmingly disapprove of political violence are as follows:
- Correcting Democracy Misperceptions (the most impactful intervention) is an eight-question game in which participants are “asked to guess what people from the other party believe when it comes to actions that undermine how democracy works.” One of these eight questions is essentially: “Would most (members of the other party) support using violence to block major laws (from your party)?” The answer was “never” in both Republican and Democratic cases.
- Pro-Democracy Inparty Elite Cues (the second-most impactful intervention) involved reading and summarizing a “fictional op-ed with real quotes and statistics…[Participants] learn that the leader of their party (Biden or Trump) condemns violence and supports democratic processes (e.g., right to vote, freedom of the press). The op-ed also cites data from researchers Nathan Kalmoe and Lilliana Mason about how at least 90% of their own party do not support violence or breaking the rules to help their party win.”
Other details and results of other interventions are uncertain
Without further research, some details about mechanisms and what works are unclear.
Even in the two interventions above, does it matter that those interventions used data on support for political violence in the context of political goals, such as blocking major laws or helping a party win? Would it be equally effective to show low rates of support for political violence in general?
The mechanisms are quite unclear for the other three interventions that significantly (but barely) reduced support for political violence.
- Correcting Division Misperceptions (the third-most impactful intervention) is a video showing fellow Americans learning “the extent to which Democrats and Republicans agree is much more than they expected.” Communication with a lead researcher has shown that data correcting dramatic misperceptions of “meta-dehumanization”—assuming that those in the other party dehumanize those in one’s own party much more than in reality—accounted for nearly all of the benefits from the video. A potential mechanism linking this to violence would be historical examples of sub-human name calling, such as “cockroaches” during the Rwandan Genocide. However, that follow-up study did not show significant impacts to support for political violence, so maybe this mechanism does not reliably work.
- Pro-Democracy Bipartisan Elite Cues (the fourth-most impactful intervention) is a real 2020 campaign ad “with a Democrat and a Republican candidate who were running against each other to be [Utah’s] governor. Each candidate emphasizes that all votes will be counted and they will honor the peaceful transfer of power.” Both candidates say they will fully support the results of the 2020 election. Because the Strengthening Democracy Challenge happened after January 6, it is possible that some participants interpreted that statement as in opposition to violence but this is simply conjecture without further testing.
- Outpartisans' Willingness to Learn (the fifth-most impactful intervention) involves hypothetical exchanging of “messages with someone from the other party who is seeking an open-minded exchange.” The exchanges start after the individual sees answers on a survey from the supposed conversation partner, showing that they are really interested to learn about their partner’s point of view and do not want to convince their partner that they are right. The connection to political violence seems quite indirect and unclear, but it does present those from the other party in a better light.
Taking a step back, it is hard to draw other conclusions. Of the five interventions that significantly reduced support for political violence, three focused on all Americans, while two focused on political leaders. The method of intervention was all over the map—while there were two videos, there was also a multiple-choice question game, an op-ed, and a supposed exchange of messages.
What can be done: Use Polarization Research Lab data to show that very few in the other party support political violence
So far, the only clear initiative that has come from these interventions is Disagree Better, originally a National Governors Association initiative spearheaded by Utah Governor Spencer Cox, who was part of the 2020 Utah campaign ad described above. In the aftermath of Charlie Kirk’s assassination, this is getting more media attention than ever since Gov. Cox is so focused on it.
This initiative does have value in encouraging people not to lash out and make their thoughts and feelings about each other worse. This said, it does not directly deal with violence. (It also does not relate much to the video itself, which was much more about commonalities and shared opinions, rather than how to disagree.)
Despite the uncertainties in the research, I draw the conclusion that showing very low rates of support for political violence among all Americans—especially among those in the other political party—seems to be the best option. While further research can refine this conclusion, waiting for perfect results is a dangerous “paralysis by analysis.”
The only possible pitfall I see is a misunderstanding that this messaging about political violence is saying that violence is not a problem. Yet messaging can acknowledge that a small fraction of the public can cause terrible damage in terms of violence, while also communicating to the American public that the vast majority of their fellow citizens oppose political violence. Good messaging can convey both of these points.
According to the Polarization Research Lab, Americans dramatically overestimate the share of those in the other party willing to condone political violence. In terms of out-party murder as of mid-September 2025, 1.3% of Republicans and 1.9% of Democrats claim to support it. This is admittedly troubling, but it is vastly better than Americans’ estimates of the other party. Both Republicans and Democrats think roughly one-third of those in the other party support political murder. This is about 17x and 26x times greater than self-reported values, respectively.
My organization, More Like US, focuses mostly on college campuses. So far, talking about political murder has seemed like a somewhat extreme and provocative topic for campus, but it will now be much more front-of-mind. Since we have various initiatives and resources correcting political misperceptions, revealing misperceptions on political violence is a relatively straightforward next step.
Let’s not have these assassinations be in vain. A silver lining is that very few Americans support political violence. It is possible to acknowledge the risks of this political violence while also showing that the American public overwhelmingly opposes it.
James Coan is the co-founder and executive director of More Like US. Coan can be contacted at James@morelikeus.org
