In the wake of Charlie Kirk’s assassination, social media posts and poll results showing support for political violence have been seized upon by members of both parties to blame their opponents, justify retribution or even raise fears of a civil war.
In the wrong corner of the internet, it can certainly seem as if there’s widespread support for political violence. And in recent polls, as many as one in five Americans say political violence can be justified.
But while a small percentage of Americans do endorse violence, that percentage is much smaller than either the polls or social media make it seem. Only a vanishingly small number of Americans support political violence in any meaningful sense.
In theory, public polls seem like a reasonable way to measure support for political violence. Pollsters ask voters whether they support various ideas and policies all the time. And unlike social media, which is dominated by a tiny number of extreme, widely shared and algorithmically promoted voices, political surveys obtain a reasonably representative sample of people across the population.
But support for political violence is not easy to quantify. The measurement issues are so serious that The New York Times/Siena Poll has never once asked about it over the last decade (plenty of people have asked us to do so). The usual poll questions — for instance, “Do you think it is ever justified for citizens to use violence to achieve political goals” — aren’t quite up to the task.
For starters, what’s “political violence”? In the context of this article, it evokes the assassination of political opponents simply for their political views. But that’s not the only thing poll respondents might imagine when they hear the question. It might not even be the only kind of assassination they imagine. What about the baby Hitler problem? What about violence that’s nothing like an assassination, like revolt against a colonial oppressor after nonviolent protest failed? What about violence that isn’t murder? The violent destruction of property — like dumping tea into a harbor — could plausibly be political violence. Nowadays, some people even call speech “violent” if it causes psychological distress or reinforces supposed structures of power and domination.
Are these examples too imaginative? Not given the wording of these questions. Polls usually ask whether political violence is “ever” justified or “sometimes” justified. These terms prompt respondents to imagine a range of circumstances when one might resort to political violence and the forms it might take — and then to evaluate whether there are any cases where it might be justified. If they can imagine any cases, they might well respond that political violence is indeed justified.
Whether political violence might “ever” be justified sets a particularly low bar for an imaginative respondent. “Ever” may not even require them to imagine a case; they may need only to believe it’s possible that they could imagine a case. I’m personally such a respondent. As my colleagues can attest, I recoil when “ever” is suggested for a poll question. Ever?
Even the term “justified” is not quite as clear as pollsters would like. There are two slightly different possible meanings of “justified” in this context: One means that the action is good and just; another is merely to say there’s valid reason or explanation for the action. As a consequence, one can believe there’s a defensible justification for violence — “eye for an eye” logic, to take a classic case — without believing the action is just.
There are other ways that polls might overstate support for political violence. Maybe most surprisingly, survey respondents may not be paying close attention. This is especially important in online surveys, where people can get into the habit of quickly clicking through questions. In a great illustration of this phenomenon, a Pew Research study found that 12 percent of online survey respondents ages 18 to 29 said they were licensed to operate a nuclear submarine.
What you put all of this together, there are a lot of different ways for a poll respondent to appear to offer support for heinous acts of political violence, even when they don’t support them at all. A handful of studies have tried to tackle these challenges by asking more specific questions and by adding attention checks, like asking respondents what they were asked in a prior question. When they do, the support for political violence plunges.
Among attentive respondents, Bright Line Watch found 4 percent who said it is “ever justified” to commit “violent felonies” to advance “political goals.” A More in Common poll found that 3 percent to 4 percent of Trump and Biden supporters thought “physically attacking” the other side was justified if they thought the election was stolen. The Polarization Research Lab found that, among other examples, 2 percent of Americans “supported arson” against the opposing political party headquarters. Only about 1 percent supported murder.
The broad pattern is clear: As the poll question gets more specific and imagined acts become more violent, support for violence falls close to zero. The requisite level of specificity makes it challenging, or even impossible, for pollsters to ask a broad question about political violence. Even with a specific question, it’s not clear that the minuscule remaining support is even real. Perhaps much of that 1 or 2 percent could be explained by trolls; people who didn’t take the question literally; people who misread the question; people who mis-clicked or misspoke, and so on.
Of course, it takes only one person — 0.0000003 percent of Americans — to commit a heinous act of violence; it takes only a single person to post something hateful on social media, which might then get put in front of millions after being amplified by algorithmic promotion or people who decide to share it. And with America so divided, these solitary acts can help the country slide further toward political turmoil. But these solitary acts — or even many solitary acts, like thousands of social media posts — shouldn’t be confused for anything like widespread support. The polls shouldn’t be confused for it, either.
Nate Cohn is The Times’s chief political analyst. He covers elections, public opinion, demographics and polling.
The post Why Support for Political Violence Is Not as High as It May Seem appeared first on New York Times.
