WEBLOG

May 2nd, 2026 (Permalink)

Doing Violence to the Data

Batya Ungar-Sargon, "Debunking the Data That Claims to Show Most Political Violence Comes From the Right", 4/28/2026

The third assassination attempt on President Trump's life this weekend has reignited a debate between Left and Right about where political violence in America comes from. The Right points to the assassination attempts on the President, the murder of Charlie Kirk, the rise of Islamist terrorism, the rabid violence of the George Floyd riots, the elevation of political violence fan Hasan Piker to celebrity status in the Democratic Party, and the recent polling showing that the more liberal a person is, the more likely they are to support political violence.
On the Left, people point to January 6 as well as data purporting to show that most political violence comes from the Right. … The problem is, the "data" that these outlets have been relying on is deeply flawed.
One of the major sources is the Prosecution Project…which analyzes felony criminal cases involving political violence and sorts them by ideology. … Yet if you pull up the data center yourself, you can see immediately that it is deeply flawed.
The data set doesn't include either of the previous two assassination attempts on President Trump's life, as far as I can tell; a search for the time frame and the names of the would-be assassins turns up zero hits. Nor does it include the assassination of Charlie Kirk. The data set is based on prosecutions, which might explain the absence of [the first would-be assassin of Trump], who died at Butler. But what explains the absence of Trump's other would-be assassin…? It's pretty easy to say that the violence is coming overwhelmingly from the Right if you overwhelmingly edit out any political violence from the Left.

The Prosecution Project's (tPP) data is inherently skewed by being based solely on violence that's prosecuted. Ungar-Sargon mentions that the lack of inclusion of the first assassination attempt on President Trump may be explained by the lack of prosecution of the assassin, who was shot and killed at the scene, but there are other reasons why some political violence doesn't get prosecuted.

The editing goes deep. During the summer of 2020, the George Floyd riots were in full swing. Political violence claimed the lives of dozens of Americans and caused $2 billion in property damage. Yet the data set from the Prosecution Project lists …just three incidents of left-wing violence during that time. Maybe you think that people murdered during political riots shouldn't count as victims of political violence. …

Well, it depends. There are three problems with data for violence committed during those riots: first, it's difficult to identify, arrest, and prosecute perpetrators of violence during riots even if the effort is made; second, in some cases the effort was not made¹; finally, many of those who committed violence were simply criminals taking advantage of the chaos caused by the riots to commit crimes.

How is one to separate the politically-motivated violence from the opportunistic? One could, of course, simply categorize all acts of violence committed during such riots as political, but then acts of self-defense against rioters would be counted too.

In any case, the tPP's data is systematically biased because it's based on prosecutions and, at least, should not be the sole source of data for a study of politically-motivated violence.

I couldn't find a complete dataset for the CSIS [The Center for Strategic and International Studies] graphs², yet in their methodology, they note some glaring absences. For starters, the information it cites is culled from data provided by the ADL [Anti-Defamation League], which "uses public records such as media reports and police filings to reach their numbers," and the Southern Poverty Law Center―the same organization that was just indicted for actually manufacturing the racist violence it was "chronicling." Needless to say, if you are relying on the media, wildly skewed to favor the Left, for your data set, it's not data; it's propaganda.
Moreover, even when acknowledging that left-wing political violence is on the rise, CSIS admits it went out of its way to absolve the Left of even more violence. In a recent report, it excluded pro-Palestinian terrorism from the Left, reclassifying it as "ethnonationalist incidents rather than left-wing ones," despite the Palestinian cause becoming the most important litmus test for belonging on the Left these days. This is just a naked attempt to absolve the Left of a signature issue because that issue inspires violence and the people tallying up the crimes want the Left to win. …
This is far from an exhaustive list of the way the Left cooks the books when it comes to its tabulations of where political violence is coming from. … Suffice it to say, it's enough to make any honest broker truly suspicious of what they're seeing out there as "data."

As I've pointed out previously³, to count something you first have to define it. For instance, if you were going to take a census of all the bald men in Baltimore, you'd first have to define "bald"―and "Baltimore". So, if you had an incentive to minimize the number of bald men in Baltimore, you'd be tempted to define both "bald" and "Baltimore" in narrow ways. In addition, no matter how you define the two terms, there will be borderline cases that could reasonably be put into one bucket or the other. Given your incentive to minimize the number of bald men, you'll be tempted to put the borderline cases in the "not bald" bin.

Similarly, to count cases of political violence you have to define both "political" and "violence", and it's sometimes unclear whether a particular case was either politically-motivated or an act of violence, which leaves a lot of wiggle room for politically-motivated manipulation of the data.

Finally, even should one succeed in accurately counting cases of politically-motivated violence, separating them into "left" and "right" requires further definitions. As Ungar-Sargon explains, The Center for Strategic and International Studies (CSIS) classified anti-Israel political violence as "ethnonationalist incidents" rather than "left"⁴. This is fine but then it classifies the violence committed by Trump supporters during the Capitol riots as "right" despite being ethnonationalist⁵.

For these reasons, any attempts to count incidents of political violence so as to blame one side or to exonerate the other should be approached with great caution, and studies on politically-charged topics by "think tanks" should always be treated skeptically. There is a lot of advocacy research from organizations with carefully neutral names, such as "The Prosecution Project" and "The Center for Strategic and International Studies", that reveal nothing of their political biases. Such groups are often funded by wealthy foundations and others with political views and goals⁶. The money that such donors donate may not come with explicit strings attached, but it always comes with implicit ones: the think tanks know that if they displease their donors, the money will stop flowing.

Now, I'm not saying that such funding skewed the research done by CSIS or tPP; but I'm also not saying that it didn't. What I am saying is that there are good reasons to be skeptical of any studies put out by such organizations. After putting both thumbs on the scale, the CSIS report admits that left-wing political violence is both on the rise and exceeded that of right-wing violence last year⁷. So far, this year is continuing the trend.

Notes:

↑ For instance, according to The Grauniad, most of the charges against "protestors" during the riots were dropped, see: Tom Perkins, "Most charges against George Floyd protesters dropped, analysis shows", The Guardian, 4/17/2021.
↑ Red flag.
↑ For instance: "Charts & Graphs: The Case of the Missing Murders", 11/10/2022.
↑ Daniel Byman & Riley McCabe, "Appendix: What Is Excluded?", The Center for Strategic and International Studies (CSIS), 9/25/2025.
↑ Byman & McCabe, "Methodology and Codebook", CSIS, 9/2025. See under "Riots".
↑ See: "Our Donors", CSIS, accessed: 4/30/2026.
↑ Byman & McCabe, "Left-Wing Terrorism Incidents Are on the Rise", CSIS, 9/25/2025.

Disclosure: I don't agree with everything in this article, but I think it's worth reading as a whole. In the above selected excerpts, I have sometimes suppressed the paragraphing and always removed the names of would-be assassins: I don't need to know their names, but they're in the original if you must know.

Previous Month | RSS/XML | Current