Chart inflates prevalence of transgender people committing mass shootings

"Who Is Behind Mass Shootings?" a February 11, 2026 post on X asks, answering with a chart that claims to assign incidents with four or more fatalities to different races and gender identities.

Image

A screenshot of an X post taken on February 19, 2026

The image accumulated hundreds of thousands of engagements across social media in February, following a school shooting in British Columbia, Canada and another shooting arising from an apparent "family dispute" at an ice rink in the US state of Rhode Island. X owner Elon Musk amplified the chart to his more than 234 million followers, writing: "Wow."

The graph has circulated online since at least August 2025, AFP found.

It stems from X user Frank McCormick, who first posted it August 27, 2025 on his "@CBHeresy" account, before adding a variation on February 17, 2026.

Although the post does not explicitly say it is referencing the United States, McCormick later shared his sources, which include the US census and a database of shooting incidents compiled by Mother Jones, an American magazine.

Experts told AFP that despite frequent mass shootings in the United States, the posts framing transgender people as the most likely demographic to commit such acts of violence are mistaken.

"It is plainly obvious that most US shootings are not being done by people who are transgender," said Aaron Chalfin, associate professor of criminology at the University of Pennsylvania (archived here).

He told AFP February 18 that in addition to mass shootings, "a considerable majority of individuals arrested for shootings and homicides are people who are born male and who identify as male."

AFP reached out to McCormick -- who has claimed that expressing his views cost him his career as a high school teacher -- but a response was not forthcoming.

Male perpetrators

A factsheet compiled by the Rockefeller Institute of Government, a public policy think tank, describes the profile of mass shooters in the United States as "white" and "overwhelmingly male" (archived here).

Further data has been compiled by the Gun Violence Archive, which carefully tracks mass shootings in the United States, recording hundreds each year.

Its founding executive director, Mark Bryant, said his organization started specifying if a suspect was transgender following debate and misinformation that surrounded a 2023 school shooting in Nashville.

They assigned researchers to examine case reports on records kept since 2013 and found the first known incident in Aberdeen, Maryland in 2018 -- a shooting that left three dead and another three injured.

The organization shared the full dataset with AFP for review. It identifies five cases where the perpetrator of a mass shooting -- which they define as an event resulting in four or more victims injured or killed by gunfire, not including the shooter -- was transgender. 

"As you can see, that gives us a rate of 5/5000, or 1/10, of 1 percent of mass shootings," he told AFP February 13.

Several other experts said the chart circulating online also makes assumptions about what share of the US population is transgender or non binary. Gender identification and fluidity can make a strict categorization flawed (archived here).

A survey from Gallup, an analytics and advisory company, suggests that about 1.1 percent of US adults currently identify as trangender, with LGBTQ+ identification estimated at 9 percent nationwide (archived here). The survey involves the largest dataset to date on the matter.

Comparing gender identity to race in an "apples to oranges" way also risks overlapping data points, said James Nolan, a professor in the department of sociology and anthropology at West Virginia University (archived here).

"This is not something a researcher seeking the truth would do. It is more like somebody using data to mislead or to fit some sordid narrative," he said February 19, adding that the same methods used to selectively analyze a different demographic group could yield entirely different conclusions.

Flawed data

Statiticians told AFP the per-capita framing presented by McCormick in his graph did not make sense, either.

Responding to McCormick's explanation of how he crunched his numbers, Jing Cao, chair of the department of statistics and data science at Southern Methodist University, said (archived here): "The sample is too small, definitions are fragile, and denominators aren't comparable."

"If you take 2 perpetrators out of 32, that is indeed 6.25 percent. If you then divide by a claimed population size of ~2.8M out of ~341M, you'll mechanically get a much larger per-capita rate for the smaller group," she told AFP in a February 18 email. "That's just how denominators work."

Maria Cuellar, assistant professor in criminology and statistics and data science at the University of Pennsylvania, agreed (archived here).

"Ratios are unstable when the denominator is small," she said. "Per-capita rates standardize for population size, which is good, but they amplify noise when the population is small, the event is rare, or the numerator is near zero."

Even so, the data do not add up, according to Ty Partridge, director of Wayne State University's research, design and analysis unit (archived here). Per-capita proportions look a lot different than what is shown on X by McCormick, per Partridge's own analysis using Pew Research Center and Gun Violence Archive datasets (archived here).

"The 'statistics' being reported are absolute fabrications," Partridge said on February 20.

Stigmatizing tropes

The claims of "trans terrorism" and viewing gender dysphoria as a "mental illness" also perpetuate hurtful and stigmatizing tropes, a spokesperson for the LGBTQ media advocacy organization GLAAD told AFP on February 18 (archived here).

"We see irrational fear and focus on scapegoating vulnerable communities in these high profile crimes," the spokesperson said.

AFP has previously investigated other false and misleading claims targeting about transgender and queer people following violent events in the United States.