The majority of white nationalists who use genetic ancestry sites aren’t happy with the results, according to a new study.

Sociologists at UCLA and the Data and Society Research Institute examined 12 million posts on a white nationalist website, Stormfront, posted by more than 300,000 users. Within those, researchers found 70 discussion threads containing 153 users who posted their test results from sites like Ancestry.com and 23andMe, with 3,000 individual posts in all, STAT News reported.

One-third of those who posted the findings of their genetic ancestry testing were pleased, according to Aaron Panofsky, a sociologist of science at UCLA’s Institute for Society and Genetics. Meanwhile, the majority who found that their ancestry wasn’t “100 percent white European” either rejected or disputed the findings. Some even claimed to know their own background better than any test could.

“They will talk about the mirror test,” Panofsky told STAT News. “They will say things like, ‘If you see a Jew in the mirror looking back at you, that’s a problem; if you don’t, you’re fine.’”

Some other users who got results they hadn’t expected claimed that they didn’t matter since they were truly committed to being a white nationalist, while others claimed the genetic testing kits were part of a vast Jewish conspiracy that’s “trying to confuse true white Americans about their ancestry,” Panofsky said.

Panofsky and Joan Donovan of the Data and Society Research Institute presented their findings during a sociology conference in Montreal on Monday, roughly 48 hours after thedeadly white nationalist rally in Charlottesville, Va. The timing was incidental, but provides a “useful, if frightening, window” into how some extremists view their ancestry, according to STAT News.

Ancestry.com — which announced its 5 millionth DNA test earlier this month — said in a statement to The Post that the company was “horrified by the tragic events” in Charlottesville and had a message for those in search of ethnic purity.

“We not only condemn the violence that occurred but we are deeply disturbed by the ideologies of the white supremacist groups who marched there,” the statement read. “To be clear, we are against any use of our product in an attempt to promote divisiveness or justify twisted ideologies. People looking to use our services to prove they are ethnically ‘pure’ are going to be deeply disappointed. We encourage them to take their business elsewhere.”

Representatives from 23andMe, meanwhile, said the company conducts ancestry testing that provides a percentage of a person’s regional origin.

“We do not report on race,” senior director of research Joanna Mountain said in a statement to The Post. “In part this is done by using an algorithm to compare large segments of their genome to 31 different reference populations — both publicly available reference data as well as reference data that we have internal to 23andMe. From this, 23andMe is able to estimate percentages of a person’s genome that come from these areas, for instance, sub-Saharan African, Iberian or European.”

23andMe performs genotyping — the process of identifying genetic variants within key parts an individual’s genome — rather than sequencing, which determines the exact order of the four bases in DNA.

“When a 23andMe research participant tells us they have four grandparents all born in the same country — and the country isn’t a colonial nation like the US, Canada or Australia — that person becomes a candidate for inclusion in the reference data,” 23andMe product specialist Jhulianna Cintron told STAT News.

The company then excludes close relatives who could distort the findings and removes outliers whose genetic data doesn’t correspond, she said.

An associate professor at the University of Michigan who has studied commercial genetic testing told STAT News that companies are typically reliable at identifying genetic variants. But interpreting the results as far as ancestry or possible looming health risks is less certain, he said.

“The science is often murky in those areas and gives ambiguous information,” J. Scott Roberts said. “They try to give specific percentages from this region, or X percent disease risk, and my sense is that that is an artificially precise estimate.”