Elizabeth Warren’s DNA test fact check: What journalists got wrong about senator's Native American ancestry

After being egged on by President Trump, Senator Elizabeth Warren, D-Mass, released the results of a DNA test 15 October that indicated she had a Native American ancestor.

The results - which identified Native American DNA from six to 10 generations ago - were immediately misinterpreted.

It started with a Boston Globe report, which initially indicated the test showed she was at best 1/32nd Native American and possibly just 1/512th Native American. After confessing twice to a math error, the Globe corrected the numbers to 1/64th and 1/1,024th Native American. That would translate to between 98.44 per cent and 99.9 per cent not Native American.

The Republican National Committee then issued a news release directing reporters to a 2014 New York Times report that said "European-Americans had genomes that were on average 98.6 per cent European, 0.19 per cent African and 0.18 Native American." So it sounded like Ms Warren had less Native American DNA than the average European American.

The Fact Checker even issued a tweet along these lines (at a moment when the Globe still indicated the range was between 1/32nd and 1/1,024th): "So Elizabeth Warren could be 3.1 per cent Native American - or as little as 0.01 per cent Native American."

But it turns out reporters and politicians are not very good at understanding genetics. So we will set the record straight, after reviewing the results in detail and consulting with genetics experts.

Ms Warren's DNA was sequenced and analysed by a group led by Carlos Bustamante, a well-regarded Stanford University geneticist. Researchers studied a fraction - far less than 1/1,000th - of Ms Warren's DNA, and then compared it to the DNA of 148 people from Finland, Italy, Spain, China, Nigeria and North and South America. Additional comparison was done on 185 individuals from Utah and Great Britain.

As one might expect, the vast majority - 95 per cent - of Ms Warren's DNA indicated European ancestors. But five genetic segments were identified, with 99 per cent confidence, as being associated with Native American ancestry. The largest segment identified was on Chromosome 10.

"While the vast majority of the individual's ancestry is European, the results strongly support the existence of an unadmixed Native American ancestor in the individual's pedigree, likely in the range of 6-10 generations ago," the report said.

Here is where the reporting went off course. The report said Ms Warren had 10 times more Native American ancestry than the reference set from Utah and 12 times more than the set from Britain. The report also said the long segment on Chromosome 10 indicated the DNA came from a relatively recent ancestor.

Those are significant findings. But reporters focused on the language indicating a range of between the sixth and 10th generations. That raised the prospect of an ancestor amid hundreds of great-great-great-etc-grandparents. That is because most people, looking at the contributions of generations over time, would imagine an even distribution, with each successive ancestor contributing equally to the DNA of an individual.

But ancestors do not contribute genetic material equally over time, especially when you move past great-grandparents. Some ancestors contribute a lot - while others do nothing at all. In other words, as you go back in time, the number of your ancestors keeps increasing but not nearly as fast as the number of genealogical ancestors. In the sixth generation, a genetic chart of an individual would show some strong contributors of genetic material - and many weak ones.

The most important point is this: The results in Ms Warren's DNA test are static. The percentage of Native American DNA in her genome does not shrink as you go back generations. There could be one individual in the sixth generation - living around the mid-1800s, which is similar to Warren family lore - or possibly a dozen or more ancestors back to the 10th generation, which would be about 250 years ago. Her results are consistent with a single ancestor, however.

(Note: Mr Bustamante did not have access to Native American DNA because of mistrust in the community that DNA results could affect tribal identity, so he relied on samples of indigenous people from Mexico, Peru and Colombia - populations in the Americas with high Native American genetic ancestry. There is research showing that using these groups as references is accurate when differentiating between genetic ancestries at a worldwide level. But no tribe for Ms Warren could be identified, only that she had an ancestor or ancestors descended from indigenous people.)

This basic error in understanding the test results was compounded by the RNC's reference to the 2014 New York Times article, which was about a genetic profile of the United States based on a study of 160,000 people drawn from the customer base of 23andMe, a consumer genetics company. With reporters believing Ms Warren's genome was only as little as 0.01 per cent Native American, the article's line that "European-Americans had genomes that were on average 98.6 per cent European, 0.19 percent African and 0.18 Native American" made it appear as if Ms Warren's sample was even smaller than that of the average American. But the majority of European Americans in the study have zero.

Not so. Remember we said that the Bustamante study said she had 10 times more than the individuals from Utah? That is the relevant statistic, indicating her claim to some Native American heritage is much stronger than most European Americans.

In fact, the 23andMe study used a different methodology, so it cannot be compared to the Bustamante report. Moreover, the reference to an average "European-American" is misleading, because there are wide variations in the genetic makeup, with the vast majority of European Americans having no Native American ancestry. The small percentage of European Americans with more than 2 per cent Native American ancestry are concentrated in a handful of states, such as North Dakota, New Mexico and Louisiana.

Mike Reed, the RNC spokesman who circulated the Times article, said in response: "The bottom line is Elizabeth Warren has, at most, a minuscule amount of Native American heritage and it is obvious she had absolutely no right to claim minority status while climbing the professional ladder to the Ivy League."

The test results certainly have not won fans in the indigenous community.

Kim TallBear, associate professor at the University of Alberta, said the "very desire to locate a claim to Native American identity in a DNA marker inherited from a long-ago ancestor is a settler-colonial racial understanding of what it is to be Native American." In an email, Ms TallBear said Native Americans' own definitions of legitimate Native American or tribal identity focus not on long-ago ancestors identified through a test but are based on a living community, or "close social and biological relations of people one can name, indeed people one probably knows (huge LOL here) - one's family, community and tribe."

We are not trying to defend Ms Warren's decision to release the test, just to set the record straight about what the test shows. The media bungled the interpretation of the results - and then Ms Warren's opponents used the uninformed reporting to undermine the test results even further. We fell into this trap as well and were too quick to send out a tweet (now deleted) that made an inaccurate comparison. We should have not relied on media reporting before tweeting.

Ms Warren's Native American DNA, as identified in the test, may not be large, but it is wrong to say it is as little as 1/1,024th or that it is less than the average European American. Three Pinocchios all around - including to our tweet.

The Washington Post