Select Page

New percentages rolling out for 10 million customers

At 11:59.59 a.m. PDT today, The Legal Genealogist was 49% British, 31% Scandinavian and 20% miscellaneous “low confidence regions.”

At noon, I was 66% England, Wales & Northwestern Europe, 19% Germanic Europe, 7% Norwegian, 6% Swedish and 2% Irish and Scottish.

At 11:59.59 a.m. PDT today, my sister was 37% British, 30% Europe West, 10% Scandinavian, 10% Iberian Peninsula, 8% Europe East and 5% miscellaneous “low confidence regions.”

At noon, she was 46% England, Wales & Northwestern Europe, 26% Germanic Europe, 10% Norwegian, 8% Irish and Scottish, 5% Eastern Europe and Russia, 4% Swedish and 1% Cameroon, Congo, and Southern Bantu Peoples.

Yep. AncestryDNA has updated its ethnicity estimates, and the results are rolling out for all 10 million plus persons who’ve tested with the company.

And the world has changed, yet again.

Ancestry Director of Scientific Communications Barry Starr explained that the update only affects the percentages that we see — how much Irish or German or Native American. It won’t affect our matches at all.

There were two changes in the update: more people — many from Ancestry’s own research studies — were added to the reference populations (the groups of people with well-documented pedigrees to whom our DNA is compared to make conclusions about ethnic origins) and the method of making the comparisons has changed.

In the last update, there were roughly 3000 reference samples assigned to 353 regions of the world. In this update, there are 16,000 reference samples assigned to 380 possible regions. This helps screen out less-likely regions and make more nuanced estimates between, say, Scandinavian and Norwegian or Swedish. There are better samples and regions particularly from Europe and Asia and, although Starr said more work needed to be and is being done there, better in Africa as well.

On the methodology side, in the last version, the bits and pieces of DNA were compared bit by bit, while the new update looks at longer stretches of DNA at a time. That also helps minimize the chances of misreading a person’s ethnic origins.

As a result, Starr said, just about everyone who’s ever tested with Ancestry will see some changes: the percentages will change and low confidence regions will generally be eliminated. And everybody will see exactly how the numbers were changed with a reference as to how thje results were refined.

As an example, here’s my change in the percentages:

AncestryDNA before and after

Ancestry Senior Director of Product Marketing Stephen Baloglu said all AncestryDNA customers will see the new ethnicity estimates today if they happen to log on today, and all customers will get an email notice that the estimates have been updated. In fact, since the emails will be rolled out over several days, you may very well see the update before you get the email.

As you check your new estimates, there will be what Ancestry is calling a transition experience. You’ll be asked a few questions (what do you expect your ethnicity to be, what changes do you expect and how close to what you expected your current estimate is).

You’ll be given the opportunity to review the previous estimates and the new estimates and, for a time, to keep the older estimate and wait to put the update into effect. You’ll be able to toggle back and forth even on the map of regions to see what’s changed. Here for example is mine:

AncestryDNA map before and after

And you’ll be asked how satisfied you are with the new estimates, and how closely you think they align with what you know about your ethnic origins.

There is help available to understand the changes including a white paper and, eventually, a scientific paper will be published explaining the methodology of the change.

If memory serves me correctly, this is now the third ethnicity estimate AncestryDNA customers have had. Because DNA analysis in general and ethnicity estimates in particular are still new and can change based on more and better data, it’s not likely to be the last.

And remember what these numbers are: the only way to get these percentages is by comparing the test results of folks like you and me — alive today — to the test results of other people who are alive today (and not to the actual DNA of our ancient ancestors!!). Because of that, it’s all a numbers game based on one fundamental assumption: that people who live in an area today (say, modern Englishmen and -women) and who have all four of their grandparents born in that area are typical of the genetic signature of that population generations ago.

The numbers are terrific at the continental level: Europe versus Africa versus Asia. But they may still be more than a little problematic at the regional or country level.

So… how did your estimate change? Better? Worse?