Ethnicity estimates are still just estimates

There’s been a bit of a buzz recently as the DNA testing company 23andMe has updated its ethnicity estimates.

For most people — The Legal Genealogist included — the estimates seem to be a bit closer overall to what’s known about family origins in the last 200 years.

For myself, for example, as a first-generation American of German paternal descent, I was pleased to see that 23andMe has raised my French and German to 36.6% and has identified it as all German. That’s not quite the 50% I’d expect with my father, paternal grandparents, great grandparents and so on all born in what is today known as Germany, but it’s considerably better than it has been in the past.

On the heels of Ancestry finally discovering that I have German ancestry (I’m all the way up to 19% Germanic Europe at Ancestry, up from a measly 4% Western European in my last ethnicity estimate there), the numbers are clearly getting better for the one-in-six Americans who have German ancestors.

And the real buzz with the new 23andMe estimates is that the company is now offering a bit of granularity in the results: the newest 23andMe estimate says I’m not just German but even Bavarian!

Except, of course, for the fact that there isn’t a single solitary Bavarian anywhere in my family tree in the last 200 years. My father and grandmother and all of that side of my father’s family can be documented in Bremen, and my grandfather’s side for at least 200 years is solidly in Sachsen-Anhalt and Thüringen — which at least neighbors Bavaria…

In other words, the estimates are a little better than they were, so maybe we don’t quite have to take them with an entire salt lick,1 but just with a whole bunch of grains of salt.


The reality here is that ethnicity estimates are pretty darned good at the continental level. All of the testing companies do a reasonable job of distinguishing between Europe, Asia and Africa, just to name three continents, in their estimates. Once they get below the continental level, to a regional or even country level, all of them start to run into issues: country boundaries have changed; entire populations have moved; people from one area have invaded and intermarried with people from another.

We can’t forget for a moment what these admixture tests do: they take the DNA of living people — us, the test takers — and they compare it to the DNA of other living people — people whose parents and grandparents and, sometimes, even great grandparents all come from one geographic area. Then they try to extrapolate backwards into time. Nobody is out there running around, digging up 500- or 1,000-year-old bones, extracting DNA for us to compare our own DNA to.

So coming up with these percentages in these tests requires this fundamental assumption: that the DNA of the reference populations — those groups whose parents, grandparents, great grandparents and more all come from the same area — is likely to reflect what we might see if we could test the DNA of people who lived in that area hundreds and thousands of years ago.

In other words, these percentages are:

• estimates,

• estimates based on comparisons not to actual historical populations but rather to small groups of people living today, and

• estimates based purely on the statistical odds that those small groups tell us something meaningful about past populations.

These limitations are true of all of the testing companies. My own results are — literally and figuratively — all over the map. I’m German with some companies, not German at all with another. Largely Scandinavian with one, only slightly Scandinavian with the others.

DNA testing for genealogical purposes is a wonderful tool. But people get disappointed when they see these percentages and they don’t match up to their own paper trail and don’t match up from company to company. And when they get disappointed, they may lose interest in genealogy or in DNA testing. And when they lose interest, we lose out on the paper trail information they might add to our mix.

Worse, ethnicity estimates that are not what people expected lead them — and their political friends — to regard all genealogical DNA testing as a fraud, missing the whole point that the real value of testing is not in the ethnicity percentages but in the cousin matching — helping us connect to those cousins who may have family Bibles or photographs or documents or stories to help us move forward in reconstructing our family histories.

Despite the advances, the bottom line remains: We need to educate our friends and our families, our DNA cousins, as to the limits of what these percentages can show — and to show them all the other things DNA testing really can help with.

So let’s repeat for the record here — everyone who’s ever taken a DNA test that provides a set of ethnicity or admixture estimate percentages, repeat after me:

“It’s not soup yet.”2

Take the estimates with at least these grains of salt — and let’s all move forward and work on those cousin matches.

Cite/link to this post: Judy G. Russell, “And still not soup…,” The Legal Genealogist, posted 27 Jan 2019 ( : accessed (date)).


  1. A salt lick, for the city slickers among us, is “a place to which animals go to lick naturally occurring salt deposits” or “a block of salt or salt preparation provided, as in a pasture, for cattle, horses, etc.” “Mineral lick,” ( : accessed 27 Jan 2019).
  2. For those too young to remember the reference, the Lipton Soup Company had a string of ads in the late 1960s and early 1970s. The mother in the ad would begin preparing Lipton soup, a child would ask, over and over, “is it soup yet?” and the mother would answer “not yet” over and over until finally she’d say, “It’s soup!” So “not soup yet” means “not finished” or “not ready.”
Print Friendly, PDF & Email