May 22, 2017

“Results May Vary:” One Family’s DNA Ethnicity Percentages

DNA ethnicity results may varyFour members of a family–mom, dad, daughter and son–tested with AncestryDNA. Their DNA ethnicity percentages vary. Why?

We talked recently about the limitations of the ethnicity results delivered by the DNA testing companies. We concluded that for the most part, these admixture results are like that short film before the actual feature presentation. That feature presentation in this case is your genealogical match list.

But, the short film is entertaining, and certainly keeps your attention for a while. It can be especially interesting when you have several members of the same family tested so you can compare their ethnicity results. Let’s consider the real AncestryDNA test results of a family we’ll call the “Reese family:”

DNA ethnicity results Reese

The Reese family has something to their great advantage: their family history indicates that they are mostly from Western Europe. This is a big advantage in the climate of today’s ethnicity results as all of the testing companies have far more data in their reference populations pouring out of western Europe than from anywhere else. That means that in general, they are going to be better at telling you about your heritage from Ireland or England than they are at discovering that you are from China or India.

Looking at the ethnicity results for the Reese family, we are tempted to start applying our knowledge of DNA inheritance to the numbers we see. We know that each child should get half of their DNA from their mom, and half from their dad. So our initial reaction might be to look at the dad’s 40% Scandinavian and mom’s 39% Scandinavian and assume that the child would also be about 40% (20% form dad, 20% from mom). You can see that the daughter did in fact measure up to that expectation with 37% Scandinavian. But the son, with only 25%, seems to have fallen short. The temptation to consider the daughter as the far better example of familial inheritance is strong (especially for us daughters, who are so often exceeding expectations!), but of course inaccurate. It is actually very difficult to look at the parent’s numbers alone and estimate the percentages that a child will receive.

Remember that these numbers are tied to actual small pieces of DNA we call SNPs (snips). Of the near 800,000 SNPs evaluated by your testing company, less than half of them are considered valuable for determining your ethnicity. The majority of the SNPs tested are working to estimate how closely you are related to your genealogical cousin. A good SNP for ethnicity purposes has to be ubiquitous enough to show up in many individuals from a given population, but unique enough to only show up in that population, but not any others. It is a difficult balance to strike.

But even when good SNPs are used, it is still difficult for the computer at your testing company to make accurate determinations about your ancestry. Take the Great Britain line in the Reese family data, for example. The dad has 21%, the mom 7%, so it would follow that the largest amount any child could have would be 28%.  We see the daughter (of course!) falling well within that range at 10%, but the son is seven points above at 35%. How does THAT happen?!

Let’s go back to some basic biology. Remember that you have two copies of each chromosome, one from mom and one from dad. These chromosomes are made up of strings of letters denoting the DNA code. That means that at each SNP location you report two letters- one from mom and one from dad. As the testing company is lining these letters up for comparison, they have to decide which letters go together- which are from the same chromosome- which set came from one single source. To illustrate how this works, let’s say we are trying to write two sentences: “The brown dog ate the bone” and “A black cat scared a mouse” where each word in each sentence represents a SNP. However, all the computer sees are two words, it doesn’t know which word goes in which sentence. You can see in this example that it would be fairly easy to get it wrong. Mixing up a couple of the words creates entirely new sentences with very different meanings. The process of determining which set of values goes on which line is called phasing. Often the inconsistencies you see in your DNA test results, weather it be in the matching or in the ethnicity, are because of problems the company has with this very difficult process:

Phasing correct:

DNA phasing correct

Phasing incorrect:

DNA phasing incorrect

While it is not valuable genealogically (for research purposes) to have children tested when both parents are tested, it is fun to compare notes to see who got what. And it can be an excellent way to involve members of the family who are curious about his genetics but not interested in genealogy, per se (like the “Reese” family’s son). We can also very tangibly see the fleeting nature of this DNA stuff. We can see that in one single generation this family has lost all traces of ancestry to several world regions. This really highlights the value of having the oldest generation of family members tested, to try to capture all that they have to offer in their DNA code.

DNA quick guides super bundle of 7Get more genetic genealogy help from one or more of my DNA Quick Guides. Try the “Getting Started” guide or the AncestryDNA guide (for autosomal tests like the Reese’s). Or save a bundle on the bundle pack: all 10 guides!

Come visit me at if you’re ready for a one-on-one consultation: consider me your travel guide for your DNA journey.

About Diahan Southard

Diahan Southard is Your DNA Guide here at Genealogy Gems! She has worked with the Sorenson Molecular Genealogy Foundation, and has been in the genetic genealogy industry since it has been an industry. She holds a degree in Microbiology and her creative side helps her break the science up into delicious bite-sized pieces for you. She's the author of our DNA guides Getting Started: Genetics for Genealogists, and Y Chromosome DNA for Genealogists (click STORE in the menu above)