Ancestry's DNA Ethnicity Updates

Ancestry has recently updated the way the predict ethnicity. I think the changes are substantial and positive. Although DNA is exact, our interpretation of it, and especially our determination of ethnicity from it, are very subjective. Since there really aren't such things as German, Irish or English genes, testing companies can only compare an individual's DNA with those of sample populations. Better understanding and larger sample sizes improve the accuracy of the testing. Ancestry has done a lot of work in those areas.

My original guess as to my ethnicity was as follows:

40% Europe West (Mesenbrink, Van Gundy, Flier)
30% Great Britain (Stephenson, Davis, Squibb)
10% Irish (Brennan and suspected numerous others)
5% Europe East (Nipkow)
5% Scandinavian (Friend - could be Finnish/Russian too)
5% Europe Jewish (Were Squibbs and Cowan's Jewish migrants?)
5% Iberian Peninsula (Squibb name from Esquivel?)

The first results from Ancestry surprised me a bit with a lower percentage of Great Britain, higher Scandinavian, and higher Irish than I expected.



The latest Ancestry ethnicity estimate changed the regions a bit.

This is a drastic change in my mind. The England category, which now includes northwestern Europe, is 50% higher than the old Great Britain category. Germanic Europe is only 17%, but I suppose with a lot of my Germany heritage from northwest Germany, it probably matches northwestern Europe. Irish and Scandinavian are down a bit, closer to what I had guessed.

Probably the most noticeable improvement I've seen was with my father- and mother-in-law. All four of my father-in-law's grandparents were born in Denmark, although he was considered only 40% Scandinavian. My mother-in-law has no known Scandinavian ancestry, yet Ancestry gave her a higher percentage Scandinavian than her Danish husband. Now Ancestry shows my father-in-law at 75% and my mother-in-law at 19%.

Ancestry's new numbers seem better to me, less surprising. They have made strides by increasing the number of samples for each region and using more regions now. Given that people have frequently moved around, there is considerable overlap between neighboring regions. Ethnicity is not an exact science, but I think it is improving.