Ancestry's latest Ethnicity update to Ancestral Regions

[DianaCanada]

OH’s father was of English, Irish (Protestant), Scottish, wish a dash Germanic (early US settlers), and his mother was Danish, Swedish, and a bit of Germanic Europe….but surprisingly, and quite coincidentally, both have a smidgeon of North African DNA. That one is pretty unlikely.

[rsel]

But to be more accurate, it would require more Germans to do DNA tests, and those Germans to also do accurate trees in sufficient numbers.

I think you are hitting the nail on the head here melba_schmelba.  We really don't know the quality of the reference population tree data that Ancestry have used, and from which locations they are actually from. If for example they have based Germanic based on American testers who show german ancestors, rather than actual german testers it could make a big difference. 
Despite what ancestry say about all the science and trying to focus less on the 'estimate' i will still just consider these as a 'bit of fun' rather than anything serious. Thats not trying to disparage the hard work that i am sure a lot of people at ancestry have done, but i think the marketing people have over emphasised what is still a very imprecise 'science'.

Richard

[Nova67]

My father is getting 91% Germanic Europe. His parentage is German and Sudeten-German. The 2023 update was 75%. It is increasing every year.

[Biggles50]

But to be more accurate, it would require more Germans to do DNA tests, and those Germans to also do accurate trees in sufficient numbers.

I think you are hitting the nail on the head here melba_schmelba.  We really don't know the quality of the reference population tree data that Ancestry have used, and from which locations they are actually from. If for example they have based Germanic based on American testers who show german ancestors, rather than actual german testers it could make a big difference. 
Despite what ancestry say about all the science and trying to focus less on the 'estimate' i will still just consider these as a 'bit of fun' rather than anything serious. Thats not trying to disparage the hard work that i am sure a lot of people at ancestry have done, but i think the marketing people have over emphasised what is still a very imprecise 'science'.

Richard

I have well over 150 DNA matches in my 4C or closer category who seem to have a common country of origin of their families.

The majority are based in the USA with sizeable numbers being in the UK and smaller numbers in Australia and New Zealand.

Of these the vast majority have no family tree of any kind, others have a tree of less than 10.

The common country of the origin of the families of these matches is Italy, the surnames is often the only clue that I have in their families origin.

Just looking through 100 of my own closest DNA matches which range down the 34cM with the hundredth I find that of those I have only linked 28 into my family tree.  The 72 “unknowns” generally have no tree or a very limited tree and as such other techniques would need to be used in determining who is actually who.

So whatever sources Ancestry is using then I do not see family trees as being any significant element in their analysis.

As it is the latest Ancestry predictions on my own ethnicity now for the first time ever they reflect my own family tree very accurately.

Conversely one of the DNA tests that I manage has a Mother who was born in Europe and their Maternal ethnicity is all over the place and there is zero documentation that I have found nor have the shared matches yet to help in revealing actual origins.

[melba_schmelba]

OH’s father was of English, Irish (Protestant), Scottish, wish a dash Germanic (early US settlers), and his mother was Danish, Swedish, and a bit of Germanic Europe….but surprisingly, and quite coincidentally, both have a smidgeon of North African DNA. That one is pretty unlikely.

Interesting you mention that, as I noticed a close relative who got this, and also a bit of Spanish, which I think looking at their fairly thoroughly traced family tree is unlikely!! I am not sure if people remember the sort of results people were getting 10 years back, but it was a bit of a joke that so many people got 1% Nigerian!! Also Brits and Irish regularly got Spanish and Basque, which was eventually corrected, unfortunately for whatever reason Ancestry's algorithm with this update seems to have gone backwards in accuracy. You saw that in similar pattern of unlikely results MyHeritage's own results until their update this year (which many people are still to get?!).

[melba_schmelba]

But to be more accurate, it would require more Germans to do DNA tests, and those Germans to also do accurate trees in sufficient numbers.

I think you are hitting the nail on the head here melba_schmelba.  We really don't know the quality of the reference population tree data that Ancestry have used, and from which locations they are actually from. If for example they have based Germanic based on American testers who show german ancestors, rather than actual german testers it could make a big difference. 
Despite what ancestry say about all the science and trying to focus less on the 'estimate' i will still just consider these as a 'bit of fun' rather than anything serious. Thats not trying to disparage the hard work that i am sure a lot of people at ancestry have done, but i think the marketing people have over emphasised what is still a very imprecise 'science'.

Richard

Yes, I wouldn't be surprised if they are using people of significant Pennsylvania Dutch ancestry as a model, as I think they probably use Quebecois for French. I know for the latter that has big pitfalls as Quebecois came from a fairly limited original population, and they came from specific areas of France in particular, and not from other areas. Ancestry do market their DNA product in Germany, so they may have a reasonable amount of samples that way, but I am not sure if they have ever released figures on that.

[DianaCanada]

Am not sure where I read this, but apparently the largest single ancestral ethnic group in the US are the Germans.  Not only the Pennsylvania Dutch, but other Germans in that (then) colony, and they also settled in the more northern Southern colonies, and New York state had a sizeable number.  They kept coming in, too, over the centuries. 
Still having ties to Quebec, although I am an Anglo, I somehow doubt the Quebecois French would be testing a lot with Ancestry - their genealogy is well documented, and as you mentioned, they come from a relatively small pool of ancestors (although many do find a non-French ancestor, there was intermarriage with Catholic Irish after the famine, some even married Anglos, my aunt married a French-Canadian but he also had Irish ancestry). 
You are right though, many Americans with French Canadian ancestry might have tested - don't forget the Acadians, many ended up in Louisiana, and they might contribute to Americans with French ancestry.  Many Quebecois headed to New England to work in the mills at the end of the 1800's and have a large number of descendants in Massachusetts and New Hampshire in particular.

[melba_schmelba]

Am not sure where I read this, but apparently the largest single ancestral ethnic group in the US are the Germans.  Not only the Pennsylvania Dutch, but other Germans in that (then) colony, and they also settled in the more northern Southern colonies, and New York state had a sizeable number.  They kept coming in, too, over the centuries. 

Yes, I think most were Palatine Germans
https://www.nps.gov/articles/000/the-palatine-germans.htm
https://blogs.bl.uk/european/2015/09/the-poor-palatines.html

Still having ties to Quebec, although I am an Anglo, I somehow doubt the Quebecois French would be testing a lot with Ancestry - their genealogy is well documented, and as you mentioned, they come from a relatively small pool of ancestors (although many do find a non-French ancestor, there was intermarriage with Catholic Irish after the famine, some even married Anglos, my aunt married a French-Canadian but he also had Irish ancestry). 
You are right though, many Americans with French Canadian ancestry might have tested - don't forget the Acadians, many ended up in Louisiana, and they might contribute to Americans with French ancestry.  Many Quebecois headed to New England to work in the mills at the end of the 1800's and have a large number of descendants in Massachusetts and New Hampshire in particular.

Well, I think I can rather definitely say that is not correct, as my own close relative has a rather vast number of Quebecois matches ! Lots of people just do the test to find relatives they may have lost contact with or to resolve doubts over their own origins, curiosity, or I suspect very often simply because someone gave it to them as a gift .

[Wexflyer]

Any regular here knows that an altogether too large proportion of Ancestry family trees are simply wrong - especially the trees of the rebel colonists in North America.
Now, Ancestry DNA ethnicity estimates are not based on "science", but on those Ancestry family trees.

Ergo, their ethnicity estimates are just as reliable as those family trees they are derived from....