Author Topic: Impressed with ethnicity breakdowns  (Read 409 times)

Offline brigidmac

  • RootsChat Marquessate
  • *******
  • Posts: 6,014
  • Computer incompetent but stiil trying
    • View Profile
Impressed with ethnicity breakdowns
« on: Tuesday 11 April 23 22:47 BST (UK) »
Recently looked at ethnicity breakdown of all the people whose  DNA I manage on ancestry

Usually I only look at ethnicity of shared matches but I see that the divisions are getting more and more accurate
And there is a way of comparing ethnicity of a group of people

6 of the people whose DNA I manage are relations so great to look at the shared Scottish in my father's side .

I'm lucky to have such a wide data base which helps understand others results

I can  see matches my mother's 2nd cousin thru mutual Latvian great grandfather her ethnicity is 99percent Jewish and 1 percent Welsh !

A descendant of Welsh great grandfather didn't know his bio father but we've narrowed it down to 2 brothers ( ethnicity wasn't a huge help on that one ) managing his DNA

Nephew s DNA shows how ethnicity amounts can be passed thru 3 generations

From my cousin's I got to look at African ethnicity from his other side

My in law is very Essex with a bit of Scandinavian from both parents
The more I learn to use the information that is there the more I respect it

Tho I do have 2 gripes with ancestry 1 is that theyve grouped channel Islands with isle of man with Cornwall

+ I'm still making sense of Haiti and Benin Togo classification
The DNA may make sense but when searching for documents
The classification for all black people  in search bar seems to be African American

The Bahamas was not American in 1850s and
But that's how to find the records

Oops that's not the pic I meant to post

 
Roberts,Fellman.Macdermid smith jones,Bloch,Irvine,Hallis Stevenson

Offline Ruskie

  • RootsChat Marquessate
  • *******
  • Posts: 26,198
  • Census information Crown Copyright, from www.nationalarchives.gov.uk
    • View Profile
Re: Impressed with ethnicity breakdowns
« Reply #1 on: Tuesday 11 April 23 23:18 BST (UK) »
Sample populations in some of those areas might be small, hence inaccuracies, and a reason (alongside not enough differences to seperate certain populations dna wise), for grouping some in with others. Maybe close enough is good enough.

Reference panel numbers here:
https://support.ancestry.com/s/article/AncestryDNA-Reference-Panel?language=en_US
(I know some areas have low numbers but these look very low - so unsure if it is the latest table. Google will tell you.)

Also remember that these numbers consist of populations who claim to have X generations in a particular location. With such low numbers in some areas presumably it would only take a few to be wrong to skew the sample. These samples are taken from people living in a region today (or having ancestors from that region for generations) - the sample populations would have been very different a hundred or two years ago.

And we all know that different DNA tests give different ethnic estimate locations and percentages. People probably tend to believe whichever seems to represent their particular family mre accurately. Doesn’t mean it is so for all.

Claims vary depending on what you read but they say autosomal DNA only takes you back five to eight generations, so ….. earlier generations came from where?

This might be of interest:
https://support.ancestry.com.au/s/article/Key-Differences-Between-Ethnicity-and-Communities

It is advisable not to take too much notice of the percentages which are guesstimates and subject to change.

Added:
I note your example above says estimated 4% but clarifies that it could be 0 to 11%.  :-\

Offline brigidmac

  • RootsChat Marquessate
  • *******
  • Posts: 6,014
  • Computer incompetent but stiil trying
    • View Profile
Re: Impressed with ethnicity breakdowns
« Reply #2 on: Tuesday 11 April 23 23:33 BST (UK) »
im not sure i understand what you are saying Ruskie

i didnt mean to post that image .to accompany this post which is celebrating how ethnicity can help

i KNOW tht 4% can be insignificant but in that case + as  a combination of 2 groups with a known African grandparent it was very significant

i cant delete the pic but the person concerned has 19 percent Togo + Benin + ALL their matches with African origin have those countries have combinations of those 2 some also have Nigerian Camaroon Senegal Scottish in varied proportions .

Some of the matches  live in Ghana + or have Ghanain names on their trees have 100percent combination of just those 2 ethnicities with Togo/BENIN much higher proportion

american and jamaican matches cant be based on wher people trees say they are from ..most do not know why they have a percentage of Togo /benin + Ivory coast /ghana

i attended an interesting talk by Turi King about a surname search  in yorkshire many had a  very small percent of African DNA (cant remember which region )that could be traced back to a settlement of black soldiers from Roman times ( my cousin has 2 percent congo from his paternal yorkshire  side )

some people are quick to dismiss unexpected + small percentages as a mistake

its as likely to be unexplained or possibly a misclassifeid ethnicity .

 many of my matches well established in  america ,south africa + australia as far back as they can trace   are 100 percent european jewish obviously descend from at least 1 lithuanian latvian jewish person +i can usually tell from shared matches which of my 3xggparents they link to
 even if they are distant matches

my nephew has only inherited 4% jewish
 my cousins + I have between 12%--14 % and my mother 25% which is  mathematically standard amount from 1 grandparent

DNA ethnicity  is easier to identify when there are close communities

hope im making sense its past bedtime I will look at the  link tomorrow

Roberts,Fellman.Macdermid smith jones,Bloch,Irvine,Hallis Stevenson

Offline brigidmac

  • RootsChat Marquessate
  • *******
  • Posts: 6,014
  • Computer incompetent but stiil trying
    • View Profile
Re: Impressed with ethnicity breakdowns
« Reply #3 on: Wednesday 12 April 23 06:34 BST (UK) »
My late aunt's ethnicity is pure Scottish on paternal side
If I find a match that has very little Scottish but matches people on that side
They can look for connections in her tree to a particular ancestor if they know which of theirs is Scottish.
It's been particularly useful for people with single parents NPE or adoptees in their lines.

Now I know how to compare more specific communities it will be even easier
Maybe we.ll finally work out the connection between a Scottish miner and a Canadian lumberjack s family .

Lowland Scotland is separate ethnic community to highland

I love the moving image that shows emigration routes

Roberts,Fellman.Macdermid smith jones,Bloch,Irvine,Hallis Stevenson


Offline phil57

  • RootsChat Veteran
  • *****
  • Posts: 644
  • Census information Crown Copyright, from www.nationalarchives.gov.uk
    • View Profile
Re: Impressed with ethnicity breakdowns
« Reply #4 on: Wednesday 12 April 23 17:10 BST (UK) »
You are still confusing Ethnicity estimates and Ancestry Communities Bridget.

Ethnicity estimates are probability ranges of a link to a particular ethnicity region around 500 years ago or greater. They are determined by DNA analysis of segments that show consistency with Ancestry's reference panel of individuals that they have allocated to that region. As Ruskie says, many of the reference panels consist of very few individuals, so the possibility for error can be large (hence to some degree the range of probability which the single "headline" percentage does not express clearly).

But the communities are a completely different association, derived in a totally different way, as I explained in our PMs, and which is also explained in the link that Ruskie quoted in her post. The only DNA link to communities is that autosomal DNA matches are used to identify individuals who are related by family and/or who have shared matches with others. The public family trees of the resulting set of individuals are then trawled by Ancestry for particular locations (or communities) that are common to those individuals at specific points in time within the last 50 to 300 years.

So the communities aren't using DNA to predict anything, other than to select the group of individuals whose trees are searched for commonality of locations. The results rely heavily on the facts recorded in related public family trees.

Ethnicity estimations are based on identification of DNA segments that Ancestry believes identify populations that would have been in common to a specific region more than 500 years ago. So they are based on DNA. But Scottish ethnicity doesn't necessarily mean that ancestors of people attributed with that ethnicity actually lived in Scotland. Although Ancestry have chosen to give it that name, it identifies lengths of DNA that they believe are common to many people now found in Scotland and other areas of the UK with what might be better described as Celtic origins. Those people migrated to the British Isles from the European continent via various routes. Although many of them settled in Scotland and the other regions that Ancestry's Scottish ethnicity encompasses, some settled in other parts of Britain and never went anywhere near Scotland. So it is quite possible to have Scottish ethnicity as defined by Ancestry, without any direct ancestral lineage to Scotland itself.

My paternal father's family have around 37% Scottish ethnicity according to Ancestry, with no evidence of any connection to Scotland the last 350 years (other than a single Scot who married into the family) and indications that they probably arrived in Britain from NW France several hundred years earlier.

When you say that allocated communities can't be based on where people say they are from, because the individuals you know aren't aware of the connection and it isn't in their trees; it doesn't have to be. Much of the location information would have been obtained from other public trees of individuals that have autosomal DNA matched to "your" individuals, and the trees of shared matches between them.

Which brings me back to the USA settler communities we discussed, that my brother and I are, according to Ancestry, a part of. But of which our only connection to is through the half cousins descended from our GGF and a woman with whom he fathered children before he married our GGM - so no direct line of descent between them and us at all.

 
Stokes - London and Essex
Hodges - Somerset
Murden - Notts
Humphries/Humphreys from Montgomeryshire