* Ancestry DNA recalculations

The place to post news about genealogy products and services that might be of interest to other Family Historian users.
Post Reply
avatar
Mark1834
Famous
Posts: 202
Joined: 27 Oct 2017 19:33
Family Historian: V6.2
Location: South Cheshire

Ancestry DNA recalculations

Post by Mark1834 » 19 Sep 2020 16:31

Anybody else had any surprising results from Ancestry's recent recalculations of UK DNA origins?

Mine hasn't changed significantly (53-62% England & NW Europe, 0-36% Ireland, 8% bar) and feels about right - strongly English with a small amount of Ireland, but I've had one first cousin go from 60% Irish 40% English (which fits their tree well) to 46% Irish, 43% Scottish, and only 5% English just by recalculation! Another first cousin with demonstrably English roots has now also suddenly acquired 33% Scottish ancestry!

I know that these numbers are more entertainment than science, and the DNA providers grossly oversell what they can do, but really.....

I'm half tempted to order a test from LivingDNA for comparison, as they concentrate on the UK and include all three DNA tests. Going direct is much cheaper than via FMP, but anyone have any experience of them and how they compare with Ancestry?
Mark Draper

User avatar
ColeValleyGirl
Megastar
Posts: 1752
Joined: 28 Dec 2005 22:02
Family Historian: V6.2
Location: Cirencester, Gloucestershire
Contact:

Re: Ancestry DNA recalculations

Post by ColeValleyGirl » 19 Sep 2020 16:55

LivingDNA do seem to have a more granular analysis of UK regional origins, but they aren't popular enough to be much good for finding matches yet.

My Ancestry analysis in its current iteration is spot on for me being half welsh but has chucked away half my English ancestry and replaced it with Scottish.

Have you seen https://blogs.ancestry.com/ancestry/202 ... estimates/ ?

avatar
Mark1834
Famous
Posts: 202
Joined: 27 Oct 2017 19:33
Family Historian: V6.2
Location: South Cheshire

Re: Ancestry DNA recalculations

Post by Mark1834 » 19 Sep 2020 17:16

Hadn't seen that, thanks - somebody needs a primer on the difference between precision and accuracy.... :)
Mark Draper

avatar
AnneEast
Superstar
Posts: 302
Joined: 20 Jul 2005 23:39
Family Historian: V6.2
Location: Cumbria

Re: Ancestry DNA recalculations

Post by AnneEast » 19 Sep 2020 18:56

I have pretty much ignored the ethnicity up to now. The new version of it is even worse than before so I shall just continue to ignore it!
They claimed they were now able to distinguish between England, Scotland, Wales and Ireland but they removed all my Irish (which I know I have) and replaced it and added to it a huge chunk of Scottish. My brother, on the other hand has kept all his Irish! I do accept that our Irish ancestors may well have come from Scotland but the % is excessive.
Anne

User avatar
AdrianBruce
Megastar
Posts: 978
Joined: 09 Aug 2003 21:02
Family Historian: V6.2
Location: South Cheshire
Contact:

Re: Ancestry DNA recalculations

Post by AdrianBruce » 19 Sep 2020 20:27

60% Irish 40% English (which fits their tree well)
I believe that this is (at least partly) the issue. Your tree quite possibly has nothing to do with your ethnicity. So far as I can see, based on what Ancestry claim are typical English percentages, the ethnicity that is being measured, is giving us clues to our ancestry back way beyond any of our trees - back into 1066 and earlier. People saying that they've got their trees back to 1750 and it doesn't go outside Norfolk (or whatever) are missing what ethnicity means. Though I doubt they've ever been told this...
Adrian

User avatar
jimlad68
Megastar
Posts: 650
Joined: 18 May 2014 21:01
Family Historian: V6.2
Location: Sheffield, Yorkshire, UK (but from Lancashire)
Contact:

Re: Ancestry DNA recalculations

Post by jimlad68 » 19 Sep 2020 21:11

There is a glaring fault with Ancestry origin calculations, and many others.
They rarely give you a time point, so it is meaningless.

At least a previous "version" of Ancestry origins gave various dates back to I think 1700ish. I suppose that might be "statistically" useful, but that is not of much use re origins of DNA and population migrations. By 1700 many "ethnicities" (or whatever you like to call them) had been very very "mixed up". Even the British Isles of the Roman period contained many from all over the Roman empire and beyond.
To my mind you need to go back at least pre roman, and even then "parts" of your DNA would be "flowing" quickly.

I don't know regarding the paid sites, but using GEDMATCH (Genesis?) there are lots of options (can't say I understand them yet) e.g.

- Archaic Matches, I think shows where bits of your DNA have been found in the deceased thousands of years ago, including Neanderthal and Denisovan.
for stand alone progs see https://fiidau.github.io/Ancient-DNA.html

- Admixture (heritage) I think shows population % for things like North_Atlantic, Baltic, West_Med, West_Asian, Sub-Saharan etc
for help http://genealogical-musings.blogspot.co ... guide.html

I'm sure there are more, hopefully simpler!
Jim Orrell - researching: see - but probably out of date https://gw.geneanet.org/jimlad68

avatar
Mark1834
Famous
Posts: 202
Joined: 27 Oct 2017 19:33
Family Historian: V6.2
Location: South Cheshire

Re: Ancestry DNA recalculations

Post by Mark1834 » 19 Sep 2020 21:31

I think the way Ancestry present their story is very muddled, perhaps deliberately. One minute they’re describing how their estimates are based on reference panels of known trees, and in the next paragraph talking about Roman and Norman invasions! It’s not at all clear to me whether that is just fluff to make us think we can get insight into ancient times, or whether they apply some sort of fudge factor to all the calculations to try to account for large scale historical migrations.
Mark Draper

User avatar
AdrianBruce
Megastar
Posts: 978
Joined: 09 Aug 2003 21:02
Family Historian: V6.2
Location: South Cheshire
Contact:

Re: Ancestry DNA recalculations

Post by AdrianBruce » 20 Sep 2020 17:35

Mark1834 wrote:
19 Sep 2020 21:31
... One minute they’re describing how their estimates are based on reference panels of known trees, and in the next paragraph talking about Roman and Norman invasions! ...
I don't ever remember them talking about "reference panels of known trees". But it's always possible that I'm reading what I think ought to be there.... But the ethnicities are based on references populations (i.e. the people who have submitted their DNA) and where those people live now. Nothing to do with their trees, since many of them haven't even linked to a tree.

As far as I understand it, the first task is to identify "markers" in people's DNA that seem to come particularly from a specific area (hand waving over how big that area might be). Bits that can be identified to people living in a specific area now go forward to working out your ethnicity. (For instance, many of us across the world are probably carrying DNA from the Neanderthals - but that's so generic it's no use to Ancestry's ethnicity).

If you're living in the UK, with long roots there, and you have DNA markers that are found in some English, and in lots of (hand-waving) Swedish, then somehow this seems to get identified as Swedish ethnicity. Now, that could be because the English invaded Sweden or the Swedes invaded England. And I think that's where the interpretation comes in and says that we know that the Swedes invaded the British Isles (as Vikings) but not the other way round (Abba concert-goers aside).

This is the sort of timescale that we're talking about - Vikings, not a timber merchant from the Baltic in the 1750s.

Yes, it's all a bit flakey because with England in particular, there's just a melting pot of ethnicities but, as I've said - the academics have identified bits that can be found only in certain areas even of England. Whether these are individual markers or groups of markers in combination, I have no idea. Whether Ancestry use those analyses, I've no idea.
Adrian

User avatar
trevorrix
Famous
Posts: 161
Joined: 17 Nov 2002 20:27
Family Historian: V6.2
Location: Suffolk, England

Re: Ancestry DNA recalculations

Post by trevorrix » 24 Sep 2020 06:59

Ethnicity predictions and estimates are just that. Some people may receive results that match their expectations - others may not. This science is stiil evolving as more reference populations are added to the databases and the algorithms improve. We are not there yet.

https://dna-explained.com/2018/12/28/et ... yes-really
Trevor Rix

avatar
NickiP
Famous
Posts: 149
Joined: 26 Feb 2013 12:36
Family Historian: V6.2
Location: UK

Re: Ancestry DNA recalculations

Post by NickiP » 28 Sep 2020 01:21

I uploaded my parent's raw DNA to LivingDNA and until they revised their results earlier this year they pretty much matched the paper trail which had been researched extensively over the last 20 years. Unfortunately the last update changed a lot of the results considerably and while I appreciate their could be NPEs or issues with the paper trail, I'm not so convinced with their results now which is a shame because prior to that they were a good option for anyone with predominantly UK ancestry. These are obviously different to the ethnicity results from the other companies which is why they were useful.

User avatar
AdrianBruce
Megastar
Posts: 978
Joined: 09 Aug 2003 21:02
Family Historian: V6.2
Location: South Cheshire
Contact:

Re: Ancestry DNA recalculations

Post by AdrianBruce » 28 Sep 2020 16:03

NickiP wrote:
28 Sep 2020 01:21
... issues with the paper trail, ...
The only issue with your paper trail will be that it doesn't go back 1,000 or 1,500 years to where the ethnicity matches were generated. Ethnicity is not your tree. Ethnicity is back way beyond that to when Angles, Saxons, Vikings entered the British Isles and created the melting pot of DNA we have in England now.... Lots of avoidance of questions in what I just wrote...
Adrian

avatar
Mark1834
Famous
Posts: 202
Joined: 27 Oct 2017 19:33
Family Historian: V6.2
Location: South Cheshire

Re: Ancestry DNA recalculations

Post by Mark1834 » 29 Sep 2020 08:46

I think there is some misunderstanding in this thread on what Ancestry and similar "ethnicity" predictions are saying. It is worth reading the Ancestry White Paper on how they make the estimates (https://www.ancestrycdn.com/dna/static/ ... 0paper.pdf). What they call Ethnicity is based on reference populations with documented trees showing the family essentially stayed in one region over the period documented. Relatively few trees reliably go back more than about 5 generations, so this is looking at the past couple of hundred years, not thousands. Remember that these regions are pretty large buckets, basically entire countries. This gives them typical composition of "French DNA", "Korean DNA", etc.

Mass movement of people through migration, invasion, and the like are complicating factors, as they reduce the differences between these populations. For example, the British Isles are now divided into four distinct regions, but the DNA differences between these are relatively small due to extensive internal movement over thousands of years.

If Ancestry and the rest sell a service that tells a customer that their DNA is 95% from the British Isles, for example, that's not very exciting, as a reasonable response would be "I knew that anyway!" They therefore attempt to make these reference populations as tightly defined geographically as possible. This is where it start to get on slightly shakier ground. I get very nervous when I read things like "outliers were removed to improve the correlation". It's fine to remove outliers if you know why they don't fit your model, but not otherwise.

The more precise they try to make the reference groups geographically, the less accurate the determination becomes. To understand the difference between accuracy and precision, consider the statement that "this posting was made at about half past nine in the morning". That is accurate, but not very precise. If I say it was made at 09:31:27, that is precise, but not very accurate.

I don't know how other DNA companies operate, but Ancestry also give sub-region breakdowns. However, these are based on common ancestors, not DNA analysis. For example, my report emphasizes East Anglia as a possible area of interest. That is not because I have "East Anglian DNA", but their tree matching service reports common ancestors in this part of the country.

My take on all this is that what Ancestry misleadingly call an "ethnicity prediction" should broadly correlate with my paper trail. All it is saying is that my DNA looks most like that of people who lived in that area over the past couple of centuries or so, and actually says very little about the millennia timescale, other than through the high uncertainty caused by population mixing. But that correlation is only at a relatively high level - "95% British Isles" is defendable from the science, but a more precise breakdown between English/NW Europe, Scotland, Wales, and Ireland is little more than educated guesswork with huge uncertainty.

Extrapolate this to a company like FMP/LivingDNA who specialize in breaking the UK down to even smaller regions, and you inevitably end up with more precision, but less accuracy. No matter how large and UK-specific the reference panel becomes, it is always fighting the fact that regional DNA differences in a small country like the UK with extensive population movement are tiny and very difficult to resolve.
Mark Draper

User avatar
jimlad68
Megastar
Posts: 650
Joined: 18 May 2014 21:01
Family Historian: V6.2
Location: Sheffield, Yorkshire, UK (but from Lancashire)
Contact:

Re: Ancestry DNA recalculations

Post by jimlad68 » 29 Sep 2020 11:25

Mark, yes nicely put, confirming my points above.

From a "building your tree" perspective, their statistical results may be helpful. What I cannot understand is why they do not also make use of all the "ancient" DNA samples out there and give places/migration routes in historic+prehistoric timelines".
Jim Orrell - researching: see - but probably out of date https://gw.geneanet.org/jimlad68

User avatar
AdrianBruce
Megastar
Posts: 978
Joined: 09 Aug 2003 21:02
Family Historian: V6.2
Location: South Cheshire
Contact:

Re: Ancestry DNA recalculations

Post by AdrianBruce » 29 Sep 2020 15:13

Well, I actually find no confirmation in that White Paper about how far back the reference samples' trees need to go back in one area. Lots of pious hopes but... For instance:
When asked to trace familial origins, most people can only reliably go back one to five generations, making it difficult to find individuals with knowledge about more distant ancestry. ... In the recent past, it was much more difficult and thus less common for people to migrate large distances. Because of this, the birthplace of a person’s recent ancestors often represents the location of that person’s deeper ancestral DNA
This describes a lot of statements, with which I wouldn't argue. But I've still no idea from this bit - or later bits - what the qualification is to go into a reference panel. For ease of discussion, let's say it's equivalent to 200y.

Now, one of the Ancestry blogs said that, for people living in England, most people are (ethnically) a little bit Scottish. The question then becomes - where does that Scottish ethnicity come from? It's matching reference population people currently living in Scotland but surely we're talking common ancestors somewhere along the line...

The immediate result of this "everyone's a wee bit Scottish" was that people living in (say) East Anglia whose trees went back 300y solely in East Anglia, started say that this is all nonsense. Their trees, note, go back beyond my (arbitrary guess at) the reference panel qualification so surely they ought to match the reference panel for their area - i.e. they ought to be wholly English. And, remember, Ancestry are saying most English people are a little bit Scottish, so it's not just the usual statistical variation if most English are a bit Scottish.

There are several ways that I could attempt to make sense of most English being a bit Scottish:
  • The numbers are pointless. Well, perhaps, but I'm trying to be polite and assume that there is at least something in it;
  • The reference populations are meaningless because they're not long standing families after all;
  • The Scots migrated through what is now England on their way to Scotland (true for the Saxons and the Strathclyde Welsh), leaving ancestors behind, but somehow their DNA characteristics are now found mostly in Scotland not England;
  • Various Scots moved south into England over the centuries and their DNA characteristics survive in their descendants who now cover most of England. Crucially, this happens before most people's trees start in any comprehensive manner.
My own guess is that the last bullet point there explains most of the "wee bit Scottish". And crucially, on that basis I would not expect the percentages to match my paper trail. I fear this is where Mark and I differ.

If my ethnicity is to match my paper trail, then I'm back looking - not for Scottish, because G-GF came from Dundee - but for 4% Norwegian, 2% Swedish and 3% Germanic Europe - and I'm looking for it in my paper trail! Really? Surely I'm much more likely to be looking at (say) Viking remnants from those areas that might have wandered over when Vikings did their wandering. (Come to that, it might, of course, be via my Irish ancestry of 6.25% on paper but which otherwise is almost invisible at 2%)

So, for me, the only sensible interpretation of the ethnicity numbers - especially the "wee bit Scottish" for so many - is to look back beyond my paper.

And if anyone wants to say that this is all very tricky to get these fine gradations in as mixed up an area as England, well, I'll not argue but I am trying to come up with a reasonable explanation, not a plausible denial.
Adrian

User avatar
AdrianBruce
Megastar
Posts: 978
Joined: 09 Aug 2003 21:02
Family Historian: V6.2
Location: South Cheshire
Contact:

Re: Ancestry DNA recalculations

Post by AdrianBruce » 29 Sep 2020 15:15

jimlad68 wrote:
29 Sep 2020 11:25
... What I cannot understand is why they do not also make use of all the "ancient" DNA samples out there ...
I strongly suspect that the tools and methodologies are incompatible.
Adrian

avatar
Mark1834
Famous
Posts: 202
Joined: 27 Oct 2017 19:33
Family Historian: V6.2
Location: South Cheshire

Re: Ancestry DNA recalculations

Post by Mark1834 » 29 Sep 2020 17:36

I'm not sure we do disagree, but I do think many people try to over-interpret what the DNA results are telling them. If I step back and take a look at my data from a global perspective, it tells me that I am roughly 97-100% British Isles, 0-3% German/Swedish and no discernible contribution from any of the other 74 regions that Ancestry characterise. That's what I mean by broadly correlating with the paper trail, and on a global scale, that's a pretty good correlation. Does the 0-3% mean that I definitely have a German or Swedish gggg-grandparent? Of course it doesn't, that level of detail is just sales hype. Does that mean that DNA results are actually of limited use in determining exactly where on this small island my gg-grandparents lived? I couldn't possibly comment ;).

I think the "wee bit of Scottish" in most "English" results is one of two things (probably a bit of both) - firstly, the shared ancestry over the millennia scale, and secondly, there is no such thing as pure English and Scottish DNA. There will be natural variations in DNA even within an isolated population, and if this natural variation is similar in scale to the typical difference between two isolated populations, it will be virtually impossible to assign an individual to one group or the other unambiguously. Current technology can resolve between the British Isles and Eastern Europe (for example), but not reliably within the British Isles.
Mark Draper

User avatar
jimlad68
Megastar
Posts: 650
Joined: 18 May 2014 21:01
Family Historian: V6.2
Location: Sheffield, Yorkshire, UK (but from Lancashire)
Contact:

Re: Ancestry DNA recalculations

Post by jimlad68 » 29 Sep 2020 17:54

Adrian,
I strongly suspect that the tools and methodologies are incompatible.
If a free site like Gedmatch can do it, and there are many very good amateurs out there that can do it, and some also paid for sites; for Ancestry it should be easy. Could the reason be that for Ancestry their business model is selling data from the last few hundred years, not 1000s of years.
Jim Orrell - researching: see - but probably out of date https://gw.geneanet.org/jimlad68

User avatar
AdrianBruce
Megastar
Posts: 978
Joined: 09 Aug 2003 21:02
Family Historian: V6.2
Location: South Cheshire
Contact:

Re: Ancestry DNA recalculations

Post by AdrianBruce » 29 Sep 2020 19:25

Mark1834 wrote:
29 Sep 2020 17:36
I'm not sure we do disagree, but I do think many people try to over-interpret what the DNA results are telling them. If I step back and take a look at my data from a global perspective, it tells me that I am roughly 97-100% British Isles, 0-3% German/Swedish and no discernible contribution from any of the other 74 regions that Ancestry characterise. That's what I mean by broadly correlating with the paper trail, and on a global scale, that's a pretty good correlation. ...
OK - I can agree with everything that you write there...
Adrian

User avatar
AdrianBruce
Megastar
Posts: 978
Joined: 09 Aug 2003 21:02
Family Historian: V6.2
Location: South Cheshire
Contact:

Re: Ancestry DNA recalculations

Post by AdrianBruce » 29 Sep 2020 20:23

jimlad68 wrote:
29 Sep 2020 17:54
... If a free site like Gedmatch can do it, ...
D'uh, Jim. You are, of course, correct. For some reason, I had fixed in my head the ancient DNA results coming to Ancestry to be interpreted by their current algorithms looking at the very specific (at least, I think they are) areas that Ancestry look at.

Whereas, as GEDMATCH shows, the logical equivalent of the reverse direction is perfectly possible. Logical equivalent, not necessarily physical equivalent. I do wonder about all sorts of legal aspects - Ancestry couldn't transmit data to GEDMATCH because (a) it would drown them in data volume but also (b) it would almost certainly send DNA stuff out of the legal jurisdictions in which it should be kept.

Which would suggest building the equivalent analysis tools within Ancestry's data centres, etc. And I can see all sorts of similar issues might crop up in that direction! Academics co-operating with big business? :o Well, cross their palm with enough silver and they'd say yes - but I'm actually unclear how authoritative those ancient analyses on GEDMATCH are... Are they supported by major institutes? Or is it just commited (professional) enthusiasts carrying out their own analyses?

Since this stuff is just a fun aside for me, I've not put effort into trying to understand the Ancient DNA analyses on GEDMATCH but, for instance, the Eurogenes K13 Admixture Proportions analysis under Admixture (heritage) gives me 50% North Atlantic, 24% Baltic (there's that Norwegian / Swedish again! Is it?) and so on down to 1.6% South Asian, 1.22% Amerindian, 0.99% Oceanian... If we struggle with being "a wee bit Scottish" - just what does 1.22% Amerindian mean for the ordinary Brit?

I also had a look at Archaic DNA Matches on GEDMATCH. I have even more difficulty with this - as there's more detail. I seem to have a good solid match to "LBK Stuttgart" (7k years) but also a goodish match to "Ust-Ishim, Siberia" 45k years. But there seems to be a blank right down the middle of Chromosomes 1 & 9 where I match nothing. (Maybe that's my Vulcan heritage that led to my logical ability with maths and computer programming? ;) )

But, but, but, reinforcing my concerns over the authority of this stuff, I find this quote on a FamilyTreeDNA page
Ancient samples at GedMatch have the threshold set so low as to be meaningless. Some individuals do match ancient samples today at Family Tree DNA
(My emphasis....) So maybe we shouldn't imagine everything is crystal clear elsewhere either.....
Adrian

User avatar
jimlad68
Megastar
Posts: 650
Joined: 18 May 2014 21:01
Family Historian: V6.2
Location: Sheffield, Yorkshire, UK (but from Lancashire)
Contact:

Re: Ancestry DNA recalculations

Post by jimlad68 » 29 Sep 2020 21:26

Adrian, my point is time and place. For instance, my Y-dna shows most similar to some people that were found in Sweden ~2000BC, 22000 years ago Siberia, and my autosomnal dna the usual 1-2% Neanderthal. But this relies on samples of DNA from those times, the more, the "statistically" more likely.
I've been everywhere, man, along with the rest of you.
Not important, but fascinating.
Jim Orrell - researching: see - but probably out of date https://gw.geneanet.org/jimlad68

avatar
Mark1834
Famous
Posts: 202
Joined: 27 Oct 2017 19:33
Family Historian: V6.2
Location: South Cheshire

Re: Ancestry DNA recalculations

Post by Mark1834 » 30 Sep 2020 09:33

Purely out of curiosity, I paid the £29 to LivingDNA to analyse my Ancestry DNA according to their more granular UK regional breakdown. The results came back within a couple of hours and were very much a mixed bag. The detailed numbers were way off what the paper trail would indicate (over 50% assigned to regions where I have no little or no known connection), but if I ignore the actual numbers and just overlay the birth locations of 14 of my 16 gg-grandparents onto the LivingDNA map of the regions identified, the correlation is reasonable. The 15th was Ireland (which was listed), and the 16th unknown but probably Staffordshire.
Capture.PNG
Capture.PNG (43.8 KiB) Viewed 951 times
I think the takeaway message is the same as for Ancestry. If you just look at the broad picture rather than the detail, it can at least give you a rough indication of where your roots lie, and it is less likely that there are recent ancestors outside the broad regions highlighted.
Mark Draper

avatar
NickiP
Famous
Posts: 149
Joined: 26 Feb 2013 12:36
Family Historian: V6.2
Location: UK

Re: Ancestry DNA recalculations

Post by NickiP » 04 Oct 2020 12:32

AdrianBruce wrote:
28 Sep 2020 16:03
NickiP wrote:
28 Sep 2020 01:21
... issues with the paper trail, ...
The only issue with your paper trail will be that it doesn't go back 1,000 or 1,500 years to where the ethnicity matches were generated. Ethnicity is not your tree. Ethnicity is back way beyond that to when Angles, Saxons, Vikings entered the British Isles and created the melting pot of DNA we have in England now.... Lots of avoidance of questions in what I just wrote...
Yes I accept that but unlike Ancestry, LivingDNA is based, if memory serves me right, to the Population of the British Isles research project where the participants had four grandparents living in a specific area. The use of this database initially was supposed to give more accurate results and prior to this year's it appeared to. I believe they've also used the same methodology for the other countries they include. However, I'm wondering if they've started expanding their sample database outside this methodology and perhaps this explains the change in results. The restricted nature of the sample database really was the appeal of LivingDNA over the other companies. Even now their matching database isn't very large so if they're just expanding their sample database for ethnicity style results to use just anyone, there isn't really much of a reason to use them.

User avatar
AdrianBruce
Megastar
Posts: 978
Joined: 09 Aug 2003 21:02
Family Historian: V6.2
Location: South Cheshire
Contact:

Re: Ancestry DNA recalculations

Post by AdrianBruce » 04 Oct 2020 18:09

NickiP wrote:
04 Oct 2020 12:32
... unlike Ancestry, LivingDNA is based, if memory serves me right, to the Population of the British Isles research project where the participants had four grandparents living in a specific area. The use of this database initially was supposed to give more accurate results and prior to this year's it appeared to. ... However, I'm wondering if they've started expanding their sample database outside this methodology and perhaps this explains the change in results. ...
I suspect that will very much depend on the areas in question. Being born in the railway town of Crewe, I've never had much faith in the four GPs idea. If you take that idea, I should be, roughly speaking, 100% North-West England. Yet my paper trail is 12.5% Scottish, 6.25% Irish, to take the broad brush categories. If you were to look at tighter areas, something immediately becomes apparent when you compare my mother's qualification under 4GPs - which she passed - to my father's where he had just one GP from Cheshire, one from Central Lancashire, one from Bristol and one from Scotland.

I think I'm right in saying that the Population of the British Isles research project was aiming to say things about groups of people in an area. But I suspect it all goes a bit flakey once you start using the area level data to look at individuals.
Adrian

Post Reply