How does DNA demo ethnicity?

+3 votes
I'm sorry to ask this, or probably will be sorry I asked, but how does a result from a DNA test show your ethnicity? Where does this judgment come from? What is it that stands there waving a flag that says x % Irish and x % Greek and x % Polish?
in Genealogy Help by Susan Smith G2G6 Pilot (262k points)
So, Peter, Dina, Frank, I could do almost as much to demonstrate my ethnicity with a solid paper trail of various ancestors.

So would you say DNA testing is best for medical considerations and for tracing the mother line(mtDNA) and the father line (Y-DNA)  in terms of lineage, rather than ethnicity? It seems so to me.
The only other thing about DNA that might not clarify is where in the lineage 2 siblings had married another set of siblings from another family and at some point a descendant from each couple then married and had children  

It has been pointed out that the DNA of the descendants from two sets of siblings who married in such a fashion would be quite similar and would not clearly define the biological ancestry. One would see there was a relationship but not see clearly which particular parent/child were biologically tied, because of the similarities of the DNA

An example would be where the imaginary Albert and Ronald GORGE married the imaginary sisters Betty and Mary FULLER.

Albert and Betty had a son Herbert Gorge and Ronald and Mary had a daughter Lily Gorge.

Herbert and Lily did not marry each other but they did each marry and have a child and those two childrent did so, when Herbert's daughter married Lily's son.  

I don't know how often such a situation would arise globally  although I myself have half a dozen or more such cases among relatives "near" and "far"

 I can imagine that in cases of polygamy where one man has married two sisters as may have married several other women, the DNA of the women (mtDNA) would help in sorting out which child had which mother, except in the case of the two sisters he married.

So there seems to be some inherent limitations in the applications of DNA test results -- at least it seems so to me.
Susan, I'd go so far as to say that a "solid paper trail" is far superior than these admixture estimates, and that while the latter is sold as a short cut to the former, that it is no substitute at all.

For genealogy, DNA testing is best used in concert with the "paper trail". The former not only can be used as a confirmation for the latter, but it can also provide leads that reveal branches of your family that might otherwise have been practically impossible to uncover. In my own case, for example, an ancestor had a brother named "William Johnson". He apparently moved out of the area that he was born in, and there were no obituaries found to tell you where he (as a surviving sibling, for example) had gone. But his descendants showed up, plain as day, in my brother's DNA matches.

Of course, for more immediate relatives, you can also expose secrets regarding who your biological relations are, if applicable. It's especially gratifying to help an adoptee find their biological family.

Yes, there are Real World cases that pop up that throw a bit of a monkey wrench into how it works, but those are the exceptions. One wouldn't want to "throw the baby our with the bath water".

A pair of my gt-gt-gt grandparents' descendants are especially "complicated". It seems like it was a family tradition, or something, to marry your second cousin. On one line, I noticed something like 4 generations in a row, marrying distant relations - all descended from the SAME ancestor. It's a mess.
Frank, I can imagine (shudder) what 4 generations of "familial marriage" might "do" ...

Susan, the situation you described is a special case of endogamy.  There are other ways endogamy can arise, and it is always a possible complication in DNA analysis of any sort.  If you look at the collection of your 30-g-grandparents (i.e. look at the 33nd row up in your tree), there are over 4 billion different positions.  There weren't 4 billion people alive back then, so you must see some people (lots of people, in fact) duplicated in many positions in that row.  This duplication is endogamy. If it happens at a level much lower down in your tree, then it gives you less work identifying all the people above that level in your tree but also increases the work in identifying which lines down from that level gave you your different genes.

Susan, sometimes this endogamy stuff doesn't add up to as much as it seems. I have a pair of gt-gt grandparents who were 2nd cousins, for example, and I calculated that my matches to 3rd cousins on that side should only be enhanced by a factor of 33/32 because of it.

I think the record I saw was that somebody had that common ancestor on their tree 5 times. He's way back in a generation where there's 64 pairs of ancestors, so maybe it's not so bad. This guy have astounding longevity, living to 90 in a time where 60 was the norm, and he did it on the frontier. About one in six of my grandfather's cousins on that side lived to at least 89, with some making it to over 100! If you're going to have somebody on your tree 5 times, he's the right guy to have (although you may need to save a lot more for retirement)!
I have a complicated intermarriage farther back in a couple of lines of my family (Curtis and Mann).

A researcher shared: "Kinship report shows the immediate family had close blood relations:  1) Groom and bride: Sgt Wm Benson Curtis 5 (1748-1793) and his wife, Deborah Curtis (1756-?) were 2nd cousins.  2) Groom's mother (Martha Mann 1725-1769) was bride's mother's paternal 1st cousin (Deborah Mann 1721-?) 3) Groom's mother Martha Mann was also bride's father's maternal 1st counsin (Abner Curtis 1730-1799).  4) Groom's mother's (Martha Mann) father (Benjamin Mann Sr 1697-1770) was the brother of bride's (Deborah Curtis) mother's father (Thomas Mann 1681).

"This was a recipe for disaster for the next generation when William Benson Curtis 6 (1777-1851) married Olive Isabella Stubbs (1779-1849) and for of the the twelve children born to this couple were deaf and dumb. 1) George W Curtis (1802-1869). 2) Olive Stubbs Curtis (1807-1881), 3) Ann Curtis (1813-1887), and 4) Ebenezer W. Curtis (1820-1894) (History of Hanover, Genealogies, page 119).  The family appears in the US Special Census of Deaf Family Marriages and Hearing Relatives, 1885-1895."

I figure this accounts for a few issues in my own family possibly.
So, basically, the groom's mom was a 1st cousin to both of the bride's parents - call it a double second cousin. Seems like I've seen worse, with no effect. Sounds like you have a pair of ancestors on your tree three times, WAY back. No biggie.

To get the kind of problem that Wm B & Olivia experienced, though, you'd probably need recessive genes from BOTH Wm B & Olivia. Their problem can be blamed on intramarriage only if Wm B and Olivia were related - if nothing showed with Wm B, then the aforementioned endogamy wasn't the sole problem. Otherwise, it's just bad luck, I should think.

Again, if the problem could be caused by Wm B's  genes alone - a dominant gene - then he would have been effected himself. For the problem to carry forward from just one ancestor it's the same - it's have to be dominant, but it clearly wasn't. These genes must have PART of the recipe for problems, being passed on, but you'd have to run across a spouse with the rest of the recipe for it to have effect. It seems unlikely that it would cause problems in YOUR generation, to me.
Correction! I guess that means you have a couple of instances of having a pair of ancestors in two places on your tree - one instance from Martha Mann's father's side, and another from her mother's. That's even less of a problem, I would think.

6 Answers

+5 votes
Best answer

Here is a four part guide to ethnic admixture at GEDmatch

It is free to upload to GEDmatch and adding your GEDmatch ID to WikiTee automatically integrates the two.

If all your recent ancestry is from the same country then GEDmatch’s Oracle utility is pretty accurate in determining which country that is without knowing your ancestry. See

by Peter Roberts G2G6 Pilot (513k points)
edited by Peter Roberts
It's amusing, and perhaps a reflection of human nature. If you read the introductory paragraph of that article, it says, essentially, the same thing as my answer.

One would thing that, reading that paragraph, that people - having been warned - wouldn't want to go to the time a nd effort to even read on. Yet, it tells them a story, and - even though that story is admittedly a bit hokey - they'll take that in lieu of nothing (or, perhaps, rather than do real work). It's remarkable, how much people get into that stuff, given its severe shortcomings.
+4 votes
It's interpretation from populations that share similar makeup.

Sometimes it is frustratingly incorrect. For about ten years on FTDNA I showed as having around 20 percent Native American ancestry -- which is quite a lot, and totally false. Luckily at the time I knew of the issues between reading West/Central Asian DNA (usually Turkic) as Native American DNA and was not sent on a wild goose chase.
by Dina Grozev G2G6 Mach 4 (47.4k points)
+6 votes
They use various sorts of mathematic witchcraft to attempt to identify genes that were common in variation populations throughout the world. It's not as precise and scientific as they'd like you to think, and even if you could identify exactly which country every cM of your genes came from, the percentages still could not possibly be very accurate, since you get an uneven amount of genes from any ancestor further back than your parents.

In short, it can give you a rough estimate, and tell you pretty precisely what continent our people originated from, but it's no substitute for real genealogy. The real value of a DNA test for genealogy is mostly in your matches with other people..
by Frank Stanley G2G6 Mach 6 (60.8k points)
When you say "the percentages could not possibly be very accurate", you mean they are not an accurate reflection of the distribution of your ancestry through your tree.  They are presumably more accurate a reflection of how your ancestry is distributed through your genes (but still not very accurate).

23andme correctly identified the exact canton in Switzerland where my Swiss great-grandmother Elisa and her and her ancestors lived -- much more precise than just a continent. But it also currently states that I am 33.7% Swiss and French.  I have no recent French ancestry, and it's only a weak match, so presumably the bulk of this percentage comes from my Swiss.  This is far more than the 12.5% expected value from a great-grandparent, so if I eventually get most of my chromosomes mapped I expect to see Elisa overrepresented.
I have also noticed that the more insular a population group was over time the more likely that it will show up as a pinpointed area on these estimates.

As to your 1st point: "Yep."

As to your 2nd point, yes, that's definitely worth bringing up. The thing is, those localized regions it comes up with don't have any percentages associated with them, so - in a way - it's actually a different subject!

I've seen the results of a woman who is 1/8 Italian, and it said she was 0% Italian (it appeared to really be non-zero, but perhaps rounded down to zero). But at the same time, it identified the exact area in Italy that her people were from!

So, yes - and I think this important - those localized regions they give you (the ones specifically that do NOT have a percentage given), in my experience, have been "SCARY-ACCURATE" (as I put it). Basically, anything that DOESN'T have any connection to the admixture is accurate!
+4 votes
DNA ethnicity tests are extremely unreliable. As an example if it says you are Italian what that actual means is that your ancestors could be from any of a dozen locations around that area of the Mediterranean. The DNA testing companies conveniently forget to include that fact. It has little value as a tool to construct your family tree. Stick with the paper trail.
by George Churchill G2G6 Mach 8 (85.1k points)
Italy - Excellent example George.

Another messed up one - Denmark, Every test I have seen lumps it with varying percentages of Norwegian/Swedish/General Scandinavian, Germanic, Northwestern European, and British Isles.... When you look at it its the areas that the Anglo/Saxon/Jutes and later the Vikings migrated to and/or pillaged and plundered.

The tests are picking up on the 30%+/- genetic contribution from them in the other populations and saying a percentage of the DNA is from those areas when its really the other way around. They just do not know how to properly sort it all out at the moment.
Don't trade your haggis in for a bowl of spaghetti.

Haha! laugh

+3 votes
In my opinion it is more marketing hype than real science.

It is based on the size and contents of their database so it is not realistic to all countries as some databases may have more of one country than another.
by Laura Bozzay G2G6 Pilot (614k points)
+1 vote
DNA tests have been a huge help for me in the search for cousins; other than that, not much help.  Ancestry DNA indicates I'm almost 1/3 Norwegian (per paper trail, I'm less than 5%); however, reading the chart, the Norway identification encompasses Norway, most of Sweden, northern Denmark, and a small portion of northwest Finland.  Germanic Europe (on Ancestry) could mean any one or more of the following: England, France, Belgium, Switzerland, the Netherlands, Liechtenstein, Luxembourg, Germany, Denmark, Sweden, the Czech Republic, Austria, Slovakia, Italy, Hungary, Croatia, Slovenia, Poland, and Serbia.
by K. Anonymous G2G6 Pilot (118k points)
edited by K. Anonymous

Related questions

+12 votes
1 answer
+3 votes
2 answers
+14 votes
3 answers
+4 votes
1 answer
99 views asked Oct 4, 2018 in Genealogy Help by Angela Herman G2G6 Mach 1 (15.6k points)
+8 votes
4 answers
+7 votes
1 answer
190 views asked Feb 7, 2017 in Genealogy Help by Gigi Tanksley G2G6 Mach 3 (31.5k points)
+21 votes
4 answers
+10 votes
2 answers

WikiTree  ~  About  ~  Help Help  ~  Search Person Search  ~  Surname:

disclaimer - terms - copyright