Ancestry 2018 Ethnicity Update

When ethnicity estimates were first produced by vendors, they tended to resemble the wild west.

Today, results are becoming more refined and hopefully, more accurate as reference populations grow and become more reliable.

The Ancestry ethnicity update has been in beta for several months, but this week, Ancestry rolled out the ethnicity update for everyone.

Checking Your New Results

To see your updated results, sign on and click on the DNA Story to the left with Ethnicity Estimates.

Ancestry then explains that while your DNA doesn’t change, the estimates (pay attention to that word) do as the science improves.

Ethnicity Estimate Aren’t Precise

I’ve said this before, and I want to say it again. Ethnicity is the least precise and the least accurate of DNA tools for genetic genealogy. Ethnicity estimates are the most accurate at a continental level. Within continents, like Europe, Asia and Africa, there has been a lot of population movement and intermixing over time making the term “ethnicity” almost meaningless.

I know, I know – ethnicity estimates are also the simplest because there isn’t much learning curve and they’re easy to understand at a glance. This deceptive “ease of use” also makes them interesting to people who have only a passing curiosity. That’s why they attract so many test takers who either love of hate their results, but never fully understand the true message or utilize any other genetic genealogy tools.

Let’s take a look at how ethnicity estimates have changed over time and if they have improved with the latest version.

Ethnicity Estimate Changes

In my case, my original Ancestry ethnicity estimate in 2012 was:

  • British Isles 80%
  • Scandinavia 12%
  • Uncertain 8%

To say it was really bad is an understatement.

In 2013, Ancestry introduced their ethnicity V2 version which provided a lot more granularity.

Version 2 was dramatically different, with the British Isles moving from 80% to a total of 6%. Like a pendulum swinging, neither was accurate.

Ancestry introduced new features and combined their Genetic Communities with their ethnicity estimates in 2017.

In this new 2018 version, Ancestry has divided and recombined the British Isles and Western Europe differently and the resulting differences are significant.

My mystery Scandinavian is entirely gone now, but sadly, so is my Native American.

The New Results

I just got really boring – but the question is whether or not the new results are more accurate as compared to my proven genealogy. Boring doesn’t matter. Accuracy does.

Various Ancestry Ethnicity Versions Compared to Proven Genealogy

I created a chart that reflects the three Ancestry ethnicity versions as compared to my proven genealogy.

For the current version, I also included the ranges as provided by Ancestry.

As you can see, generally, the results are much more accurate, but the regions are also fairly broad which makes accuracy easier to achieve.

Until this current version, Ancestry didn’t show any Germanic, but now the Germanic estimate is exact at 25%.  The Germanic range is also very tight at 24-26%, right where it should be.

The England, Wales & Northeast Europe category is somewhat high, but that could be accurate because I do have some ancestry that is unknown.

Unfortunately, my Native is proven, both through Y and mtDNA and by triangulating the Native segments to others descending from the same Native ancestors. That portion is now missing in my Ancestry ethnicity.

Ancestry V1 Test Versus the V2 Test

For the record, I’m using my Ancestry V1 test because I’ve used that test version for all previous ethnicity comparisons.  My Ancestry V2 test ethnicity results are approximately the same, as follows:

  • England, Wales and Northeast Europe – 76%
  • Germanic – 22%
  • Ireland and Scotland – 2%

The same tree is attached to both tests.

On my V2 test, which I seldom use, I had to answer a couple of question regarding my expectations about ethnicity testing changes and how accurate my previous results were perceived to be before I could access my updated results.

Regions Changed

In Ancestry’s FAQ, they provided this list of how the regions were and are defined.

Previous Region New Regions
Scandinavia Norway, Sweden
Iberian Peninsula Spain, Portugal, Basque
Europe South Italy, Greece and the Balkans, Sardinia
Europe East Baltic States, Eastern Europe and Russia
Caucasus Turkey and the Caucasus, Iran/Persia
Europe West Germanic Europe, France
Native American Native America—North, Central, South; Native America—Andean
Asia South Southern Asia, Western and Central India, Balochistan, Burusho
Asia East Japan, Korea and Northern China, China, Southeast Asia—Dai (Tai), Southeast Asia—Vietnam, Philippines

Ancestry has addressed lots of other questions in their FAQ as well, and I suggest taking a look. I particularly like their comment, “Some places are complicated.” Indeed, that’s true with population churn both in historical times along with unknown pre-history and that complexity is exactly what makes intra-continental ethnicity estimates so difficult. Of course, people whose ancestors are from Europe, for example, want as much granularity as possible.

Previous Ethnicity Versions

For the first time, Ancestry explains what happened between versions, at least at a high level.

Click on the little “i” in the upper right hand corner of your ethnicity estimate box.

You’ll see more information.

Click on “View Previous Estimate” at the bottom.

Your previous ethnicity estimate is shown.

To see how your estimate changed, click on “Compare these results to your most recent Ancestry DNA estimate.”

This display shows you the differences compared to the previous version. In my case, England, Wales and NE Europe increased by 69%, but that’s because Ancestry redefined the regions. Note the little slide box underneath the regions on the map. You can slide back and forth from previous to current (update.).

I do wish Ancestry had told us where the “Scandinavian” went, what category it fell into. Are those segments, as a group, included in another region? Was the previous estimate simply flat out wrong? Was Scandinavian a vestige of Vikings who invaded much of Europe? What happened?

New Regions and Reference Samples

By clicking on “See other regions tested” at the bottom of your Ethnicity Estimate box, you can view the locations of Ancestry’s current reference populations.

The regions tested in which you have results are colored, and the regions where you aren’t showing results are shades of grey. This is an improvement over the previous version which people routinely misinterpreted to mean that they had results in those tested regions.

Best Features

In my opinion, the best feature of the combined ethnicity and Genetic Communities is the combined mapping. For example, the screenshot below combines the ethnicity regions with the ancestors from my tree who immigrated from that region in that timeframe.

By clicking on the 1700 box, the people from that time period in my tree are displayed. I can enlarge the map to make the display larger, until finally individual “people” icons are displayed, as shown with Johann Peter Koehler, below. Clicking on the individual person pin shows that individual in the box at right.

By clicking on the “Lower Midwest and Virginia Settlers,” I see this region and Ancestry tells me where those settlers likely originated.

You can then scroll down to the bottom of the information box where you see “Ancestry DNA Members.”

Click on the 1000+ link and you will then see the people who match you in a specific region or migration.

It’s worth noting that this isn’t always accurate. My 2nd cousin match is showing as a “Lower Midwest and Virginia” match and our ancestors came from the Netherlands directly to Northern Indiana. Ironically, she shows up in three of the 4 regions I can select from. This feature is not 100%, but it’s still nice to be able to see where that match is grouped in terms of ethnicity and Genetic Communities, according to Ancestry.

Given this combined functionality, I do wonder if Ancestry’s new ethnicity isn’t simply population genetics, but a combination of population genetics, ancestors in my tree, my matches and corresponding DNA Circles with their associated history. If so, that would make sense, both in terms of what I’m seeing as my new ethnicity results and the map functionality as well. Could that be where my Germanic came from, and why it’s so precise at 25% which matches by tree exactly?

In Summary

For me, Ancestry’s ethnicity estimates are significantly improved with the exception that my Native disappeared. I’ve worked long and hard on the Native aspect of my genealogy, and I know that part of my ethnicity mix is valid. However, that is a very small percentage overall (about 2%), and the combined improvements certainly outshine that one negative.

Of course, your mileage may vary. What are you seeing in terms of your new ethnicity estimates as compared to your known genealogy? Better? Worse? Did you lose any categories that you know are valid? What about small amounts of minority heritage?



I receive a small contribution when you click on some of the links to vendors in my articles. This does NOT increase the price you pay but helps me to keep the lights on and this informational blog free for everyone. Please click on the links in the articles or to the vendors below if you are purchasing products or DNA testing.

Thank you so much.

DNA Purchases and Free Transfers

Genealogy Services

Genealogy Research

170 thoughts on “Ancestry 2018 Ethnicity Update

  1. Mine was huge. I went from 60 percent western European to only 3 percent. My great grandfather could only speak german,and my other great grandmother had 2 German names.

  2. I can’t believe the results from my new updated DNA estimate!! The results are ridiculous. They show that I’m supposed to now be 46% French even though I have NO ancestors nor family from France. My Spanish ancestry went down to 9% even though 2 of my grandparents are from Spain. It shows 2% Native American Andean even though we have NO ancestors nor family from Peru.

    On the other hand, my 23andme results have always been very accurate from what we know of our extensive ancestry records.

    • As far as I’m able to tell, Native American -Andean doesn’t necessarily mean you have ancestors from the Andean Mountains, it means your Dna matches their sample population, I know I have Native American ancestry, Cherokee and I received Andean also, for these new results are more accurate, I only showed Central Asia on the previous one, and every test I’ve taken showed Native except the old ancestry results.

  3. I had commented the other day about my having lost my interesting ethnicity report, and since then I’ve studied my husband’s updated results. They are a lot more interesting than mine! His Germanic Europe from the old results has been refined from Europe West from 52% to 6%, his England, Wales and NW Europe has increased by 49% to 55%, and Ireland and Scotland increased by 6% to 37%. His 4% Scandinavian has now turned into 1% Sweden. And he had 2% Europe South before, although no Iberian, but now he has 1% Portugal. We don’t have any evidence at this point for any country of birth other than Scotland and England, and I do wonder if we will ever find evidence that an ancestor of his was born in Sweden or Portugal. I wonder how far back these two areas of 1% might go?

  4. I had Iberian peninsula as a trace region in my original version, which made sense as a relative discovered my ggg grandparents were Spanish immigrants. Also, I had Scandinavian which also made sense as I have Scottish heritage and norse gaelic surnames in my family. Also I had central asian which I thought was fascinating. All of these have disappeared now and with the updated version I am 100% Irish and Scottish. I found the update to be a bit disappointing and not really reflective about what I know about my family history..

  5. We can’t let this go, Roberta!! I know my folks have always told me I’m Scotch-Irish: I’m red-headed with green/hazel eyes. And yet with ancestryDNA, I’ve now got only 19% Ireland and Scotland on ancestry’s new DNA Story, with the rest (81%) England, Wales, and NW Europe. But just today I received an email from 23andMe, “13 New Populations Added
    to Ancestry Composition Report.” So I went over there to look, and they have got me with NA and Sub-Saharan African. Oh my goodness! But tiny amounts, like: .2% Congolese, .2% Nigerian, .1% Southern East Africa. Native American .2%. They aren’t saying whether those estimates are speculative, or if they are, to what extent. But it’s working, if it’s marketing. I think many of us actually do want to have interesting ancestors! Wow, do you think the 23andMe update is a response to the ancestryDNA updates??

  6. As Roberta said “Your mileage may vary”. I heard a radio commercial from an American car producer a few years ago that actually stated in the sales promotion of their product “your mileage WILL vary” (my emphasis added). At least they were being honest. Maybe no one noticed that change of one little word, as I did?

    One thing that all genealogists must pay close attention to, be they of an experienced or of an amateur level, is the time frame involved in their search. Did your sought ancestral event happen recently, such as a generation or two ago? Or a century ago? Or several centuries ago? Or a millennium ago? Or several thousand years ago?

    Time matters, and someone’s location at a point in time also does. So does anyone’s ethnic heritage pertain to them – although some try to ignore or deny it; and some tend to hype or be boastful of it as being of an overriding or superior significance. Either extreme ignores the entire picture: that of being unable to see the forest because of the trees, or an unwillingness to admit it. People often see only what they want to see and ignore whatever they find discomforting. It’s selective editing. OTOH, estimates and predictions are only of limited value in that they are probable, but not absolute.

    Genealogy is family history. History is everything that has happened up until this point in time – including your place in time and your being here.

          • The update seemed to be valid for me and my parents, but I also uploaded my raw data (just mine) to MyHeritage and it was way off. Totally sideways. I decided to give Gedmatch a try again but I don’t know what calculators to use.

          • My inquiry was relating to the phrase “reference populations”. If by that is meant the calculations are being made from the customer base of a genetics website, then obviously it will vary from site to site.

          • Reference populations are made up of different things. Some are public, like the 1000 Genomes Project. Adding on to that, each company uses their own customers with ancestry from specific areas. So yes, the reference populations are not, in total, the same at any vendor. Neither are the analysis algorithms.

        • I don’t know about, but FTDNA uses DNA from several sources (including their customer database, 1000 Genome, Human Genome Diversity Project, and others) to construct reference populations (eliminating related individuals and using principal component analysis to remove outliers). As FTDNA’s white paper explains, “Though most human genetic variation is apportioned within, rather than across, populations, there have been barriers to gene flow on an inter-continental scale. This has resulted in genetic distances between groups being great enough so as to identify them accurately with molecular markers.” The key here is “barriers to gene flow.” In regions of the world that have been major migrational crossroads (as opposed to locations that have been effectively physically isolated, such as islands or other relatively inaccessible places), it is much more challenging to identify valid location-specific clusters of genes. Surely, none of the testing companies simply trusts the reports of their users about their ethnic origins!

          • If all 4 of your grandparents were born in Spain, for example, companies like LivingDNA and the POBI project make the presumption that they are “from” there historically.

          • Thanks for explaining the phrase and the concept involved. But when dealing with history and genealogy, time can never be ignored. As I see it, some sites have not or do not address this accurately, as it does not benefit their business model. They usually provide estimates or prognostications that are based upon computer modeling with no hard evidence except that which is mostly or entirely generated from their customer base. Their methods go unquestioned, and it becomes that of trust or belief by their clientele. No matter how sophisticated the ‘white papers’ or the “principal component analysis” becomes, neither describes the time frame involved in these analyses. The reliance upon algorithms may sound or claim to be ‘scientific’ but they are not hard proof, nor are they that of ‘hard’ (provable) science..

            Some sites try to provide prognostications much further back in time beyond that of written record. 500 years ago, 1000 years ago, or more. To their credit, Ancestry’s time line of predicting ethnicity now takes that into account, more than before. In my case their report seems to be taking into account that which is provable with regard to extant paper records. In my case, their stated time line is not attempting to predict what happened before documentation (AKA ‘paper records;’) were extant. (Some in the current world of genetic genealogy tend to dismiss or denigrate “paper” genealogy as being inherently dubious or worthless. Are so called paper records now being replaced with what are now called academic ‘white papers’?)

            How does any business including that of genetic genealogical websites make money? It’s simple: Find out what the marketplace of customers want and sell it to them. In this case, it often amounts to telling them whatever it is that they want to hear.

            This is salable and effective merchandising, but let’s recognize it for what it is. The estimates of GG sites are plausible or probable, and even perhaps often correct. As they old saying goes, even a broken clock is right twice every day. But please, let us not consider such estimates as being infallible, always accurate or factually provable.

            “They really are nothing more than estimates.” 😉

          • One can question to what extent it is even possible to reconstruct the genetics of past populations, but I don’t think it’s fair to say that the methods are based on “no hard evidence” or “go unquestioned.” Plenty has been written about the methods, and, at least as described in the literature, they are a reasonable approach to using the only evidence we have (present-day genetics combined with history and what we think we understand about prehistory). Whether certain companies are too uncritical in their use of customer data or are manipulating their analyses in an attempt to pander to their customers’ biases in order to boost sales are separate questions.

            Personally, I think admixture analysis at the level customers seem to seek is probably an unrealistic goal, and I advise people not to take their results seriously, but to consider them mainly a form of entertainment. Certain companies that advertise widely on television need to scale back their claims! At this point, we simply don’t know to what extent to believe the admixture estimates. They seem most accurate for people whose ancestry comes at least in part from relatively isolated, inbred populations (for example, all the companies seem to get fairly consistent results for Ashkenazi Jewish ancestry) and wildly inconsistent for pretty much anyone with other European ancestry, because of the difficulty of constructing valid reference populations for most parts of Europe. For example, my three brothers and I have extremely different ethnic origins according to FTDNA analysis (the only company we’ve all tested with), and I don’t believe that the huge differences can be accounted for by each of us just happening to get DNA from very different combinations of distant ancestors (as has been suggested by defenders of the test). After all, we do test as full siblings and share large amounts of DNA. It’s much more likely that the analysis results have huge uncertainties.

          • I agree you have to take these tests results with a grain of salt, even though you shouldn’t believe everything these tests show, some of the results they do get right I also have tested with FTDNA, 23andMe and AncestryDNA, for me FTDNA is the least accurate it used to be AncestryDNA that came in last place before their new update, I’m a person of very mixed ancestry so I know they all have a difficult time placing me 23andMe right now is the most accurate, ancestry comes in a very close second, I know they are getting it right sciencticly how else would they know I have Native American ancestry, I have a problem with their percentages but other than that most of it is correct, I’m African American with European ancestry also but FTDNA seems to think I mostly European ancestry weather than North West European

          • Your experience is pretty much what I would expect — all the companies do MUCH better at picking up Native American, African, and Ashkenazi Jewish ancestry (the types of ancestry for which there are reliable reference populations) than they do correctly attributing ancestry to specific parts of Europe and the British Isles. I also have found that 23&Me is most accurate, followed by Ancestry (both before and after the update), and then FTDNA. But the only thing any of the companies really get right is my approximate percentage of Jewish ancestry; on FTDNA, the estimated small percentages of Jewish ancestry actually correlate with the numbers of Jewish matches my brothers and I have. My point is that no one whose ancestry is predominantly from Western Europe and the British Isles should take the specific percentages attributed to the various regions seriously — they are so approximate as to be meaningless.

        • Regardless of what you have deduced about an algorithm problem, there are undeniably all sorts of problems with how reference populations are identified.

  7. Extremely disappointing update in my opinion. Went from 9 regions to just two very broad regions with no detail whatsoever. How can they say that the results are more granular/detailed then? Also, I believe many, including yourself, have lost their proven minor ancestries with this update.

  8. Have to agree with the broadly Southern European (Spain/Italy) complaints above. Ancestry now wants me to believe that my ancestors who mixed with natives in Mexico were actually . . . British! By removing all my Southern European ancestry and replacing it with NW Europe, they’ve essentially re-written history. Who knew the Conquistadors were actually Anglos!

    The irony is that Ancestry has made Southern Europe more granular (breaking up Spain into three regions) while also saying that separating Southern and NW Europe is just, well, “messy!” Is there a new white paper circulating somewhere? Not impressed so far.

  9. I lost Iberian Peninsula which made up about 8% at least. I lost trace regions (3% and below. I was pretty sure I had some Spanish in my family from Florida. Also, 10% Scandiniavia disappeared , but I have 3% Swedish and 2% Germanic. My NE Europe, Wales and England was a whole new thing, unless it took the place of Western Europe which had been 31%. The new NE Europe, England Wales is like 58% and my Scottish Irish Wales went down from 43% to 36%. Almost all my great grandparents are Irish immigrants and married to the same. So, I anticipated that would of stayed the same or much higher.

  10. RE the ongoing discussion, in reply to Susan Daikin:

    We agree on many or most points, but any discussion addresses differences, rather than total agreement. The mention of what “has been written about the methods, and, at least as described in the literature” is that of the ‘white paper’ or that which is generated by academic analysis. Theoretical discussion may or may not be entirely accurate. In genealogy, some “paper” records are inherently more accurate than others, and an experienced genealogist understands that veracity is discovered and proven by examining ALL available records and evaluating them, putting them into perspective based upon evidence found.

    You mention that the estimates are “mainly a form of entertainment”. Yes, but not entirely, in that they do have some basis in modeling, one that is statistically based. My point is that that basis is not merely or mainly one of entertainment, but it is not one that is based in hard science (meaning provable fact), either.

    You mentioned that you only tested at FTDNA. I have tested at 23andme and Ancestry, with my results also transferred them to FTDNA. In my particular case I have a very limited admix, that of only two very similar but distinct ethnic groups that were living for centuries in the same region. (Note that I said “region” rather than ‘nation’ because national boundaries can and did change over time.) Thus I have an advantage that most others do not, with my very limited admixture, and I have established proof of it dating to before 1800. Before that there are no records extant. If and when an ethnicity estimate does not coincide with that record, I know it is inaccurate – no matter how sophisticated the model purports the estimate to be. Before that time period, before which there are no documented records, the estimates become nebulous – the stated estimate is possible, maybe even probable, but not provable.

    One thing I will say about the latest estimate from Ancestry is that their estimate for me is now entirely accurate back to their stated time frame of 1800, and they do not make a definite claim for whatever time preceded it. Before that time frame they are now including a somewhat broader area of territory that is geographically adjacent, regionally. That is believable, rather than other assessments that were assigning distant areas that are remote from my known continental locale. Plausibility tends to engender acceptance of an estimate, especially when it aligns with known evidence.

    • No, I said FTDNA was the only service where all my brothers and I have tested; I myself have tested with FTDNA, Ancestry, and 23&Me. The statistical methods used in admixture analysis are valid and rigorous, and the “white papers” simply explain these methods. I don’t doubt the appropriateness of the approach. However, one key limitation is that we are basing estimates of the past genetic composition of populations on what present-day people think they know about their ancestral origins, and another key limitation is that in many parts of the world, no long-isolated populations exist for comparison. It’s unrealistic to expect that the testing companies will examine the paper trail of every customer in detail and judge the soundness of the genealogy. Shortcuts must inevitably be used (e.g., having four grandparents reportedly born in the same country), which inevitably means compromise, and I’m sure the details of how customer data are used differ among the testing services. Your situation is unusual; the majority of customers cannot attribute much, if any, of their ancestry to distinct ethnic groups that were living for centuries in the same region. For this majority, admixture analysis amounts to a parlor trick. It’s dishonest for the testing companies to create expectations of greater accuracy than is possible. BTW, my brothers and I all obviously share the same ethnic origins (mainly a mix of English, Scots Irish, and German, much of it documented back to the 1600s, with one Jewish 2G grandparent, and tiny amounts of Welsh and French, and some brick walls in the early 1800s), but in the FTDNA analysis, a given ethnic origin can range from 0% to over 70% among the four of us. Such huge differences among four full siblings simply cannot correspond to real biological differences, but rather illustrate the enormous uncertainty in the estimates. Customers whose siblings or parents have not tested, and who therefore have no basis for comparison, are left with one estimate that might or might not have any correspondence to reality.

      • You points are very well stated. The problem is that no matter how sophisticated (“valid and rigorous”) the methods are, the estimates and predictions stated are often flawed due to other factors that you described. Not everyone is willing to recognize or admit that those deficiencies exist.

        There are GG enthusiasts who embrace any estimate with fanatical zeal. Some have faith in the reports as being irrefutable. Many even admit that current methods are imperfect, but that over time “the science” will eliminate all inaccuracies. This remains doubtful, and the wide range of reported results from being very accurate to wildly inaccurate reflects that.

        My particular pet peeve is that of the time frames involved in stated estimates: the more distant the century, the more speculative or theoretical the estimate becomes. Added to which the number of distant ancestors anyone has increases exponentially as we go further back in time, as does the increase or broadening of regional areas due to possible migratory patterns. The ultimate brick wall is that of reaching a point in time before which there are no applicable records as a basis for proof. If estimates generally need to be ‘taken with a grain of salt’, those that are made about more remote time periods become a very salty stew, indeed

        • Well said! Unfortunately, most customers have no clue where the estimates actually come from, or, therefore, the huge inherent uncertainties. What offends me is when the “enthusiasts” accuse people of “denial” when they question the estimates — I’ve even seen people claim that “DNA doesn’t lie” and lecture others about “facing facts.” But what offends me most is that companies knowingly make misleading claims about their products in order to sell kits.

          • Most of the customer base of any of the GG companies is comprised of amateur genealogists or curiosity seekers. The sales pitch for kit sales is geared toward luring them in. I realize that without attracting the masses these companies probably would not exist, but little is done to educate these subscribers after they are enrolled. (Kudos to Roberta for trying to bridge this gap.) Those who become enthusiasts (or we might call them fanatics?) have a *belief* in “the” science. This is oxymoronic as their unquestioning devotion is an expression of faith and they have no understanding about what science actually is, or how it functions.

            There is a currently popular trend that anyone who questions or voices criticism of unswerving belief in “the science” or of any other form of unchallenged group-think deserves being labeled as a “denier”. Then browbeating and calls for silencing of any questioning or disagreement begins. Critical thinking is under assault in today’s world.

            The matter is not predicated by the statement that ““DNA doesn’t lie”. The problem is that those who use that statement as a foil cannot face the reality that the *misinterpretation* of DNA can lie. Whether intentional or not, at least the interpretation can be inaccurate or misleading.

  11. I am disappointed because this now tears apart the relationship of genetic connection to DNA Circles; if my ethnicity is different then also is everyone else. Therefore, the genetic relationship markers are going to be off. Sorry, just lost all interest in Ancestry DNA and their credibility. I have with drawn from the DNA Matching and Circles, I have with drawn from the human genetic study program, and I no longer have any wish to continue genetic testing of my family members with any DNA Company. It just seems to be a big fraud!

    • I think there may be some confusion. The ethnicity estimates are ONLY interpretations by Ancestry of portions of your DNA. The DNA itself didn’t change. Who you are related to didn’t change. Ethnicity estimates are just that – and they are the most difficult part of DNA testing. The science of matching is much more straightforward. DNA testing, at least not at the major companies (Family Tree DNA, MyHeritage, 23andMe and Ancestry) is not a fraud. It’s fine if you don’t want to participate. Everyone has the right to make their own decisions. However, the circles are a result of DNA matches with those people and that fact that you show a common ancestor in a tree. If you are interested in genealogy, they are incredibly valuable hints. Ethnicity results are really only for entertainment value, and little more, in my opinion.

      • I’ll try to say this in as few words as possible.

        Genetic genealogy is only a tool, as are all its related features. It’s a recently invented tool, but it does not replace all other tools in a genealogist’s toolbox.

        If the only tool someone has is a hammer, he will tend to think of every problem before him as being a nail. If a mechanic ignores all the other tools he has, then he has been seduced by the latest, shiniest tool touted in the marketplace. Also, remember that “it’s a poor workman who curses his tools.” An experienced genealogist utilizes many tools, and knows the value and purpose of each one in his tool kit.

        Now think of what estimates actually are. They are predictions, just as weather reports are: they are best guesses based upon scientific data available. (The database itself is always much less than complete, by definition, and is subject to constant, ongoing revision.) Are your weather reports ALWAYS accurate? Mine aren’t. in fact their accuracy rate is often poor, or the reports issued are very often downright WRONG. But by tomorrow we tend to forget about their lack of accuracy and hope that their next forecast might be correct. If you think of these ethnicity estimates, including predictions about the degree of relationship of your distant relatives as being similar to weather reports, you might be able to put them into perspective and not think of them as being infallible.

    • As Roberta says, the accuracy of Ancestry’s ethnicity estimate has nothing to do with the validity of your matches. You and the people you match DO share DNA sequences, and most likely share common ancestors, and there is nothing fraudulent about these results. They are very valuable for providing genealogical clues. The ethnicity results, on the other hand, are a sideshow and a distraction.

  12. The ancestry estimates they are trying, on the new update mine is more accurate than AncestryDna ever had them the percentages and the ethnicities they are mostly guessing, one thing I haven’t figured out yet why is my Native put in the category of Native American-Andean, all of my cousin matches I viewed so far shows Native American, North, Central.

    • I’ve seen ethnic estimates of Native American included with that of Asian. But what time frame is relevant in estimating that? Thousands of years? The land mass of two continents is a VERY large geographical area (as Columbus discovered the hard way).

      Gee, maybe they got the continent right for you. But don’t bet on it… 😉

      • Yes they do sometimes include Native American as Asian, on the old estimates they only showed my Native American as Central Asian on AncestryDNA, and FTDNA old estimates said Siberian.

  13. I’m not impressed with the update. My grandfather was born on the Frisian island of Amrum, off the coasts of Germany and Denmark, average population of 500 throughout history. My line is traced back to the 1600’s. I have hundreds of DNA matches with Danish, German and Frisian individuals. My great-great aunt wrote a book about the family history; the DNA matches correlate with it. National Geographic Hapalog states we are a rarer group. Now, Ancestry states we are Swedish. I am just shaking my head

    • It is fine to call them all Scandinavians, but Danes are not Swedish. They need a Danish reference population. My Mom’s Danish heritage is assigned to Sweden, Norway, and England by Ancestry and that is ridiculously incorrect.

      • Same thing happened to me, I went from 72% Scandinavian, to 39% Norwegian and 33% Swedish. My grandparents were all born in Denmark. I can’t figure why Denmark is left out.

    • Amrum Island is one of the Frisian Islands. It is just off the west coast of Denmark and North of Germany. It is nowadayslisted as part of Schleswieg-Holstein.

  14. I think they did clarify some things more efficiently and to understand the dominant DNA story of a region, you have to look more closely at the regional “zones” listed under your new estimates, to see how far outside the new, clarified boundaries certain DNA is seen present in a region.

    For instance, someone might have grown up with first generation, German immigrant grandparents. Their whole life, ethnically, they’ve identified as 50% German because “one side” of the family comes from Germany.

    However, 300 years before their Grandparents arrived in The USA, travelling diasporas of genetic groups could have relocated to a German region from somewhere else for work, to avoid persecution or just to settle and work a considered uninhabited place, etc. In 1500, there were legal rules about property ownership of course but not the same kind of building permit, land ownership boundaries and regulations you’d see in 2018 Kentucky. In some places, at some times, just showing up and working the land meant that you could stay, so long as you paid aristocrat owners for the right to use their land. And if your people developed a particular skill in a highly competitive way, you might be the only ones able to exploit a region’s natural resources in a particular way, therefore ethnically dominating the place for some time because of what you and your ancestors offered the region.

    Let’s say a group of mostly, genetically Slavic people relocate to an isolated region in The Black Forest; let’s say ethnic Poles re-located to an isolated region in Bavaria and for a 100 years, mostly married within their ethnic-genetic group, occasionally marrying the offspring of local families scattered across the forest, etc. Also, some social acceptance could be dependent upon affluence, local performance, having to work your way up to a point in a society where you were deemed “acceptable” to marry a local girl or boy. Even then, you could be facing prejudice from people who would still see your family as outsiders, even if they’d been in a region for 50 years.

    For the first century your family was in a region, they could have been fairly genetically isolated from the “native” peoples. As time went on, they were considered assimilated and more frequently married with genetic “natives” until their cultural and ancestral identity couldn’t be considered anything else but German. Genetically, however, those remnants of family history can come up for descendants in the genetic lottery, where one side of the “German” family can have lots of genetic influence from a Slavic genetic past.

    The scale of time in regards to genetics can be daunting compared to a single, human lifespan, as well. You have to look at the topography of a region and think of the logic of travel for a time, in addition to social prejudices and religious group genetic isolation from others (with the tradition of people only marrying within their family’s religion for most of modern, human history): Are there treacherous mountains to scale? Deserts to cross? What was the travel technology like at the time your ancestors lived in a region? Today, someone from California can fly to Japan in an afternoon to meet and marry a spouse. At one time, it would be a difficult trip for someone from Austria to travel to parts of Italy. A coach ride from Austria to France could take a week. This meant only the most adventurous or wealthy would be making certain trips for long swaths of history of a region.

    These conditions also mean that genetic isolation for some peoples was steady for a great deal of time until technological advances allowed for new kinds of travel, opening the world to previously isolated groups. As people traveled and relocated, two groups that were genetically isolated for centuries might suddenly jump up in instances of two genetic groups coming together in marriage and in producing offspring. And regarding imperialism and war, genetic groups might suddenly come together for the first time in a millennia, under less positive circumstances.

    But what could seem like an eternity of genetic representation in a region for a family from the point of view of a single, human lifespan, is only a moment in the timeline of human genetic history of a region.

    Your “English” family has been in England for 500 years, which seems like a very, long time. But before that, they may have been from The Netherlands. Before that, Frisian and Viking genetic influence from Norway could have made it into your pool. So that means Grandma Jones might find out that her Englishness is the version of Englishness with some genetic background tied to groups present in a region far enough back to be considered native (like Picts or Celts) with some Dutch and some Norwegian genetic influence because of those groups’ history in England. Another “English” person could have a different mix of ancestral, genetic influences, despite both families living in the same community together for centuries.

    But 500 years in a country, inter-marrying with locals and assimilating into local culture means that by the time her family arrived in America, they couldn’t think of themselves as having an ethnic heritage that was anything other than English. It’s the timescale of individual, human lifespans vs. the overall timescale of human genetics that can distort our perspective on this.

  15. Something else I wanted to add, based on my own genealogical research and DNA results, is that some of the most gene-sleuthing you’ll do is when you try to understand how pieces of DNA ethnicity results would likely be assigned by parent, grandparent because sometimes, a mystery “bridge” result could be divided between both sides of the family where you received a lower percentage of an ethnicity than you were expecting but that “bridge” result can make storytelling sense of the difference.

    For instance, growing up I was told my Paternal Grandfather was just German, period. My Maternal Grandfather was just Italian, period. But when I got my results back, “Germanic Europe” only appeared as 8%; Italian was even more of a surprise, at just 4%. Yes, I was aware that DNA was a wildcard and that I could have just gotten the lesser of the genetic representation of these regions in my DNA but did it make sense that those two areas would be SO low?

    But then I saw that my “Eastern European and Russian” region came in at a whopping 39%. Now, I knew my Maternal Grandmother had a Father who was both Polish and Russian and that there may have even been some Polish on my Paternal side. Then I saw that “Austria” was included in that classification and that it was — in both the older percentage and updated one — the #4 or #5 country in my particular percentage in that classification group, which meant I matched highly with Austrian genetic populations in that region.

    I had no known family history of being Austrian but it wouldn’t have surprised me at all, finding some on the German side of the family because Austria is so close to Germany and historically, Austrians shared many of the same cultural, industrial, political and religious traditions as Germany. It was very likely that people from these two countries would find common ground, marry and have children along the way. Austria also borders Italy and some of my Italian ancestors could have been from the Northern region, perhaps with some Austrians travelling to Italy for work or the daughters of Austrian co-workers marrying Italian ancestors, Austrian DNA finding its way into the family line that way.

    That means it’s very possible my Austrian DNA percentage can be split in some way between my “German” side and my “Italian” side. I also had no knowledge of any Scandinavian DNA or ethnicity, no Scandinavian immigrants or names found three generations back but wouldn’t have been surprised if on my German side, some distant migration of Scandinavians traced back and I had some remnants of Scandinavian DNA from that side.

    However, I was able to trace my entire English side back to Vikings, every line. One line traces back to early settlers of Finland and Iceland. So we’re talking about very early periods. My original percentages came in at 3% Scandinavian and less that 1% Finnish. The new percentages have trimmed Scandinavian to 2% Norwegian. The Finnish is lost but I’ve gained 5% in Baltic States. What’s really fascinating about this, is that in the stories and archaeological research surrounding my very distant Finnish ancestors, they appeared to pop-up in the region out of nowhere and researchers believed they may have originated from Estonia.

    Estonia was my top representing region in the “Baltic States” category. Some of that might be from my Russian side but was a little, archaeological mystery closer to being solved, there, too? There has also been debate about whether Rollo The Viking was technically Danish or Norwegian. Well, he’s one of my ancestors through a DNA confirmed and heavily documented, full of official records bloodline. I have no recent family history tied to Scandinavia, nothing known for centuries, yet Norway shows up in my DNA results at 2%. Is this possibly some more evidence in a DNA story of this figure and his genetic ancestors, something that could be added to the solving of an archaeological mystery?

    So again, the more you learn about the stories associated with your ancestors, the more even surprising DNA results can make sense.

  16. I admit to be a total novice when it comes to this, but the update seems pretty lacking to me. My father is 100% Kurdish, and my mother 100% Flemish (I have trees going back hundreds of years, all in the Flemish area). Of course both are assigned with neighboring markers etc, but for some reason now I’m listed also as 6% Italy, 4% French, and 2% Greece/Balkans, when NONE of that shows up in either of my parents’ DNA tests. Both my mother and I lost our Iberian DNA as well, even though 23andme lists me with a very small percentage of that as well.

    • Let’s think about the time frames involved: 100 years? 200 years? 500 years? 1000 years? 1500 years? 2000 years? 5000 years?

      Virtually everyone can relate to ethnicity. That aspect is relatively modern. As admixture increases, it becomes increasingly more diffuse or diluted, and thus more complicated. Everyone wants the predictions to tell them something useful, something that they can relate to. But it seems that no one thinks about time, because most people can’t fathom long and distant spans of time, nor what the world was like long ago.

      For purpose of a general example, was England the same in the year 500 as it was in 1066, or 1500? Or in 2018?

      Genetic genealogy is useful in providing information for some things, but it can’t explain all things, as it is often lacking in specificity. Don’t expect it to explain everything, and you won’t be disappointed. 😉

  17. You can’t help but wonder if some of the surprises of ethnicity are a result of unknown adoptions or pregnancies of unwed mothers, or pregnancies due to affairs of married people. Certainly, society hasn’t changed much in respect to sexual promiscuity. And, of course, may be much more prevalent today than in recent generations. I have friends who have recently discovered that they were adopted or that the person they thought was their father, is actually not their father. Just saying.

    • Possibly in a few cases, but I’m sure it’s mostly problems with the way in which ethnicity is estimated, especially for European descent. There are plenty of people with well-documented trees confirmed by autosomal DNA matches with distant cousins who are still getting bizarre ethnicity results.

  18. Ok let me state emphatically there is some sort of drastic mathimatical error with the Ancestry September 2018 ethnicity estimate updates. My specific results are completely out to lunch in a single generation. While there were questionable results with the previous itineration these are demonstrably incorrect. I could go on and on about how my maternal grandfather is verifiably from Italy and I went from 20% to 0% Italian, and I have found numerous matches back to the Turin region etc… In my case I had my mother sequenced as well as myself, in part to serve as a proxy. This has worked well and I recomend it to others. All right I will cut to the chase, My mothers mother is from Gemany and her father from Italy so she should be 50% and 50% Italian which she was until the so called update. Now she is 50% Germanic Europe, 37% France and 13% Italian. Ok still possible as our Italian ancestors are from Pedimonte a region of Italy that borders France.

    So then that aside I now show 5% Germanic Europe down from 57%. Seemed very over estimated to ridiculously low. When something changes >11x that is a flag right away. Then there is 0% Italian despite my grandfather and many generations before being traced to Italy on multiple branches. Another bright red flag. Then there is the 11% French, which is plausable if not convincing based upon what I know. So then if we add up the ethnic regions I have in common with my biological mother we see only 16% in common. That my friends is a substantial problem as even Ancestry DNA acknowledges we inherit 50% of our autosomal DNA from each parent, which is a fundamentel biological fact. This eliminates reference population misinterpretations as the root cause, but implies a gross computational error from a severely broken algorithm. We are talking magnitudes of order not vary percentages folk. That Ancestry could allow such a thing erodes all confidence in their technical staff. Why wouldn’t you use one generation matches to act as a fail safe to prevent this. BTW I have called them, I have explained this on their surveys, and emailed Ancestry and all I hear are crickets. I tried to be discreat and give them a chance to rectify this and not a word. Needless to say I will no longer be recomending Ancestry to anyone.

    • Hi. I had a similar experience as have others. I wrote my own reply on this post. Bottom line is many have experiemced a negative bias on Italic ethnicity. In my case, I knew all of my paternal relatives up to great grand parents, and my great grand parents told me of their parents, grand parents, and great hrand parents. We are all from the same small Latin village in Lazio, born and raised, and we have lineage to the ancient Italic tribes (Latini, Volsci, Samniti, Arabi, Greek, Sicilian, Calaabrian). The original Ancestry estimate was 37% Italic (close but not correct) but the “new and improved” estiimate slid to just 15% Italic, which is orders of magnitude WRONG. To validate this I was tested in four other DNA labs which show 50% Italic (including Greek, Arabic).

      Here is a Newsweek article where others with Italian DNA had their results drastically reduced:

      Some data on Italic tribes of Latium (Lazio):

  19. These new algorithms are very wrong. My father and I, my praternal aunts, uncle, grandparents, and several generations up the paternal tree are native to Lazio (Latium), Italy. I have been tested in FOUR other DNA labs in addition to Ancestry, and all of the others got it exactly right (small margin of error) while Ancestry is off by ORDERS OF MAGNITUDE. These four other labs show 50% Italic, with the original tribes of the area (Latini, Volsci, Samniti, etc) and tribes from southern Italy, Greece, and Arabia. The other four were virtually equivalent.

    On the other hand, Ancestry went from an original estimate of 37% Italic to just 15%, with the balance now being Russian and Eastern Europe/Slavic. My maternal side is Polish, Russian, and possibly German. In summary, I went from 50% Eastern European/Russian to 85% and my Italic blood line was almost nullified. Such drastic changes which conflict with my known ethnicity on both sides means that the new algorithm is orders of magnitude worse than the original estimates, which were wrong to begin with but at least “more” accurate than the new algorithm.

    When I approached Ancestry with this data they said that “You need to get in touch with your fasmily tree on the paternal side”, even after I told them that I was born and raised in Casalvieri (Lazio) and I PERSONALLY KNEW all of my relatives, including all of my great grand parents. They were all born and raised in the same town, whose inhabitants have lineage back to the ancient Italic tribes. My great grandparents told me of their parents and grand parents, all of whom have gravestones in the local cemetary. Ancestry still won’t admit that they got things HORRIBLY WRONG.

    Here is a Newsweek article which demonstrates the Anti-Italic bias in Ancestry:

    Iitalic tribes:

    The fact that four other independant labs got it right but Ancestry was off by orders of magnitude proves that Ancestry’s “new algorithm” has SERIOUS BUGS.

  20. Totally disappointed with this “update”. I went from having 51% Ireland/Scotland/Wales, 18% Scandinavia, 16% Iberian Peninsula, 8% Great Britain and several low confidence regions to 58% Irish and 48% British. In other words, they wiped everything. I don’t understand how the results could change so dramatically, especially going from only 8% British Isles to 48%…it’s a joke in my opinion.

    I don’t think these new results are accurate and I have my doubts about the original estimate too, since I know for a fact on my paternal grandfather’s side his family emigrated to the UK from Italy yet there is not a single trace of Italian on my DNA estimate…at first I considered that maybe it was lumped in with the Iberian Peninsula (I know Italy is not in this region but its the closest one) but now that that has disappeared, I don’t even know what to think. I think at some point I’ll sign up for 23andme and see how the results compare.

  21. I think the new version is way off for me. I am mostly Germanic on all branches of my tree. There are a few people from Ireland and England, but those are further back than my Native American blood which doesn’t show up at all. I went from 52% Germanic to 2% Germanic and 36% Irish/Brit to 73% Irish/Brit and moved my Danish blood to some Swedish blood I don’t have. I think they need to re-think their algorithm. Maybe adding in all that DNA from one region skewed the results.

  22. Ancestry,

    I accepted the new interpretation of my DNA test. I am sorry that I did this as it Groups (lumps) too much together. I realize that I can see the previous analysis but some of the details have been lost, or I simply can’t find them.

    The new analysis is too “lumped” and as such it has lost most of its interest and intrigue. If the data had been homogenized a little more Ancestry could have simply lumped everyone’s DNA as human from planet earth. This would have the advantage of making testing unnecessary and Ancestry could layoff all those pesky analysists.

    Ancestry vs Ethnicity
    noun: ethnicity; plural noun: ethnicities
    the fact or state of belonging to a social group that has a common national or cultural tradition.
    “the interrelationship between gender, ethnicity, and class”

    Ethnicity and DNA profile have very little to do with anything.
    Ethnically speaking I am American, I was born and raised in America under the customs of the US. My Ancestors came from other counties. These countries have their Ethnic Customs and practices. By saying my DNA profile came from from different countries doesn’t mean I practice their ethnicity or customs.
    So the conflation of the terms Ethnicity/Customs and Genetic Profile has very little meaning.

    On the earlier version I realized that all the low % were part of being British and Western European. I always wondered why my percentages did not add up to more than 100%. By this I mean that my recent ancestry is very English but that the English had many parts from elsewhere (100% English with 10% Western European with 12% Scandinavian …) adding up to 156%. The Pre-historic British Genome would be European(Celt, Pict…) along with singlets of other minor regions. These sniglets would have contributed to my ancestry and would indicate that I am a low % descendant of these other regions, however slight.

    ?Is there a set of sequences of alleles (Amino acids) that is uniquely British. By this I mean if I have these sequences I have British ancestry?

    ?Can you tell if my Sequences Indicate whether I descend from Pict, Celts, Saxon, Anglo, Danes, or a mixture of these?

    In conclusion; It does not seem that Ancestry DNA knows what they are doing.

    From the British fellow
    Wm Glover, good luck I think Ancestry will need it.

  23. I just found this. I tested at FTDNA and am Western European with 21 % for Italy and Greece, and Ancestry had me at 7% on Italy and Greece and now Ancestry has deleted my Southern European. I think Ancestry has to be wrong and told them so.

  24. What I’m not understanding, in question of ethnicity tied to geography, is “when?” By that I mean: when DNA readout tells me my ethnicity is 90 percent geographically tied to, say, Great Britain, does that mean my forebears were located there 500 years ago?1000 years ago? 2000 years ago? We all descended from humans who originated in Africa, but that was more than 150,000 years ago (other estimates too). Obviously, the algorithm does not account for that. But what time frame DOES it reflect? I have looked in FAQs, but not found the answer.

    • There is no answer. Ethnicity is an estimate. Both in terms of amounts and time. There is no reliable way to estimate.

  25. I got my first results last year and just got the updated results. I’ve also done 23andMe and FamilyTree. FamilyTree was so far off I don’t pay any attention to it, but the others I do. I should be 50% German or Germanic, and I always assumed 50% British Isle with some Dutch. My German is always strong in both but lower than I’d expect. 23andMe has it at 37% at 50% confidence, but with a 90% confidence level that shrinks incredibly to 1.9%, while the British and Irish is at 34% but with a slightly higher confidence level at 90% of 7.9%. In the new Ancestry, 48% England/Wales/Northwestern Europe with a 44-60% confidence level and Germanic Europe at 34% and a very tight confidence level (33-34%). My brother only has 26% Germanic Europe in Ancestry. My mother’s grandparents both came from Bavaria, and I have verified lots of English, a fair number of German and Dutch, some Irish/Scots/Welsh and French on my father’s side through research. There is supposed some Swedish and I see nothing so far. The lower estimates on Germanic in both tests has been slightly surprising, but 23andMe has a big chunk at “broadly Northwestern European.” I’ve also used Gedmatch’s admixture tools and Eurogenes comes up with some odd estimates, which keep pushing Danish in both my and my aunt’s estimates–and as far I know we don’t have Danish ancestry.

    • In case any of y’all want to troubleshoot the numbers:
      Ancestry 2018
      Ethnicity Estimate
      48% England, Wales & Northwestern Europe
      34% Germanic Europe
      13% Ireland & Scotland
      4% Sweden
      1% France

      23andMe 2018 update
      French & German 37.0%
      British & Irish 34.1%
      –United Kingdom
      Scandinavian 1.6%
      Iberian 1.3%
      Finnish 0.9%
      Ashkenazi Jewish 0.4%
      Italian 0.3%
      Broadly Northwestern European 19.4%
      Broadly Southern European 2.7%

  26. I just got my Ancestry DNA results. In 2017, I took the Family Tree DNA Family Finder and the mtDNA tests. Family Finder originally reported me as 91% Ashkenazi Jewish, 6% Western & Central European, and 3% North African. However, after Family Tree revised their reference populations, my results changed to 99% Ashkenazi and 1% East Central African. Ancestry reports my results as 100% European Jewish. All 4 of my grandparents were Jewish immigrants from Eastern Europe. My 1% East Central African disappeared. Perhaps Ancestry includes part of East Central Africa with the Middle East. (My first cousins have some African, SE European, Middle Eastern, and Caucasus ancestry in their results.)

    Ancestry increased the borders of the European Jewish reference group significantly. The borders now extends from Northern Spain and the Atlantic coast to Russia. It covers most of Europe. My mtDNA is Sephardic Jewish, probably from Northern Spain.

    I really liked the way Ancestry broke down the huge European Jewish group into two major groups – “Western and Central European” and “Central and Eastern European.” Each subgroup contains three specific reference populations. I belong to the NE Poland, Baltic group under Central and Eastern European Jewish and to the Romania, Moldova, Western Ukraine group under Western and Central European Jewish. Two of my grandparents were from the Ukraine, although one had an Arabic surname. Another grandparent was from eastern Poland near Belarus and another grandparent was from Eastern Romania.

  27. Just got my update DNA. They added Irish-Scottish 2%. I was born in the Philippines, but my dad’s side was originally from Iberian peninsula back in the 1800’s and immigrated to the Philippines. Not sure hos the Irish-Scottish popped up. I know from history that Spain and United Kingdom (now) were allies somewhat. King Henry Viii married Catherine of Aragon back then. I guess I have a reason to celebrate St. Patrick’s day.

  28. My Jewish 1st cousin and her daughter both got 1% Irish/Scottish on their Ancestry DNA tests. Surprised at those results. Her daughter also got 1% Iberian.

  29. I was 38% Irish, 32% Scandinavian and 10% Iberian and only 6% British — that didn’t make much sense as I am English and my family tree is largely English with an unconfirmed story of some black mixed race heritage.

    The update has made me 71% British 24% Irish 3% African and 2% Swedish — so the Scandinavian and Iberian has completely gone.

    This seems to match my recent family history more, but it doesn’t match the older family history of migration from the north and my Norse surname.

    Confusing why the change should be so drastic.

  30. Pingback: 2018 – The Year of the Segment | DNAeXplained – Genetic Genealogy

Leave a Reply