Thursday, April 23, 2020

More evidence that Ancestry.com’s formula for Italian ethnicity is skewed

Author's note: This post is now out of date! Ancestry has updated it's database and formula. I will leave this post online for posterity, but Ancestry has made huge strides with it's April 2022 update. Read about it here: 

---------------------------------



It’s official. Ancestry.com’s formula for determining Tuscan ethnicity is seriously fouled up—and I now have DNA ethnicity tests and genealogical data I needed to prove it.

I’ve written about this before in Are new Ancestry algorithms ignoring northern and central Italians? But even as I criticized the company, I wondered what would happen if one of my Italian cousins from Tuscany took the test. I and all of my first cousins in America can be considered tainted because each of us has one parent who came from somewhere other than Tuscany. All of our parents who were born to Tuscans Michele Spadoni and Anita Seghieri have passed away, so we can’t test them.

Most of my Italian American cousins come out more French than Italian, which is confusing, since I’ve researched most of Michele’s and Anita’s family lines back to Tuscany from at least the 17th century, and many lines much further—some even to the 1200s. I have found birth records for every Italian ancestor on the Italian side of the family going back to all 16 of my third great grandparents. Each one was born in the same valley in Italy, the Valdinievole (roughly between Lucca and Montecatini), or just a few miles away. All of the surnames are common to our little region of Tuscany. All of my first cousins should be approximately half Italian, but none test more than 11%. One tested 0 percent Italian and 45% French.

Anna Giuntoli Hughes
However, I recently made contact with Annamaria Giuntoli, a second cousin of my dad. She was born in Italy, and her parents were also from families rooted in Tuscany. Names in her family history fill up the Valdinievole regional archives: Giuntoli, Magrini, Grassi, Capocchi, Montanelli, Pinelli, Pieretti, Bellandi, Pucci. Six of those names are also in our direct line of ancestry.

So what does Ancestry.com say about Anna’s ethnicity, which should be close to 100% Tuscan Italian. It says she is 49% Italian and 51% French. The ethnicity estimate becomes even more inaccurate with her son Marco’s test. Anna married a British man, so one would think that Marco would test around 25% Italian and 25% French, right? Nope, his Ancestry.com test says he is 54% British, only 2% Italian and 35% French—indicating that the genes he inherited from Anna were actually much more French than Italian.

Another cousin who should be close to pure Tuscan is Joan (Seghieri) Reiling, born to Dante Marcucci Seghieri and Maria Luisa Togneri. Both surnames have long roots in Tuscany. Joan tests 50% Italian, 44% French and 6% from Greece and Balkans. Her grandson Michael tests as 0% Italian and 10% French.

Still another cousin Vilma Ferranti Mott, now deceased, was born to Gabriella Montanelli of Montecarlo and Giuseppe Ferranti of Villa Basilica, both small towns in the province of Lucca. Her results: 44% French, 54% Italian.

This explains a lot about why the ethnicity results for me and my cousins are so skewed toward French. Somehow, Ancestry.com’s algorithms find Tuscans to be roughly a half-and-half mixture of French and Italian. History does not support this odd admixture. Except for the invasion of the Gauls in the years 200 to 400 BC, inland Tuscany has never received an influx of French immigrants. If anything, the opposite is true, as social scientist Robin Cohen reports: “About 5 million French nationals are of Italian origin, if their parentage is retraced over three generations.” And according to official Eurostat data for 2012, the number of Italian citizens residing in France was 174,000. Wikipedia says of Marseille, France, that “in the first half of the 20th century, up to 40% of the city’s population was of Italian origin.”

Why is this discrepancy important? I find it disturbing that so many Italian Americans with Tuscan roots, most of whom speak proudly of their heritage, are disappointed and shocked to be told they are more French than Italian. No offense meant to our French neighbors, who also have good reason to be proud, but isn’t it better to know the truth of our origins and have our pride placed in the right country?

Another sad result of the problem is that some people now wrongfully suspect their grandparents of infidelity. One of my cousins commented, “My mom and several of her siblings have had their results come back as French, with no trace of Italian. It has us all flummoxed. We we were thinking my grandfather must have had a different father (out of wedlock).

I’ve experienced a lot of pleasure from my hobby of genealogy, and I give credit and high ratings to Ancestry.com for its researching tools. It has been a kick connecting with new relatives that I’ve found through DNA matching. But I sincerely hope the company irons out the problems in its methods of determining Italian heritage.
------------
Update (Sept. 17, 2020): In the summer of 2020, Ancestry revised its formula again. Big improvements! They now recognize the ethnic group of Northern Italy. They still have too much France in there, but Anna Giuntoli is now listed as 69% Northern Italy and 5% Southern Italy, a total of 74% Italian. The French is still in there at 21%, but this is a major step in the right direction! In addition, her son Marco is now listed as 18% Northern Italy and 10% France.
Even more encouraging is that Vilma Ferranti was changed from 44% French and 54% Italian to 84% Northern Italian and 14% Southern Italian. Keep it the good work, Ancestry.



30 comments:

  1. Thanks for this article, Paul. It's been the same for my family's Ancestry.com tests. We were so confused to be classified as French. Our family hails from the Lucca region, and my 1st cousin, Sandi, sent me here to read your article.

    ReplyDelete
  2. My 3 x Great-Grandfather was from Tuscany, which means I'm only 3.125% Italian.

    But my test came back with nothing at all from Italy.

    71% England, Wales and North-Western Europe

    24% Ireland and Scotland

    1% other regions.

    I was disappointed not to show any Tuscan heritage (albeit small).

    ReplyDelete
    Replies
    1. You might get different results from 23andMe, which actually has a category for Tuscan Italian.

      Delete
  3. My gradparents (on my dad's side) were from Sicily yet NO Italian showed up in my DNA test results!! WTH??

    ReplyDelete
  4. My question would be were the ancestors of all these families all living in Italy 500 years or more prior. My own heritage varies depending on the DNA test. There was a great deal of migration in Italy and this may be the reason why some Tuscans show as french unless of course you know all of your ancestors from the past 1500 years....https://en.wikipedia.org/wiki/Genetic_history_of_Italy

    ReplyDelete
    Replies
    1. I've done a lot of research on my ancestors in the parish archives in Pescia and found many ancestors--but obviously not all--back 300-500 years. They all had surnames common to that region, which is a relatively isolated area, not near a port city. I've also studied the history of the area, and see no evidence of much migration. For roughly half of the population to be considered French is totally unbelievable. In fact, I read the link you gave me, which says: "It is generally agreed that the invasions that followed for centuries after the fall of the Roman Empire did not significantly alter the local gene pool, because of the relatively small number of Germanics, or other migrants, compared to the large population of what constituted Roman Italy."

      Delete
    2. Same here, my ancestors, Filippi & Parisi, go back 800 + years or more, as recorded in church records. Despite the fact that they lived in the remote and isolated high Dolomites, A.com makes me Spanish and French, broadly referencing all of the coast from the Adriatic to the Rock of Gibraltar. I guess they made us sailors not shepherds, haha

      Delete
    3. Yes, it would seem that 800 years in the same area should be considered Italian! However, in your case, it is a little ironic that one of the surnames is Parisi, so there should be at least a tiny bit of French. But if a person from France moved to Italy and he and all his ancestors married locally, the French DNA would mathematically be less than 1% within 7 generations.

      Delete
    4. I know my maternal grandfather was half Italian. I took a 23 and me test and it seemed to confirm that, saying I was about 14% Italian. Took an Ancestry DNA test and they say I'm 3% Italian and oddly, 10% Scottish. Another Italian friend of mine said his DNA results from Ancestry said Scottish as well. I think Ancestry has got something confused somewhere. I don't think they're last update is accurate at all.

      Delete
  5. Yes, I had the same thing happen to me. My father's side of the family have all been traced back to Italy, but instead of my DNA showing close to 50% Italian, it shows me as 27% Italian and 22% French. Glad to see I'm not the only one!

    ReplyDelete
  6. Some of my Italian Family came from Provence in the South of France when they fled to Italy during the Spanish Inquisition, this is what I learned about my French influence in my Italian Family.

    ReplyDelete
  7. On the surface, it sounds reasonable that this would have happened. However, about 14 generations have passed since the Spanish Inquisition. A single ancestor who moved from France to Italy in 1490 would have passed on about .006 percent of his/her dna to you. So you must have had a LOT of ancestors who moved from France to Italy!

    ReplyDelete
  8. Paul HollingsworthMay 19, 2020 at 3:08 PM

    I took the Ancestry.com DNA test several years ago and it was relatively accurate in mapping with my DNA what I knew to be my family history on both maternal and paternal sides. That included a nexus in Tuscany through my maternal grandparents (Ada Seghieri nee Pantera and Severino Seghieri), both of whom were born in Montecarlo near Lucca. However, in summer 2019 Ancestry sent me an email with "updated" results based on what they said was refined techniques. That wiped out the Italian part of my ancestry and placed it in southwestern France. I wrote for an explanation and was simply informed that this was new and better information. So I took the 23andMe test and it pegged my family tree perfectly, right down to pretty close locations in Tuscany for my mother's side and elsewhere for my father's side. I do not understand the mathematics involved, but I do know that I would not recommend Ancestry based on my experience. (I gave 23andMe tests to many of my relatives and they all came back with very similar results.)

    ReplyDelete
    Replies
    1. I had the same experience with ancestory.com. An early test showed me as Italian and after the update it refigured most of my Italian DNA to French. I found it very disappointing since I am very proud of my Italian heritage.

      Delete
  9. Paul, I apologize that only today I saw your comment and moved it to public status. But thanks for adding to the evidence that Ancestry needs to re-do their metrics to stop putting those of us with Tuscan blood as having French bloodlines! And you didn't mention that you and I are distant cousins (I'm guessing you knew that?). I'd like to ask you a few questions, but it would be better if we could write by email or Facebook. Can you contact me?

    ReplyDelete
  10. Hello, I have also had the same issues with Ancestry DNA giving me (originally) only 2% italian dna first, when my grandfather and his family have come from Firenze and villages around Siena traced back to 1400's. As with others it changed last year to no Italian dna and my dna (other than British & Irish...was French and German.
    I found a way to download my raw data to LivingDNA.com.
    I was so happy when my results came back 25.1% Southern Europe - Italian..this was broke down to 22.6% Tuscan and 2.5% Sardinian. :)

    ReplyDelete
  11. Much Better for me too!

    Results for Suzanne DiVita Louis


    Northern Italy
    85%
    Northern Italy
    France
    10%
    Southern Italy
    4%
    Eastern Europe & Russia
    1%

    ReplyDelete
    Replies
    1. That makes so much more sense. The French is still a bit questionable, but otherwise seems very close.

      Delete
  12. My father's entire family is from dozens of small villages in northern Tuscany and we have lots of family trees documenting these lineages (sourced from local church records) which date back to the 1400's.
    Initially, Ancestry had my Italy DNA as 20% and my French ancestry as 43%. The 2020 update put me at 31% Northern Italy and 18% France, which is better than the prior estimate, but I still don't know where this 18% French comes from since neither my mom nor dad have any French surnames (or localities) in their extensive genealogy family trees.

    ReplyDelete
  13. My experience closely matches yours. Nice to see improvement, but still not exactly there. Another problem is that much of this Italian seems to drop out quickly in subsequent generations, meaning that if your children are tested, the Northern Italy percentage is likely to be minuscule or even not there at all.

    ReplyDelete
  14. I also tested my DNA with 23andme and their latest update has me at 45% Italian with a "highly likely" match for the region of Tuscany (which is pretty accurate given the fact that I’m 50% Tuscan). I also have 8.5% Broadly Southern European (which could account for the extra 5% Italian that I have). My niece (who's my identical twin sister’s daughter) also tested with 23andme. Even though she should be 25% Tuscan, her DNA only showed 12.1% Italian (she also has a "highly likely" match for the region of Tuscany). She has a 5.8% Broadly Southern European (which could account for some extra Italian DNA). As you noted, the Northern Italy percentage does seem to drop out quickly with each additional generation.

    ReplyDelete
    Replies
    1. I have also noticed that 23&Me tends to be more so accurate in that regard. I should be about 12.5% or 1/8 Tuscan as my great-grandmother's parents were both Tuscan. Anyways, 23&Me also shows me as highly likely Italian with a connection to Tuscany. Ancestry on the other hand shows me with 2% Northern Italian & 19% France.

      Delete
  15. Thank you so much for this post! I'm 12% French and am baffled -- I'm pretty much a mutt on the other three sides, but my mom's father and her grandparents are from Southern Italy and I have DNA matches on both sides (Chiodo, Lupinacci), so I am almost certain the biological relationship is correct. However, I only show up 2% Southern Italy. I was curious so I did a bunch of digging on the history and got down a rabbit hole -- and learned all about Gallo-Roman culture. Doing more research on that led me to your blog post! I'd wished I found it first, but I was happy to see that my own search led to where you went! I did my original test on Ancestry, so I am going to load my raw data to a few other sites to see what they make of it. I really appreciate the research you did and also glad to discover it's not just me having this oddball experience! Thanks!

    ReplyDelete
  16. Thank you for this! I have long suspected that our Northern Italian DNA was not being identified but was instead being grouped with French or Swiss DNA. Especially if your family traces back to areas around Trento or Verona.

    ReplyDelete
  17. Have you considered that the French Armies, which occupied Italy and definitely influenced some of the dialects, may influence the dna of people in the region? Italy has had Sardinians, Austrians, French and Spanish occupations in modern history so it is unsurprising that the dna would reflect this.

    ReplyDelete
  18. My Italian DNA from 23andme is now at 52% even though only my father is Italian (northern Tuscan) and not my mother. My mother's DNA (from 23andme) does not show Italian DNA, so this doesn't make sense that I could have over 50% Italian in my DNA.
    My Italian DNA from 23andme went from 12% to 52% in 3 years. They must be adding lots of new DNA sources from southern Europe.

    ReplyDelete
    Replies
    1. Interesting observations. It's kind of cool to see how the companies make progress in their accuracy, even if there are some bumps along the way. I'm sure it's not easy for them to get the formulas precise enough to be accurate for every person.

      Delete
  19. Wow, I am very glad to have found this! my father's maternal grandmother was born to parents from the Lucca province in northern Tuscany. I had always wondered why on ancestrydna I recieved very little Italian & such heavy french ancestry. My father recieved recieved his results and recieved 12% Northern Italy & 12% France, so equal parts Northern Italian & French. After the April 2022 I recieved 5% Northern Italy & 15% France. With the most recent update in August having me at 19% France & 2% Northern Italy! My mother has neither France or Northern Italy, and it's impossible for me to recieve more France% than my father. So clearly, ancestry does indeed have an issue with Tuscan & overall Northern & Central Italian, despite a few minor improvements.

    ReplyDelete
  20. Good observation. Yes, Ancestry seems to have trouble being consistent in this way. I and many of our cousins are should be around 50% Italian (Tuscany), and with the latest updates Ancestry is now recognizing this. But our kids, who should be about 25%, are testing under 10%, with some showing no Italian at all.

    ReplyDelete
  21. That's exactly it. As you mentioned, Tuscan, and from what I've noticed most Northwestern Italian tends to show up as half French & half Nothern Italian, like in my father's case. But again like you mention, the Nothern Italian is descendants is especially read as French. When I originally got my ancestry results back in late 2015, I recieved 12% Iberian and 8% Europe south. My paternal grandmother was born in Cuba to a Tuscan mother & father from a small town in Nothern Spain, so this made sensd. However, since the introduction of the France region, my Italian has been non-existent to low with the various updates. Since ancestry is still able to pick up my 9% Spain despite my great-grandfather being Nothern Spanish, I do wish they were able to detect my equal amount of Northern Italian.

    ReplyDelete

Comments welcome.