Announcement

Collapse
No announcement yet.

New Origins Ethnicity is Incorrect

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    My 23and me is
    European
    99.9%
    Northwestern European
    94.0%
    British & Irish
    63.2%
    French & German
    9.1%
    Broadly Northwestern European
    21.7%
    Southern European
    3.5%
    Balkan
    0.8%
    Broadly Southern European
    2.7%
    Broadly European
    South Asian
    < 0.1%
    Broadly South Asian
    < 0.1%

    My FTDNA myOrigins is,
    European 97%
    British Isles 55%
    West and Central Europe 22%
    Iberia 11%
    Scandinavia 9%
    Middle Eastern 3%
    Asia Minor 3%

    My DNAland is,
    West Eurasian 100%
    Northwest European 87%
    Mediterranean Islander 5.9%
    Southwestern European 3.5%
    North Slavic 1.9%
    South/Central European 1.4%


    I don't see how I got 0% Iberian and Scandinavian at 23andme but I have 9% and 11% here? Also 3% Turkey? That's a fair amount to go undetected at 23andme. I'm not sure what the accuracy of DNAland is but the Mediterranean results include Cyprus so that might say something. I just don't understand how 20% of my DNA was 0 at 23andme or how middle eastern by way of Turkey was undetected as well.
    I'm having gripes with 23andme as well because its very misleading with some of its presentation. It seems the South Asian is very likely genetic noise and disappears once you go over 50% confidence. Their timeline turns out to be ridiculous in its claims. I don't know why they would include something as misleading as that.

    Comment


    • #17
      I just don't understand how 20% of my DNA was 0 at 23andme or how middle eastern by way of Turkey was undetected as well.
      I'm having gripes with 23andme as well because its very misleading with some of its presentation. It seems the South Asian is very likely genetic noise and disappears once you go over 50% confidence. Their timeline turns out to be ridiculous in its claims. I don't know why they would include something as misleading as that.
      The predictions for me at 23andme are provably wrong, not merely questionable or inaccurate. I'm 100% Eastern European, with only two ethnicities; both are Slavic, with no other admixture. I can prove all my ancestral lines, from only two villages in Eastern Europe back to pre-1800. 23andme predicts I am only 75% Eastern European during part of that time frame. Both Ancestry and FTDNA have my reports correct, at 99% Eastern European and 1% Siberian as an ancient possibility.

      23andme's reports are not merely misleading, they are wrong.

      Comment


      • #18
        use GEDmatch, its free. they don't try to guess what your ethnicity is, they simply use the science of the chromosome breakdown to tell you where it is from.

        Comment


        • #19
          Originally posted by JerryS. View Post
          use GEDmatch, its free. they don't try to guess what your ethnicity is, they simply use the science of the chromosome breakdown to tell you where it is from.
          Absolutely. Gedmatch is not in this game for the money. Their service is provided for free and/or for voluntary contribution. It's for those who have a serious interest in genealogy.

          Gedmatch doesn't provide "eye candy" graphs and charts, or try to lure people to their site by providing fallacious ethnicity or regional predictions.

          Comment


          • #20
            worthless

            I have used most of the calculators on GedMatch, as well as two other autosomal DNA companies with their own ancestry estimators. Suddenly with the update last year, ftDNA decided that I am 7% "Sepahrdic Jew", even though both my parents have tested (and ftDNA confirmed that they are my biological parents) and are reported as 0% Jewish ancestry. None of the other test variants I have taken have suggested Jewish ancestry (which I would be excited to learn about, if it were real). The test also now reports that I am otherwise mostly British Isles. This test is a joke and a fraud.

            Comment


            • #21
              Whether it is actually a fraud or even remotely humorous, I can't say, but your observation definitely falls into the category of "this shouldn't happen".

              Ancestry.com issued a "white paper" about their own ethnic origins algorithm (it's on the internet), in which they discussed various ways to test the performance of the improved algorithm. One of the best ways to assess the algorithm turned out to be the test of "consistency": For the case of children whose two parents had also been tested (a large number of cases were gleaned from their own DNA database), the expectation is that no components should be detected in the children that were not also detected in at least one of their parents. Ancestry reported that the frequency of these consistency exceptions was reduced in their new, improved algorithm, but exceptions were still present. Ancestry did not actually present numbers on this effect, as I recall, but only discussed the issue qualitatively.

              If we assume that there is some amount of "statistical error" in FTDNA's algorithm - not an unreasonable idea, I think - , perhaps the occurrence of these exceptions is telling us something about the amount of statistical variation that we should expect. If it is as high as 7 percent even when there is no real Sephardic ancestry, that tells us, roughly, that Sephardic scores as high as 7 percent should be regarded as lacking statistical significance - 7 percent might be real, but it might also be entirely a statistical artifact!

              And that's as close as I can come to understanding the limits of ethnic origins!

              Comment


              • #22
                Your match list is FAR more important than your "ethnicity estimates". I don't know why these companies can't just put that information in big bold letters. Matches > ethnicity estimate.

                Comment


                • #23
                  My results are like nergui's

                  Father 0% Sephardic
                  Mother <2% Sephardic
                  Me 14% Sephardic

                  Amount of shared DNA proves the really are my parents. So the results are false.

                  You'd think they would have used families in their own database to test the reliability of the results before they released the new version.

                  And my mother has way to much Eastern European for someone whose ancestors came from the east of France and the west of Germany, right along the Rhine, and from Luxembourg, Switzerland, The Netherlands, Scotland and England.

                  Comment


                  • #24
                    While I tend to agree with the notion that when both parents have 0 percent and the child has, say, 10 percent, it is more likely that it is the child's score that is anomalous, it cannot be ruled out that the parents' scores are "off" and the child's "accurate". Either way, however, this sort of inconsistency tells us that we need to add very large error bars on ALL of the ethnic origins scores, something like +/- 10 percent.

                    Comment

                    Working...
                    X