No announcement yet.

the validity of Family Finder matches

  • Filter
  • Time
  • Show
Clear All
new posts

  • the validity of Family Finder matches

    I invite comment on the question of what degree of validity we should attach to our Family Finder match lists.

    Consider how this hypothetical scenario might play out in your match list. Let's suppose your list offers a truly representative sample of your actual family tree. I concede that it probably doesn't because those who buy a Family Finder test more likely constitute a random assortment of whoever's interested for whatever set of reasons. But let's imagine, for the sake of conversation, that your list really does reflect the array of your ancestry and your relatives.

    If that's so, then one fourth of the people on the list would be related, however distantly, to any one of your grandparents. Going back just one generation, one eighth would be related to any one of your great-grandparents.

    The latter possibility is what raises this question in my mind. My principal hope, as I explore my family tree, is to identify one of my great-grandfathers. There's no record of his identity--no name, no place, no information of any kind. So I delve into my four thousand autosomal matches thinking that somewhere among them must be a group of people who are related to Mr. X and therefore related to each other. I've studied the submitted family trees, and I've used the chromosome browser from a number of starting points. So far, I haven't found what I need.

    So I'll concede that the FTDNA algorithms suggest that I'm related to everyone on my match list, even if too distantly to correlate with a paper trail, but I have to wonder if that's really the case.

    Any thoughts?

  • #2
    From what I've understood, I think that you have to take FTDNA's estimated relationship ranges with a grain of salt, I mean they said that one of my siblings could be one of my other siblings' grandparent and that my parents could be my parents or my children! So you are in fact related to your matches, but the odds that one fourth of them would be second cousins is slim to none. Some matches can be several generations back in relationship.


    • #3
      First, the degree of relationship between you and your matches becomes increasingly uncertain as the total shared cM decreases -- and I strongly recommend subtracting "shared" segments less than about 7 cM from the total. The 7 cM cutoff point is arbitrary, different vendors and genetic genealogy web sites (GEDmatch, for example) use different values, but it seems best to place the cuttoff point somewhere in the range of 7 to 10 cM. Then compare this "scrubbed" shared total cM with one of the charts that shows the likely relationships associated with any particular total shared cM value. The one I use, now all over the internet, and the one chart that I always keep next to my computer, can be easily found with a Google search on "DNA Detectives Autosomal Statistics Chart". I have learned to ignore the suggested relationships from FTDNA and other vendors, as they tend to be wildly optimistic for relationships more distant than about first or second cousins.

      Second, discovering genealogical connections with your matches is often very difficult, but with great patience and persistence, breakthroughs do happen. I've had several good ones, over a period of years, using mainly the tools on GEDmatch. The general method is to build up an inventory of verified triangulated segments, and, through careful comparisons with known relatives and application of genetically informed logic, attempt to assign each such segment to a parent, grandparent, etc. The more known relatives you can enlist in this search, the better!

      An important part of the method is to avoid jumping to conclusions! Rather, when trying to work out the exact relationship of a promising match, we want to sketch out all the possible family trees that are consistent with the genetic evidence, and then seek additional evidence that may rule out one or more of the possible trees, until only one tree, possibly the correct one, remains. Then we look for more evidence! The more distant the relationship, the more difficult it is to prove that you have really found the correct relationship. But each time we can prove a relationship with another kit, we have also verified that part of both of our pedigrees.

      Serendipity can happen at any point. Somewhere in the world, a genealogist may happen to have exactly the information you lack -- a record that documents a connection completely forgotten by your immediate family. Staying active in genetic genealogy seems to increase the chances that you will eventually bump into that unknown genealogist!


      • #4
        The number of matches that may be related to a particular grandparent or great-grandparent also depends enormously on whether they were of UK origin and how recently they or their ancestors immigrated to the United States. My mother's grandfather immigrated in 1905 from France and I persuaded one of her 1st cousins to test here. Other than her, I don't know that my mother has any matches descended from relatives who remained in France. She has a couple of French Canadian matches. I don't see any common ancestors, but there may be one way back. Most of her other lines came from France and Germany between 1840 and 1886. She has only a handful of matches that I can trace to these lines. Probably about 85% of her matches are related to her one grandfather who had colonial American ancestors. My fathers parents were Italian immigrants. I have fewer than 1400 matches and most of them are probably related to my one colonial American great-grandfather.


        • #5
          Not sure that I am related to all my FF matches ...
          I have roughly 3800 FF matches.
          I have roughly 1700 FF matches in common with my father, and roughly 1400 in common with my mother.
          From that, either "In common with" does not work, or I am not really related to 700 of my 3800 matches, given that any match I have should be a match to one of my parents, and just some randomness.


          • #6
            The matches that are not assigned as either paternal or maternal are those where there is not enough information to reach a conclusion. And even then, as you also suggested, there is always a possibility that some of your matches, particularly the weaker ones, are not related to you in any meaningful genealogical sense, but are the result of chance combinations of small segments that are common in some ancestral population.


            • #7
              Originally posted by John McCoy View Post
              ...assigned as either paternal or maternal
              In my case, I am not talking about "Maternal" and "Paternal", which I think are only worked out from matches you make in your tree - and I would not expect them to add up.
              I am referring to the "In Common With" filter. All my matches should also be matches of either my mother or father.


              • #8
                Interesting! Clearly the "in common with" filter has some limitations that are not obvious. Or, put another way, the "in common with" filter isn't doing the same thing that the assignment of paternal and maternal kits is doing. I wonder if some of these matches that don't "common with" as expected can be studied further to see if they involve segments that are too short, or known problem areas, or some other feature that sets them apart. Even better if FTDNA could be more explicit about exactly what that filter is doing.