Sunday, March 19, 2017

Six Siblings - Part Three: Close Matches

Something is wrong with the matches at Family Tree DNA. At least with the matches they call "Close Matches," which is what they define as "2nd Cousin - 3rd Cousin."

For the last year or more, I have been getting project administrator notices like this

that appear to be seriously skewed towards the newer tests. That is, one sibling who tested recently seems to be getting lots of these while others who tested earlier are getting many fewer. I discussed this with Janine at RootsTech and she sent me to the Help Desk people. All they have been able to say - and I paraphrase here - is that different siblings match different ways.

But this evades the question.

Just to make it clear, everyone I am discussing here has the same settings for notifications, but that shouldn't even matter. It's not a matter of notifications, it's a matter of the actual matches.










I ran the numbers for three sets of full siblings - my paternal second cousins Marshal and Lee, my maternal second cousins Sam and Beverly and my own siblings, a group of six. I could have done Aunt Betty and Uncle Bob, but neither of them is what I'd call a recent test.

I looked at the matches through the filter shown on the right.

I removed the known cousins who are third cousins or closer from the numbers below, though it really shouldn't matter.

The absolute numbers are large, but this an endogamous population.

Lee's test was received at FTDNA on 2 April 2014. He has 41 close matches.

His brother Marshal's was received nearly two years later on 11 February 2016. He has 89 close matches.

89 vs 41. Hmmm. More than twice as many. Maybe it's an outlier.

Sam's test was received 12 January 2015 and he has 29 close matches.

His sister Beverly's test was received 5 December 2016, also nearly two years later. She has 95 close matches. More than three times as many as Sam. Another outlier?

My brother Dan's test was received in Houston on 7 January 2017. He has 103 close matches.

The dates and numbers for the rest of us:
  • Sarajoy 8 July 2014, 31 close matches.
  • Amy 9 July 2014, 30 close matches.
  • Jean 31 December 2014, 26 close matches.
  • Judith 5 February 2015, 27 close matches.
  • My own kit was received 30 March 2011 but that was for my Y and MtDNA tests. My first autosomal matches were 13 May 2012. I have 22 close matches.

I'm no statistician but it appears obvious that the recent tests are getting close matches that the earlier ones are not getting. The siblings angle isn't the issue - it's just the way to make the proof argument.

I can't be the only person with results like this. What say the rest of you who have tested full siblings, both recently and say two years ago?

If it's just me, they can try to ignore me.
And if two people do it, in harmony, they can try to ignore both of us.
And if three people do it! ... They may think it's an Organization!
And can you imagine fifty people a day? I said FIFTY people a day.... Friends, They may think it's a MOVEMENT,
How about it, FTDNA? What's wrong and how are you going to fix it?

Previous posts in the Six Siblings series are here and here.

12 comments:

  1. Seems like you're onto something here.

    My father (test completed 1 Aug 2013) 42 close matches
    My uncle (test completed 26 Jun 2013) 38 close matches
    My aunt (test completed 11 Jan 2017) 103 close matches

    ReplyDelete
    Replies
    1. Same pattern. Thank you.

      Delete
    2. Hi, I don't even have any close matches. 295 matches in total. Tested in July 2016. Maybe they've changed the chip between 2013 and 2015?

      Delete
  2. Perhaps they haven't reprocessed the older tests under their new algorithm.

    ReplyDelete
    Replies
    1. Have they even admitted to a new algorithm?

      In any case, they should certainly get everything on the same basis.

      Delete
  3. Could it be all the ancestry transfers are raising the number of matches but not being run against old tests?

    ReplyDelete
    Replies
    1. I'd be surprised. But then I am already surprised!

      Delete
    2. No, I just checked an ancestry transfer, and she's close matches to my father and both of his siblings.

      Delete
  4. I noticed that the two first cousins who tested the latest keep getting many more close matches than the other 4 of us. Here is the information on the 6 of us first cousins (my maternal side)
    My results 5/6/2011 85 close relatives (I think that I may have asked a question that had them redo my results), my sister 6/7/2013 42, the cousins through mom's brother, HP results 2/6/2012 35, his sister SP 7/31/2014 30, the cousins, the cousins through mom's sister, HS 2/4/2016 204 close relatives and SM 2/15/2016 160 close relatives.

    ReplyDelete
  5. Okay, here's a weird one - My father has 2345 matches; he did a YDNA67 12/14/2015, a free atDNA txfr from Ancestry 9/4/2015 and an unlock onthe atDNA 3/28/2016; his full brother who I transferred to FTDNA on 2/17/17 his test from Ancestry that was completed 9/30/2016 only has 241 matches.

    ReplyDelete
    Replies
    1. Here's the answer I got from FtDNA about the discrepancy in the number of overall matches between two male full siblings:
      "Thanks for contacting us. Ancestry's V2 and 23andme's V4 tests analyze fewer and different autosomal SNPs than our Family Finder test analyzes. We were able to make transfers work with our Family Finder product, but because of the SNP discrepancy, Speculative Matches are not available with V2 or V4 transfers. We will be rolling out the ability for customers to order a kit to test for their Speculative Matches (6th-Remote Cousins) within the coming weeks."

      Delete
    2. Let me dumb down the question for them.
      Fred's test results come in today. Fred matches two siblings, John and Jane. Jane has more cMs and longer segments with Fred than John does. But FTDNA calls John a close relative of Fred and Jane a less-close relative of Fred. In what world is this right?

      (I made up these people don't go looking for them.)

      Delete