Multiple AncestryDNA Tests

[muttgenes]

I'm not sure of what everyone's opinion is on this but has anyone taken 2 AncestryDNA tests to compare the results? If you did, how did the results differ? I'm not looking for your opinion on doing this, I'm just simply curious about the findings.

[geebee-1015]

I'm not sure of what everyone's opinion is on this but has anyone taken 2 AncestryDNA tests to compare the results? If you did, how did the results differ? I'm not looking for your opinion on doing this, I'm just simply curious about the findings.

I took both the v1 and v2 versions of Ancestry's test.  (Now I have no idea what version they may be on.)  I eventually deleted my v1 results, but if I remember correctly, one difference was that total match length was reported differently in some cases.  For example, on the v1 test I was reported to share 92 cM in 5 segments with my 3rd cousin "TC".  

On the v2 test, this is now reported as only 42 cM.  However, the real difference was only 3 cM, and then Timber chopped off a full 47 cM of the match; our longest shared segment is reported as 49 cM -- so a bit longer by itself even of the vandalized Timberized total.  Frankly, this underscores what a piece of junk Timber really is.  I might note that Ancestry reports my daughter's match with TC -- who is of course her 3rd cousin once removed -- is 62 cM in 4 segments, with a longest reported segment of 48 cM.  Unweighted sharing is 85 cM.

Obviously, the segment I passed on to my daughter must have had a negative length, since Ancestry reports her sharing with TC is longer than mine is.  And lest you think that perhaps she's related through both parents, my wife shows no sharing with TC on direct comparison.  It's just that Timber somehow sees a significant amount of our sharing with TC as "excess" for me but not for her.

It's also worth noting that I share 109 cM "MB", who is another 3rd cousin with the same MRCA as I share with TC.  TC and MB, however, are not 3rd cousins to each other, but 1st cousins.  It would definitely be interesting to see how much DNA the two share with each other.

Other than some differences in reported sharing with my matches -- which may also be at least somewhat time-related, since I don't know what Ancestry would report now based on that v1 test -- I can't remember any huge differences.

[Jalisciense]

In reddit, there was a guy (Afro-US American) who took 2 AncestryDNA tests a couple of years apart, both v2, and the results were different, in the last one he scored more European, less SSA and like 1% less Native American if I'm not wrong.

[muttgenes]

In reddit, there was a guy (Afro-US American) who took 2 AncestryDNA tests a couple of years apart, both v2, and the results were different, in the last one he scored more European, less SSA and like 1% less Native American if I'm not wrong.

That's part of what I'm curious about. My father and his cousins score a variety of different SSA results in varying percentages as well as some indigenous. We've also tested our children and my wife's family and there are results our children get (like Scottish) that neither of us had but we know exists in both of our ancestries. Since the price was fairly inexpensive for Father's Day I've submitted my test and post my results in comparison once I have them. My first test was V2.

[ArmandoR1b]

What are your complete results and what is the ancestry that you were expecting based on your documented genealogy? Have you also tested at 23andme?

[muttgenes]

What are your complete results and what is the ancestry that you were expecting based on your documented genealogy? Have you also tested at 23andme?

Here are my current results. I'm expecting Scotland, Italian (Sardinian), and Eastern European results based on my documented genealogy. Also possibly other SSA results. I've done 23andme and have Eastern European, Siberian, and North African through their testing. I would also suspect different communities for Appalachian and African American that family members get which correspond exactly with documented history.

[muttgenes]

What are your complete results and what is the ancestry that you were expecting based on your documented genealogy? Have you also tested at 23andme?

Here are my current results. I'm expecting Scotland, Italian (Sardinian), and Eastern European results based on my documented genealogy. Also possibly other SSA results. I've done 23andme and have Eastern European, Siberian, and North African through their testing. I would also suspect different communities for Appalachian and African American that family members get which correspond exactly with documented history.

My second v2 test was done very quickly under and under a different name with no connection to the original test. The results were identical with identical communities. I ran the "cheat" and it's the same down to the hundredth percent.

[muttgenes]

K12 in Gedmatch is different from the initial test before Ancestry did updates but it's identical to the test after the updates they did in the fall of 2023.

[geebee-1015]

I just got my results for a 2nd test.  I'd actually had two sets of results before, for two different versions.  But I'd eventually ending up deleting my results for the v1 kit, which turned out to be the one for which Ancestry considered me the "owner".  So for a while Ancestry stated that I didn't have any DNA results -- since the results I had were for someone whose kit I managed, but who had exactly the same name and DNA.  For the new results, I'm the owner.

Anyway, I haven't been able to do a deep dive yet.  But one thing I've found really interesting is my sharing with my 3rd cousin TC.  In the managed GB kit, Ancestry reports my sharing with TC as 42 cM in 5 segments, with a longest shared segment of 49 cM.  But this is only after Timber lopped off 47 cM of the unweighted amount!  Weirdly, at the same time Ancestry was reporting my daughter's sharing with TC as 62 cM in 4 segments, with a longest shared segment of 48 cM.  Unweighted sharing for my daughter was only a little less than my own -- 85 cM for her but 47 cM for me!  This makes no sense, but that's Timber for you.

So now what does Ancestry report for my sharing with TC?  92 cM, which makes it Timber-safe -- even though it's actually only two cM more than my unweighted sharing had been on the previous test.  But this is not at all a unique case.  Timber often is worst the useless, for all that Ancestry insists on it being some sort of "improvement" over what other companies do.

[muttgenes]

I just got my results for a 2nd test.  I'd actually had two sets of results before, for two different versions.  But I'd eventually ending up deleting my results for the v1 kit, which turned out to be the one for which Ancestry considered me the "owner".  So for a while Ancestry stated that I didn't have any DNA results -- since the results I had were for someone whose kit I managed, but who had exactly the same name and DNA.  For the new results, I'm the owner.

Anyway, I haven't been able to do a deep dive yet.  But one thing I've found really interesting is my sharing with my 3rd cousin TC.  In the managed GB kit, Ancestry reports my sharing with TC as 42 cM in 5 segments, with a longest shared segment of 49 cM.  But this is only after Timber lopped off 47 cM of the unweighted amount!  Weirdly, at the same time Ancestry was reporting my daughter's sharing with TC as 62 cM in 4 segments, with a longest shared segment of 48 cM.  Unweighted sharing for my daughter was only a little less than my own -- 85 cM for her but 47 cM for me!  This makes no sense, but that's Timber for you.

So now what does Ancestry report for my sharing with TC?  92 cM, which makes it Timber-safe -- even though it's actually only two cM more than my unweighted sharing had been on the previous test.  But this is not at all a unique case.  Timber often is worst the useless, for all that Ancestry insists on it being some sort of "improvement" over what other companies do.

Timber is very problematic. I've been able to confirm relatives through gedmatch where we uploaded our AncestryDNA raw data but Ancestry had not found the match or had greatly reduced the relationship.

[geebee-1015]

I just got my results for a 2nd test.  I'd actually had two sets of results before, for two different versions.  But I'd eventually ending up deleting my results for the v1 kit, which turned out to be the one for which Ancestry considered me the "owner".  So for a while Ancestry stated that I didn't have any DNA results -- since the results I had were for someone whose kit I managed, but who had exactly the same name and DNA.  For the new results, I'm the owner.

Anyway, I haven't been able to do a deep dive yet.  But one thing I've found really interesting is my sharing with my 3rd cousin TC.  In the managed GB kit, Ancestry reports my sharing with TC as 42 cM in 5 segments, with a longest shared segment of 49 cM.  But this is only after Timber lopped off 47 cM of the unweighted amount!  Weirdly, at the same time Ancestry was reporting my daughter's sharing with TC as 62 cM in 4 segments, with a longest shared segment of 48 cM.  Unweighted sharing for my daughter was only a little less than my own -- 85 cM for her but 47 cM for me!  This makes no sense, but that's Timber for you.

So now what does Ancestry report for my sharing with TC?  92 cM, which makes it Timber-safe -- even though it's actually only two cM more than my unweighted sharing had been on the previous test.  But this is not at all a unique case.  Timber often is worst the useless, for all that Ancestry insists on it being some sort of "improvement" over what other companies do.

So far I'm seeing that while most of my Timber-safe matches are still Timber-safe, some have experienced rather large shifts in their reported sharing.  For example, there's my 2nd cousin once removed RB who on my old test shares 145 cM with me in 9 segments, with a longest shared segment of 41 cM.  On the new test, her sharing is shown as 100 cM in 5 segments, with a longest shared segment of 37 cM.  That's a reduction of 45 cM!

This does not exactly instill confidence in things like SideView, which is what Ancestry uses to phase our DNA.  Further, how in the world can Ancestry justified using Timber to further reduce sharing -- given the magnitude of shifts that can occur even without using Timber?  Imagine if one of my 130 cM matches underwent a 45 cM downward shift.  That would make the new number 85 cM, which would leave Timber free to reduce it still further.  Keep in mind, my 3rd cousin TC is now reported as sharing 92 cM when Ancestry previously showed 42 cM.  The problem is, the unweighted sharing previously was 89 cM -- so only 3 cM less than the current number, instead of the 50 cM difference courtesy of Timber!

Granted, the 45 cM reduction between RB and me on the two tests can't be blamed on Timber; my point is simply how much worse Timber could have made such a change.

Ancestry needs to stop using Timber, at least as they do now.  Instead of using Timber to justify reporting a "theoretical" number, report the real number using conventional matching techniques.  Timber could then be used to simply "flag" matches that might seem closer than they really are, due to possible excess sharing.

[geebee-1015]

I just got my results for a 2nd test.  I'd actually had two sets of results before, for two different versions.  But I'd eventually ending up deleting my results for the v1 kit, which turned out to be the one for which Ancestry considered me the "owner".  So for a while Ancestry stated that I didn't have any DNA results -- since the results I had were for someone whose kit I managed, but who had exactly the same name and DNA.  For the new results, I'm the owner.

Anyway, I haven't been able to do a deep dive yet.  But one thing I've found really interesting is my sharing with my 3rd cousin TC.  In the managed GB kit, Ancestry reports my sharing with TC as 42 cM in 5 segments, with a longest shared segment of 49 cM.  But this is only after Timber lopped off 47 cM of the unweighted amount!  Weirdly, at the same time Ancestry was reporting my daughter's sharing with TC as 62 cM in 4 segments, with a longest shared segment of 48 cM.  Unweighted sharing for my daughter was only a little less than my own -- 85 cM for her but 47 cM for me!  This makes no sense, but that's Timber for you.

So now what does Ancestry report for my sharing with TC?  92 cM, which makes it Timber-safe -- even though it's actually only two cM more than my unweighted sharing had been on the previous test.  But this is not at all a unique case.  Timber often is worst the useless, for all that Ancestry insists on it being some sort of "improvement" over what other companies do.

Timber is very problematic. I've been able to confirm relatives through gedmatch where we uploaded our AncestryDNA raw data but Ancestry had not found the match or had greatly reduced the relationship.

Yes.  I've seen a lot of matches that Ancestry reports as paternal for my daughter, yet I'm neither in her shared match list nor even show any sharing on direct comparison.  In these cases, it's likely than Timber has reduced my sharing with the match to less than 8.0 cM, and has simply dropped it from my list.

Other times, my daughter has appeared on the shared match between me and someone with whom I share less than 20.0 cM after Timber.  Obviously, my daughter's sharing with the same person managed to stay above that threshold, while mine didn't.  In other instances, she and I will show identical unweighted sharing with someone, but one of us with have a Timber adjustment and the other won't; or the adjustments will be of significantly different amounts.

EDIT:

I keep finding more examples.  There's SK, who is a 2nd cousin once removed.  On the previous test, Ancestry reports our sharing as 53 cM in 4 segments, with a longest shared segment of 52 cM.  But this reflects a Timber reduction of 35 cM, from unweighted sharing of 88 cM.  In the new test, Ancestry says our sharing in 92 cM in 4 segments, still with a longest shared segment of 52 cM.  So, it's only a 4 cM change from the unweighted sharing, but a 39 cM change using Timber's "adjustment".

There's also my 3rd cousin PWB.  Previously, Ancestry reported our sharing as 78 cM -- after a 12 cM reduction by Timber.  Yes, Ancestry reported unweighted sharing of 90 cM, which is supposed to be the "Timber-safe" number.  But what you have to understand is that Ancestry really means 90.0 cM or more.  If a number merely rounds up to 90 cM, it's still fair game.  In the new test, however, Ancestry reports my sharing with PWB as 92 cM.

2nd EDIT:

On the other side of the coin, my matches DF, DB, and MS were all "Timber-safe" on the old test.  Respectively, they shared with me 104 cM, 101 cM, and 95 cM.  On the new test, it's 73 cM (89 cM unweighted), 85 cM (88 cM unweighted), and 55 cM (88 cM).

[geebee-1015]

There's another thing taking the AncestryDNA test again has done for me.  Previously, Ancestry and I disagreed about which side my match WV was on.  They were adamant that he was a maternal match, even though our shared matches were all on my father's side and Ancestry even reported a pair of common ancestors for the two of us.  They are 2nd great grandparents to me -- my father's paternal grandfather's parents -- and 4th great grandparents to WV.  So we're 3rd cousins twice removed.

That suggests to me that I was probably right -- that Ancestry had incorrectly identified the side for the segment that WV and I share.  At least, I think the odds are greater that they're right now than that the were right then.  We actually have a decent chance of being related on both sides, since a fraction of our shared matches are on my mother's side -- though the majority are on my father's side.

[geebee-1015]

Here's a sample of what makes me really doubtful concerning the reliability of Ancestry's matching process:

As I've said in other posts here, I've now taken the same Ancestry test two different times -- separated by a few years.  Even in comparing my very closest of matches, there are significant differences. Take a look at the two sets of results:

• my daughter KB - test 1: 3,468 cM total in 24 segments, longest shared segment 268 cM; test 2: 3,448 cM total in 28 segments, longest shared segment 277 cM
  

• my niece GL, daughter of one of my full sisters - test 1: 2,026 cM total in 53 segments, longest shared segment 119 cM; test 2: 2,018 cM total in 53 segments, longest shared segment 119 cM
  

• my niece SS, daughter of one of my full brothers - test 1: 1,937 cM total in 52 segments, longest shared segment 132 cM; test 2: 1,919 cM total in 53 segments, longest shared segment 133 cM
  

• my niece SJ, daughter of another of my full sisters - test 1: 1,801 cM total in 53 segments, longest shared segment 112 cM; test 2: 1,804 cM in 52 segments, longest shared segment 139 cM
  

• my 1st cousin GWS, son of one of my father's full sisters - test 1: 1,093 cM total in 39 segments, longest shared segment 108 cM; test 2: 1,088 cM total in 38 segments, longest shared segment 108 cM
  

• my 1st cousin DS, daughter of one of my father's full brothers - test 1: 1,048 cM total in 34 segments, longest shared segment 111 cM; test 2: 1,030 cM total in 33 segments, longest shared segment 111 cM
  

• my 1st cousin DB, daughter of one of my father's full sisters (and sister to GWS) - test 1: 1,018 cM total in 34 segments, longest shared segment 73 cM; test 2: 1,011 cM total in 33 segments, longest shared segment 73 cM
  

• my 1st cousin VM, daughter of one of my father's full brothers (and sister to DS) - test 1: 831 cM total in 27 segments, longest shared segment 116 cM; test 2: 850 cM total in 28 segments, longest shared segment 116 cM
  

• my 1st cousin RE, son of one of another of my father's full sisters - test 1: 690 cM total in 29 segments, longest shared segment 75 cM; test 2: 677 cM total in 28 segments, longest shared segment 75 cM
  

• my half 1st cousin DB.1, son to one of my father's paternal half sisters - test 1: 526 cM total in 19 segments, longest shared segment 62 cM; test 2: 515 cM total in 17 segments, longest shared segment 62 cM
  

• my half 1st cousin JH, daughter to another of my father's paternal half sisters (and full 1st cousin to DB.1) - test 1: 350 cM total in 12 segments, longest 100 cM; test 2: 327 cM total in 9 segments, longest shared segment 93 cM
  

Why should the sharing for any of these have changed?  Yet all of them have.  It is likely chiefly the result of Ancestry's SideView making a few different phasing choices this go 'round.  This underscores the fact that SideView isn't perfect.  For my daughter, especially, the "correct" amount of sharing should be easy to determine, since half of her DNA is from me.  The only part that isn't is her mtDNA, which is from her mother and also does not factor in here.

Incidentally, I could have also included in this list some ten 1st cousins once removed, two half 1st cousins once removed, and a 1st cousin twice removed.  The number of these with whom my sharing amount was unchanged?  Just one.