AncestryDNA’s new BETA

[Helen7]

But with 46,000+ matches, it is still a little disappointing that there are only sixty odd common ancestor hints in my list of DNA matches.

That's a higher return than me. I have almost 41,000 matches but only 46 common ancestor hints.

 

[Helen7]

Out of my nearly 20,000 matches 44 are shown as having common ancestors (though the figure on the front page is only 12).

Yes, I also have a 'front page' figure (shared ancestor hints) of only 12, compared with 46 showing as having common ancestors. In my husband's case, the corresponding figures are 11 and 28. Am I right that the difference is because the shared ancestor hints rely on you and your DNA match having the same ancestor in your trees, whereas the common ancestor list draws on information across all trees (whether DNA matches or not)?

 

[PhoebeW]

That's a higher return than me. I have almost 41,000 matches but only 46 common ancestor hints.

Yes, I gathered from the numbers shared earlier in the thread and those given elsewhere that my numbers seem to be quite high for a UK tester. Over half my common ancestor matches are descendants of one pair of 4 x great-grandparents - at least three of their descendants emigrated to Utah in the mid 1800s and most of the common ancestor matches are their descendants.

So that’s a source of a large number of fifth cousin matches who have a well documented family tree.

 

[PhoebeW]

Yes, I also have a 'front page' figure (shared ancestor hints) of only 12, compared with 46 showing as having common ancestors. In my husband's case, the corresponding figures are 11 and 28. Am I right that the difference is because the shared ancestor hints rely on you and your DNA match having the same ancestor in your trees, whereas the common ancestor list draws on information across all trees (whether DNA matches or not)?

Yes that’s the theory. The front page figure is from the existing system (before the Beta) isn’t it? My number there was 17 but those tended to be incorrect if you had common surnames in your tree.

 

[PhoebeW]

But not necessarily surprising, given that most of your genetic cousins are so distantly-related (see the table in my Masterclass) that it's very unlikely that the common ancestors will appear in both trees. A typical match will be with an 8th cousin, and even the most experienced of researchers won't know who most of their 512 7G grandparents are. I certainly don't!

Out of my nearly 20,000 matches 44 are shown as having common ancestors (though the figure on the front page is only 12). The fact that you have many more matches than me doesn't necessarily mean that you will have a lot more cousins with shared ancestors - it depends whether the higher number of matches is a result of larger families or endogamy (or a mixture of both)..

I do mean disappointing. I know that some of my relatives testing recently had expected that a good number of their matches would be identifiable. I used the total matches figure as it isn’t particulary easy to gauge anything other than 20cM matches or total matches. The number gives some context at least. I’m guessing that my higher number of matches results from C19th sibling/ cousin emigration to countries where DNA testing is popular now.

I think Ancestry could make it easier to gain more meaningful information on matches. I hope today’s relaxation of the shared matches algorithm is to be permanent. That would be a huge step in the right direction.

 

[peter]

I do mean disappointing.

Although it's always good to find a new cousin, and to be able to validate your records-based research, experienced researchers like you are more likely to knock down 'brick walls' by connecting with cousins who don't have the same ancestors in their tree. After all, 'brick walls' are usually the consequence of missing or erroneous records - gaps which only DNA can bridge.

 

[canadianbeth]

My ThruLines look correct, for the most part. There is one incorrect 4th great-grandfather, who married the mother of my 3rd gg, but the actual father is unknown. There are a lot of names listed as potential ancestors that I know are correct but do not know how to fix that and others I have not completely checked as yet.

I am just getting discouraged though, because right there at the top is my Dad and nothing in his paternal line. Every.single.one. of my common ancestors, all 44 of them, has been matched to a Joyce, Barratt or Bowyer, some further back to different names, but still those lines. One Joyce is back to 5th gg. With an unknown right at the beginning, how do I find a common ancestor of my paternal grandfather?

 

[PhoebeW]

Although it's always good to find a new cousin, and to be able to validate your records-based research, experienced researchers like you are more likely to knock down 'brick walls' by connecting with cousins who don't have the same ancestors in their tree. After all, 'brick walls' are usually the consequence of missing or erroneous records - gaps which only DNA can bridge.

Yes of course. I don’t really expect these Ancestry tools to help with brick walls. Having said that, and with a nod to your previous comment on profile pictures, I did find a potential 8th cousin once removed the other day on Ancestry. I clicked on an unconventional avatar in my list of matches, just to see what it was. The names in the tree were familiar and were already in a research tree of mine.

I have been intrigued by a newspaper article from the 1870s purporting to be a letter dated 1754 to my 5x great-grandfather from someone describing himself as his first cousin. The cousin emigrated to Virginia soon afterwards - what you might call a black sheep emigrant. I wasn’t sure what to make of this letter and certainly hadn’t expected a DNA link. But there it was, a 6.5 cM match to someone whose only documented Welsh ancestor was the author of this letter. Worth another look I think; with the potential to take this particular line back to the early or mid 1600s.

 

[jorghes]

I still have a lot of "Potential Ancestors" in my ThruLines (that being mine, my father's, mother's and paternal grandmother's as those are the ones I use), but then that would be because there is still a lot of doubt in a number of my lines.

Last I checked, a few of the ancestors in my tree were still replaced with "potentials" including the one entitled "Dead End".

I noted the idea of the "Shared Ancestor Hints" - they are from the old system, and Common Ancestors, unlike "Shared Ancestor Hints" does not rely on the exact same person appearing in both trees. However, as I mentioned previously, I have shaky leaf hints (Shared Ancestors) for those who share 6th/7th great grandparents, and Common Ancestors and ThruLines don't tend to show them because they only use your ancestors up to 5th. (I marked all those who had shaky leaf hints under the old system with a note to mention the link, so I haven't lost the information).

I still get the best results (Common Ancestors wise) out of my grandmother's results - the summary results give her 774 "4th cousins or closer" and 32 shared ancestor hints (remembering that four of those are my other DNA tests, so don't really count), and over some 77,000 matches altogether. Last time I attempted to count the results that were marked "Common Ancestors" it was approximately 65. I think it's gone up, but I haven't checked again.
However, my grandmother's ancestry is mixed - only about a quarter is English, the others are Scottish, Ulster Scottish and Ashkenazi Jewish. The majority of her results come from her Ashkenazi Jewish relatives (they are a fairly insular group), that side also has pedigree collapse (so a lot of double relatives!) - two of her 3rd great-grandmothers were sisters. (Ironically I just checked her ThruLines again, and they're still having trouble showing the double set of the same 4x great grandparents correctly - they're displayed properly once and then given as "potential ancestors" the second time, which doesn't make a lot of sense...).
There is a lot of emigration in my grandmother's tree - and a large proportion of those emigrants ended up in the USA - for example, her grandmother's brother emigrated to the US while she emigrated to Australia (after a trial run in the US).

My mother's results, on the other hand, which are purely from the British Isles - almost half and half English and Welsh, show a lot less results - only 9 (really 6) shared ancestor hints, and 305 "4th cousins or closer", and just over 40,000 total results. But I have found a lot less evidence of emigration in my mother's tree (she herself was born in Newport and emigrated) - although ThruLines is attempting to show me that some of her 5x great grandparents lived in the USA, which suggests somewhere, early perhaps, one of her ancestors had a child who emigrated, I have yet to find any information to back it up.
Interestingly, my mother's results give me two "New Ancestor Discoveries", of people my mother shares a lot of DNA with (i.e. a DNA circle where my mother shares a lot of DNA with people who share the Ancestor, but no links) both of whom lived in the USA, it bears more investigation!

 

[PhoebeW]

I think Ancestry could make it easier to gain more meaningful information on matches. I hope today’s relaxation of the shared matches algorithm is to be permanent. That would be a huge step in the right direction.

There was a relaxation of common matches yesterday evening but it didn’t last long. Just to show the difference this could make: my relative, a second cousin once removed, has four common matches with me (above 20cM) but yesterday I could see 40 common matches. Some of those matches are small but there are six “good” matches to me at between 16cM and 20cM, and another 13 at above 10cM. Those six “good” matches include a known fourth cousin and known fourth cousin once removed, a fifth cousin found recently, and another three matches that I may well be able to place now. I think that relaxing the common matches threshold, even to 10 or 16cM, would be fairer to UK testers who are not really able to make much use of the function now.

 

[peter]

I think that relaxing the common matches threshold, even to 10 or 16cM, would be fairer to UK testers who are not really able to make much use of the function now.

It's difficult to get the right balance between providing useful and misleading information - we all have so many distant cousins that most of them are likely to be related to each other completely independently of their connection to us.

Once you get beyond 2nd cousin the amount of DNA shared is not a good indicator of the closeness of the relationship - so setting a lower DNA threshold would be very dangerous (as anyone who has used GEDmatch will know).

 

[jorghes]

Ironically - all of you are getting more accurate ThruLines, mine just went a step in the wrong direction. My great-great grandfather, the father of my illegitimate great-grandfather has now been filled in ThruLines. ThruLines has decided that his step-father, and his step-father's ancestors should be in that spot!

This is completely wrong. My great-grandfather did recognise his step-father as the only father he had known - he put his step-father onto his marriage documents rather than leaving the space blank, and on one census was identified using his step-father's name, but they were the only times this seems to have happened. My great-grandfather used his mother's maiden name as his own surname for his entire life (other than that census), and if he was actually related to his step-father, I couldn't see the reason why he wouldn't simply change it to match that of his half siblings, mother and step-father, but he didn't. (Neither did his older half-sibling who was also illegitimate, I'm fairly sure to a different father).

I didn't mind when ThruLines offered a random, unnamed individual with a surname I'm fairly sure was my missing great-great grandfather's (I'm fairly sure my great-grandfather's interesting set of two middle names actually includes his biological father's name), and left the rest of his ancestors completely out, but I'm now annoyed that for some reason they've decided that his stepfather is a good bet. Too many useless ThruLine potential ancestors to dig through (and since this is my grandfather's father's father, they're all at the start of each section!).

 

[Helen7]

My great-great grandfather, the father of my illegitimate great-grandfather has now been filled in ThruLines. ThruLines has decided that his step-father, and his step-father's ancestors should be in that spot!

I still have a step-parent posing in ThruLines as my ancestor, but his ancestors are not there (probably because they can't be found in anyone's trees) so I can just ignore him.

Whilst I find ThruLines generally useful (especially since the replaced ancestors disappeared, the duplicates have been removed and the missing reinstated), and the 'family trees' are good for showing the connections to several DNA matches at once, I have found some misleading displays of relationships, with the same person shown twice as their own sibling or own cousin as illustrated in the attached screenshots.

In the first one, PR is my husband, and Kay and Chris are both descended from his great-grandfather George's half sister Elizabeth Jane. But the 'tree' shows both their 3x g-grandfather William and 2x g-grandmother Margaret twice, with Kay and my husband descended from one 'branch' (headed bizarrely by a 'half 4th great uncle' William) and Chris from the parallel branch. Whilst the quoted relationships (half 3rd cousin) appear correct, the relationship tree looks odd to say the least!

In the 2nd one, Edith is my mother and Robert is Barbara's father. However, my 2x g-grandfather Daniel and Barbara's 2x g-grandfather Joseph are brothers (sons of Joseph b.1809), not 2nd cousins as shown by ThruLines, and Barbara is my 4th cousin, but is described as a 6th cousin by ThruLines. I have no idea why Edward is shown as having a 'brother' - also called Edward (in a private tree) - to cause the false split in this branch (which has no 2nd marriages/step-parents to cause any confusion as far as I am aware).

I could give other examples - I have quite a lot in my ThruLines. Has anyone else come across this sort of oddity?

 

[PhoebeW]

It's difficult to get the right balance between providing useful and misleading information - we all have so many distant cousins that most of them are likely to be related to each other completely independently of their connection to us.

Once you get beyond 2nd cousin the amount of DNA shared is not a good indicator of the closeness of the relationship - so setting a lower DNA threshold would be very dangerous (as anyone who has used GEDmatch will know).

Thanks, that’s interesting. I do use GEDmatch occasionally and I was looking forward to seeing their new clustering tool. I didn’t know it was dangerous though. Is there advice or discussion on this?

 

[jorghes]

I could give other examples - I have quite a lot in my ThruLines. Has anyone else come across this sort of oddity?

That is strange - I was going to say no, and check to see how ThruLines deals with an ancestor I was descended from twice (they've added my Samuel Isaacs undertaker to the list now, and there are 6-7 DNA matches through him, so I'm happy that's now further confirmed!!) - and noticed that they have actually duplicated someone - not only that, but it's been done three times.

I will give ThruLines credit however, they are noting that both women are great-grandmothers, so that's a plus (since previously they would mark one of them as a "great-aunt" or similar).

These results are from my grandmother, Jean's ThruLines, and she won't appear because as this particular ThruLine is assigned to one of her 3x great grandmothers (Ester), she only shows up under Ester, not Sophia/Feijle, who the one who unfortunately got duplicated. And it seems very much to be confusing some of the DNA relationships here.

So all three entries here, are the same woman, Sophia. However, you can see how all three of them are written in differently with different versions of her name. "Sophia Macks" is not what her married name is, it's a bad misspelling.

Take it down a level with the middle incorrect version of Sophia, and we get this:
Ann is Jean's 2x great grandmother, and she's here twice (Ann married Ester's son).

Over to the correct Sophia, and it's all displaying normally. But look at Mark - he appears above with only one DNA link, and below with 5. Ann here (now correctly marked) only has one DNA descendant, where above she has one different one, and a further 7 (those 7 are most likely my grandmother, my father and siblings and a few closer cousins who have had their DNA tested.) And Charles didn't even show as a child of Sophia on the above screen.

I'm going to have a look at how Sophia's own ThruLine (marked as "potential ancestors") is shown, to see if they duplicate her again, or perhaps duplicate Ester this time!!

EDIT:
Nope, Sophia is still the only one duplicated. However, even on what should be Sophia's ThruLine, Jean only appears under Ester, and those 7 DNA links I thought would include her are 7 more different links.
Go up one more level to Coenraad's father, and he's been duplicated (Coenraad, not his father), by someone who appears normally in the above photos, descending from Coenraad and Sophia. (Sophia is not duplicated when viewing her grandfather's ThruLines entry).

 

[Helen7]

That is strange - I was going to say no, and check to see how ThruLines deals with an ancestor I was descended from twice (they've added my Samuel Isaacs undertaker to the list now, and there are 6-7 DNA matches through him, so I'm happy that's now further confirmed!!) - and noticed that they have actually duplicated someone - not only that, but it's been done three times.

Glad to hear I'm not alone in seeing these anomalies. Also glad to hear you got some DNA matches to Samuel the undertaker, thus confirming the connection we discussed on another thread.

It seems that ThruLines is assigning people from other trees as siblings of our ancestors - instead of actual ancestors - causing the replication. I suppose it is understandable where there are name variations (in my tree, someone has entered one of my ancestors in her married name, and she appears as her own sister) but where the names are the same (as in the case of the 2 Williams in my husband's tree I mentioned above) it seems odd. I have also found the number of DNA matches from the same person varies when viewing ancestors at different levels, which can be quite confusing.

I also noticed something a bit strange in your screenshots: I see Rebecca and Sharyn are shown as '3rd great- granddaughter', the relationship to the chosen ancestor and not to your grandmother. I haven't seen that sort of thing anywhere in my (or my husband's) ThruLines.

 

[peter]

Thanks, that’s interesting. I do use GEDmatch occasionally and I was looking forward to seeing their new clustering tool. I didn’t know it was dangerous though. Is there advice or discussion on this?

I can't comment on the clustering tool, but I certainly haven't found shared matches (at GEDmatch) of any significant value - so far the only DNA site that has helped me knock down any of my 'brick walls' is Ancestry, even though I was using GEDmatch and FTDNA for a long time before I tested at Ancestry.

In the past there have been a few people who have tried to convince me otherwise, but the evidence is somewhat thin on the ground. For me clustering (at MyHeritage) is a red flag which tells me "ignore those matches", and similarly where there are lots of GEDmatch users who share a specific segment of DNA with me (but no other) I take the same view. However I'm able to do this because, so far as I know, I have no ancestors from endogamous communities - a finding that is supported by my low number of total matches (just under 20,000). There will be many for whom the position is not as clear-cut.

 

[peter]

Have you seen this post on the Ancestry blog?

 

[Helen7]

I also noticed something a bit strange in your screenshots: I see Rebecca and Sharyn are shown as '3rd great- granddaughter', the relationship to the chosen ancestor and not to your grandmother. I haven't seen that sort of thing anywhere in my (or my husband's) ThruLines.

Actually, I've now found a couple of these in my husband's ThruLines!

 

[jorghes]

Glad to hear I'm not alone in seeing these anomalies. Also glad to hear you got some DNA matches to Samuel the undertaker, thus confirming the connection we discussed on another thread.

I also noticed something a bit strange in your screenshots: I see Rebecca and Sharyn are shown as '3rd great- granddaughter', the relationship to the chosen ancestor and not to your grandmother. I haven't seen that sort of thing anywhere in my (or my husband's) ThruLines.

I am indeed happy about Samuel the undertaker and his DNA links, as that means that his brick wall stepped back another generation (to his father!).

I have the relationship to the ancestor mentioned a lot, I suppose it may be where Ancestry is unsure of the relationship to the person to whom the ThruLines are for. Perhaps it’s a identification of a relationship without committing themselves to the exact relationship.

Thanks for the link to the blog post Peter, I’m happy to know that Ancestry has been listening.