Ancestry Matches 7cM and less

[Pauline]

I do not understand why you believe this to be a circular argument.

Because of what Peter has said in earlier posts, that he has used what is known from other sources to influence his view that the small shared segments are both valid and inherited from these particular ancestors. If you use the evidence from source A to validate the evidence from source B, then source B is no longer independent evidence and cannot be used to verify A. Doing so is a circular argument.

I agree that without a documented connection, small shared segments are more likely to be invalid than valid

That wasn't actually what I was saying. On the contrary, the fact that a match is a documented cousin generally has no bearing on the validity or usefulness of small shared segments.

 

[peter]

I'm afraid this still seems to me ominously like a circular argument, and that the two strands of evidence (DNA & documentary) are not being used independently of each other here.

It's not so much a question of how they're used, it's more a question of how the evidence was compiled. Bryman has hit the nail on the head.

Also, don't be too ready to label short DNA matches as spurious - Ancestry's White Paper esrtimates that 50% of 6cM matches and 70% of 7cM matching segments are genuine. It stands to reason that a match is far more likely to be genuine if both parties, working completely independently, have traced their ancestry to the same couple, as in my case.

If the matches were spurious, ie random, they would surely be just as likely to occur between people who aren't known to be related?

I think you quite right in saying that documentary evidence is far from infallible, although it is in my view more likely to be reliable than anything gleaned from distant DNA matches.

Your faith in the record keepers and in the integrity of your ancestors is creditable, but even if you're 95% certain about each generation, by the time you get back 9 generations there's a 37% chance of an error somewhere along the way.

 

[Pauline]

If the matches were spurious, ie random, they would surely be just as likely to occur between people who aren't known to be related?

Isn't that exactly the situation I gave as an example above, with my sister and brother in law having several distant DNA matches with these same families, while having no known connection with them?

 

[peter]

If you use the evidence from source A to validate the evidence from source B, then source B is no longer independent evidence and cannot be used to verify A. Doing so is a circular argument.

Neither is used to validate the other - both sets of evidence stand on their own in terms of proving that two individuals are related. There is no circular argument.

Remember, I'm not trying to prove that a particular DNA segment came from a specific ancestor (or pair of ancestors) - I'm simply trying to prove beyond reasonable doubt that I'm descended from that pair of ancestors. This will be just as likely as it was before if the DNA match is spurious, but far more likely if the DNA match is genuine.

the fact that a match is a documented cousin generally has no bearing on the validity or usefulness of small shared segments.

I assume you are quoting from someone else - whoever wrote that is clearly not familiar with Bayes theorem.

Finally, just a reminder that in the case I described in my newsletter I wasn't talking about a single match with a single putative cousin, but matches between me, my brother, and our 2nd cousin with three Americans who, though quite distantly related to each other, can all trace their ancestry back to John Medley, the son of Francis Medley and Mary Linnett.

EDIT: we don't all match with all of them, but all of us match with some of them.

 

[peter]

Isn't that exactly the situation I gave as an example above, with my sister and brother in law having several distant DNA matches with these same families, while having no known connection with them?

That's one of the reasons why we take DNA tests - to knock down 'brick walls' and find the connections that we don't know about. Contrary to what you are implying, the DNA evidence isn't contradicted what's in your tree and in your brother-in-law's tree, it's merely unproven.

Instead of throwing away the evidence, surely you should be filing it away for possible future use?

 

[Pauline]

I assume you are quoting from someone else - whoever wrote that is clearly not familiar with Bayes theorem.

No, I am not knowingly quoting anyone else here, and yes, I am familiar with Bayes theorem. I come from a family of mathematicians, although I freely admit probability and statistics were never my forte, and I broke the mould by doing my degree in physics.

Instead of throwing away the evidence, surely you should be filing it away for possible future use?

I doubt I will. As advised in your latest newsletter, since these matches (all at 6 or 7cM) don't show up in searches using surnames and places from my ancestry, they are probably unlikely to be of much use in my future research. And since my husband and brother in law, who are not my blood relatives, also have similar matches, it makes it seem more likely that they are just random. (The same point about surname and place searches applies to them too.)

As I said above, it's been an interesting discussion but I think it's time we moved on.

 

[peter]

I doubt I will. As advised in your latest newsletter, since these matches (all at 6 or 7cM) don't show up in searches using surnames and places from my ancestry, they are probably unlikely to be of much use in my future research. And since my husband and brother in law, who are not my blood relatives, also have similar matches, it makes it seem more likely that they are just random. (The same point about surname and place searches applies to them too.)

Yes, it certainly makes it seem more likely that they are just random when you put it that way. But then it assumes that you're not related to your husband, whereas the only question is how closely-related the two of you are (on average we're all 10th cousins or thereabouts).

 

[Helen7]

I have a common ancestor match who I know for sure is a 3rd cousin one removed. He comes in at 13cM across 2 segments, so using this algorithm would I be correct in thinking that even if he doesn't get the heave-ho he is in for demotion?

I was wondering about that too. I have several matches >8 cM with multiple segments, e.g. 24cM across 4 segments. I assume these would disappear if each segment is <8cM? Or be 'demoted' to a lower cM number once the small segments are removed.

 

[peter]

I was wondering about that too. I have several matches >8 cM with multiple segments, e.g. 24cM across 4 segments. I assume these would disappear if each segment is <8cM? Or be 'demoted' to a lower cM number once the small segments are removed.

As I understand it, unless there is a segment of 8cM or more there will be no match. So add a note to those entries just in case.

 

[Pauline]

As I understand it, unless there is a segment of 8cM or more there will be no match. So add a note to those entries just in case.

That's what I have understood the White Paper to say too, but there are a couple of DNA bloggers saying that the changes will only affect matches where the total shared DNA is under 8cM, one of whom had added this in as an update.

Is there any way of getting a definitive answer on this from Ancestry? I am not confident of getting one by contacting customers services.

One of the blogs also mentioned what Andrew raised in his opening post for this thread, that because Ancestry round up matches with 7.5-7.9999 cM, some of those listed as sharing 8cM may also be lost if you haven't already marked them.

Lastly, at the risk of stating the obvious, don't forget to keep checking back at Ancestry until the changes are made - new distant matches may be added to your list every day, some of which you may want to ensure are saved.

 

[peter]

there are a couple of DNA bloggers saying that the changes will only affect matches where the total shared DNA is under 8cM, one of whom had added this in as an update.

Is there any way of getting a definitive answer on this from Ancestry? I am not confident of getting one by contacting customers services.

I'm sure the bloggers are wrong but I'll ask - even if they're right it could only apply to existing matches, not future matches. In the meantime I'm adding notes to all matches of interest whenever the average segment length is below 8cM. Perhaps the more important question for those of us who are managing multiple tests is precisely when in August the change is likely to go into effect.

One of the blogs also mentioned what Andrew raised in his opening post for this thread, that because Ancestry round up matches with 7.5-7.9999 cM, some of those listed as sharing 8cM may also be lost if you haven't already marked them.

Again, I think they're wrong but will check. Since the change was announced Ancestry have only been displaying rounded figures.

 

[Bryman]

Ancestry’s White Paper of 2016 used the same wording to change the cut-off point to 6cM. You can read it here. https://www.ancestry.com/cs/dna-help/matches/whitepaper

Doesn't this suggest that this time will result in a very similar result except for the 8 cM cut-off point in stead of 6cM in the previous instance? Therefore just be guided by the way things are now before the imminent change. Currently, I have at least one match involving 6 cM shared across 2 segments so presumably small segments are included to give a total above the cut-off.

 

[Tim]

I'll be interested to see how this figure changes after it moves from 6 to 8

 

[Helen7]

Therefore just be guided by the way things are now before the imminent change. Currently, I have at least one match involving 6 cM shared across 2 segments so presumably small segments are included to give a total above the cut-off.

On a quick look through my 6cM matches, I spotted at least 7 where the 6cM is shared across 2 segments. As you say, this might suggest that the 8cM matches with multiple segments might be kept this time, but I don't think we can rely on that.

 

[jorghes]

Not sure if my input will help or hinder this discussion about what is considered enough re: centimorgans and the cut off point in regards to 8cM in full or in total, but perhaps something can be made of this:

I was just checking 8cM and under results for my father's test, mainly for Common Ancestor flags and to check that I'd added them to a list - and I came across one match which was "6cM across 3 segments".

I'm not sure when the 6cM cut off was, but this match was colour coded, prior to the roll out of "Common Ancestor" flags, so who knows how it managed to turn up on the list! (and prior to my discovery of this particular branch of the tree, which was a complete brick wall until a year or so ago, thanks to some help from this forum!)

 

[peter]

I'll be interested to see how this figure changes after it moves from 6 to 8

I suspect you'll still have twice as many matches as me

 

[peter]

6cM across 3 segments

That sounds very strange. It might be a result of an earlier update splitting what had previously been considered a single segment.

 

[Tim]

I suspect you'll still have twice as many matches as me

Unfortunately this is all I have close to me, maybe you have more?


My closest match has:
132 cM across 7 segments,
127 cM across 7 segments
125 cM across 9 segments
119 cM across 4 segments
109 cM across 4 segments

and then I get into 4th cousin matches
49 cM across 5 segments

And 2 of my closest matches have no trees.

The 119 cM across 4 segments really interests me, as he has my surname and I have no idea where he fits it. Maybe why he doesn't reply to me?
Ancestry is suggesting he'd quite close.

 

[canadianbeth]

I have 337 in the close matches list. The first one, after my known niece and cousin, has 198/9. I have put him and four shared matches into the unknown group. Three of them also have high numbers. I expect they are all attached to my unknown grandfather, whom I expect I shall never find. I have pretty much given up on him, especially since I have no surname to search. I doubt dumping the 6-7cM folks will make any difference to me, although I do have three possible sixth cousins in the 6cM list and seven also distant ones with 7cM.

 

[peter]

nfortunately this is all I have close to me, maybe you have more?

I have a few more (but not that many after deducting all the ones who tested at my request!).

But bear in mind that 4th cousins only share 14cM on average so most won't show up as a close match at Ancestry, and nor will some 3rd cousins and 2nd cousins once removed - so many of the most useful matches will be hidden amongst the distant matches. In fact, unless you have a very recent 'brick wall' close matches aren't a lot of use - the exception would be if you can persuade your cousin to give you access to her matches..