Level of False Positives on Ancestry and MyHertiage

[4b2]

There also might be a case where someone is adopted, but not know about it, so they have unwittingly built an incorrect tree.

Interesting story.

Also, that people make errors in their trees. I had a relative on a cluster for the surname Smith. They had Smith ancestry in Liverpool, where several sons of this family moved. But it did not link up. So I looked into it myself and found they'd made an error and that Smith line went back to my Smith line.

[ikas]

4B2 - My impression is same as yours. Anc matches above a certain level are usually through a common ancestor. No doubt Timber helps filter out the matches by chance. Always wonder if it filters out some of the real ones as well.
My experience with MH is different. My impression is there are higer level (>30cM) matches that are not via a common ancestor. I've even seen a triangulated match above 30cM on MH which on Anc is no match and we are both sure MH is wrong. When I posted on that match here there were several replies with similar experiences on MH. Caveat - I uploaded my raw data from Anc to MH so that may be a factor.

[4b2]

Timber helps filter out the matches by chance. Always wonder if it filters out some of the real ones as well.

My experience with MH is different. My impression is there are higer level (>30cM) matches that are not via a common ancestor. I've even seen a triangulated match above 30cM on MH which on Anc is no match and we are both sure MH is wrong. When I posted on that match here there were several replies with similar experiences on MH. Caveat - I uploaded my raw data from Anc to MH so that may be a factor.

I think Timber does probably cut out some. I have a cluster of matches on Ancestry with some of the same people on MyHeritage, and other people related to them. I have about ten tests for that line and they appear as matches in some of those tests, but not others (on Ancestry), but they always appear on MyHeritage.

I also think there are false positives on MH at a higher level. Not sure what point they may really start creeping in on Ancestry. But I feel even matches around 20cM appear to be very relevant.

[Biggles50]

Currently I am not aware of any accurate method of determining False Positives.

Certainly not at the level of cM being quoted.

Yes they can be Grouped.

Yes you can use the basic DNA tools on My Heritage but even then the Triangulation Tool gives spurious results, it certainly has for me with three DNA matches each a known Cousin of each other yet I have an indeterminate result.  Their Grouping Feature is something of a mixed bag in determining relationships and only gives the most likely matches that are probably going to give the desired results.

Ancestry, is useless with any form of Chromosome Browsing.

The Leeds Method is not reliable below 90cM.

As it is my false matches are all in the below 10cM shared DNA level and the four led me on a wild goose chase for a while until I determined that the branches simply should not exist in my tree.

There are probably too many ifs and buts to get meaningful answers

[SouthseaSteel]

To get back to the OP I did find this in some online academic journal FWIW:

"Segments under 10 cM have about a 15% chance of being false (i.e. not inherited from mother or father)
Segments under 5 cM have about an 85% chance of being false"

But as noted here, always fraught to identify "wrong" data

[David Nicoll]

Hi,
   That would gel with what I am finding.
   Pretty certain that sub 20 is a match, and sub 10 proven links in the 1770’s at 8 generations.
   I wonder how much of the difficulty is at at that distance pedigree collapse which is before records is making a real link nearly impossible to prove. I have a few like that, surnames link and DNA match, but very hard to “prove” by documentary evidence.
     

[ikas]

To get back to the OP I did find this in some online academic journal FWIW:

Could you identify the journal or link to the article. Thanks

[SouthseaSteel]

A quick re-research found this
 
https://whoareyoumadeof.com/blog/can-dna-matches-be-false/

but I saw it in some other "more formal" report too which I will endeavour to find .... again

[Biggles50]

A quick re-research found this
 
https://whoareyoumadeof.com/blog/can-dna-matches-be-false/

but I saw it in some other "more formal" report too which I will endeavour to find .... again

A good read thanks for posting.

If only we had better chromosome features in Ancestry it would make life so much better in filtering the DNA matches into a “priority pecking order” for research.

I agree with their writings.