Mardon <[email protected]> wrote: > [email protected] (J. Hugh Sullivan) wrote: > > > You might comment on this please... > > > > I am the only Sullivan (tested so far) who was predicted to be > > in the R1a1 haplogroup - the rest are in R1b or R1b1. I > > understand that both are (may I say subsets) of R. How likely > > is it that I would differ that much unless there was basis? An > > educated opinion is sufficient. > > I ordered a 37 Marker Y-DNA test from FTDNA and the predicated > haplogroup provided along with those results was Q3. When I paid > for the SNP test (P36+M3-), my haplogroup turned out to be Q. If > you're curious, my results are posted here: > http://www.Erbland.org/results.htm Not that it matters but we match 16 of 37. Q and R are both subgroups of P. > It's my understanding that haplogroup predictions are considered > reasonable because there's a strong correlation between haplotypes > (the Marker results) and haplogroups (the SNP results). I do not > know if this correlation is uniform with respect to different > haplotypes but I suspect not. If that's true, then estimating the > likelihood of a prediction being accurate (for example, 19 times out > of 20) would vary depending on the haplotype. I have no idea about > the actual accuracy rate of haplogroup predictions made from > haplotypes. I don't know that changing from say R1a1 to R would help or hurt because of your next paragraph. > All of this is why I finally ordered the SNP test. Bear in mind that > haplogroups don't mean much for finding relatives. I was just > curious. I'm no expert at any of this but I think that haplotype > information is useful for genealogical purposes because it mutates > over a period of a few hundred years or less. Haplogroups, on the > other hand, were defined tens of thousands of years ago. This makes > these results of very limited use for genealogical purposes. I do > not know if it's possible to have a close haplotype match and a > haplogroup mismatch; I would think not. If you are comparing your > predicted haplogroup with someone else's predicated haplogroup, the > chances of an incorrect prediction are doubled. My suspicion is > that when two people have a closely matched haplotype, but a > predicted haplogroup mismatch, something has gone wrong with the > prediction. > > Regardless of the haplogroup am I not looking for a 35/37 marker > > match, or better, to have any genealogical meaning for me since > > my provable data only extends to 1790? I have genealogies and > > logic extending further back but no provable link. > > I'm not sure why you say this. I think it's the reverse. It's my > understanding that to 'prove' (statistically speaking) a common > relative, tests using more markers are warranted as the common > relative gets closer. It must also be kept in mind that > genealogical Y-DNA haplotype matches 'prove' things only to a > certain level of statistical certainty. They do not have the same > level of statistical certainty as the results of a forensic DNA > test. Yes, my 35/37 match is 40%-60% probable at the 6th-8th generation but he is not a Sullivan which tells me hanky and panky got together sometime. Unless I ultimately match one particular line where apparently no one has yet tested, DNA testing will do me no genealogical good because I already have my genealogy further than anyone else and I have that line as far as anyone - I just can't link to them yet in the 5th to 9th ancestor generation but I'm close. If we all descend from one woman in Africa (current scientific theory I think) we may, or may not, descend from one father. Assuming a single father all groups A-R are just mutations and we are all looking for the same genealogy. > Caveat: DNA genealogy is only part of my family history hobby. I'm > the administrator of the "Erbland DNA Surname Project". I know > nothing about DNA other than what I've picked up as part of my > hobby. > Cheers, Mardon You are obviously "experter" than I. Thanks, Hugh [email protected] (J. Hugh Sullivan)