I wonder if your percentages run higher than mine because of compound segments: short segments that butt up against each other to make it look like one continuous segment. This is FTDNA's explanation for this phenomenon, but I haven't yet encountered a real-life case. I can see it would be more likely in an endogamous population like your Mennonite ancestry. Do you have contact information for some of the 7+ cM cases that are IBS? Ann Turner On Tue, Jan 10, 2012 at 11:38 PM, Tim Janzen <[email protected]> wrote: > Dear Ann and others, > I decided to download the Ancestry Finder matches from 23andMe for > my parents, my wife's parents, my wife, and me. I then sorted the data to > remove the anonymous matches. I then compared my wife's and my matches to > our parents' matches to see if any of my wife's or my matches in Ancestry > Finder were not found in the Ancestry Finder matches for our parents. > Below > is an analysis of the results: > > cMs %IBD %IBS > >10 100 (51/51) 0 > 9-10 87 (14/16) 13 > 8-9 91 (31/34) 9 > 7-8 79 (33/42) 21 > 6-7 67 (65/97) 33 > 5-6 43 (80/185) 57 > > This information complements the FTDNA Family Finder data I recently > posted > at > http://archiver.rootsweb.ancestry > .com/th/read/AUTOSOMAL-DNA/2012-01/13260977 > 07 and your data below. Overall the results are similar between both > analyses except that I got a significantly higher percentage of matching > segments that are IBD between 7 and 8 cMs with 23andMe data than I got with > Family Finder data. Collectively, your data and my data suggest that John > Walden's projections for the percentage of matching segments that are IBD > between 5 and 9 cMs are significantly on the low side. It would appear > that > almost all matches over 11 cMs in length are IBD. It seems prudent to look > for genealogical connections for all matches between 5 and 8 cMs, but we > shouldn't be surprised if we can't find a genealogical connection due to > the > fact that either the connection is too far back in time or because the > match > was IBS in the first place. If other people have two parent/one child trio > data at 23andMe or Family Finder it would be interesting to see their > analyses as well. > Sincerely, > Tim Janzen >