Is a 12 CM 1 segment match better evidence for a common ancestor if it's otherwise unlikely?

Question

Early in my research, I had guessed that the Samuel Price listed in the 1850 census for Minersville, Schuylkill County, Pennsylvania (my gg-grandfather's brother) was the same person as Samuel Price listed in the 1860 census for Washington, Daviess County, Indiana, based solely on name, age and birthplace. Although every other such naive early-research census match of this type has since been firmly disproved, everything that I subsequently learned about Samuel Price of Indiana is consistent with what's known of the actual brother. I was hoping that, eventually, DNA testing might prove or disprove the hypothesis.

A few years back, a direct patrilineal descendant of the Indiana Samuel Price submitted a DNA sample. Unfortunately, not y-DNA, but autosomal. There was no match, either with myself (4C) or my father (3C1R), although I realize that pairwise matches at this level aren't certain. More recently, I found a 12 cM 1 segment match with a reported descendant of Samuel Price of Indiana (who would be 3C3R to my father). A 12 CM match by itself is only little more than 50% likely to be IBD, but given the geographical disparity between Pennsylvania (where much of my family lived for many generations), and Daviess County, Indiana (where I have no other cousins that I know of, and I almost certainly would), is it reasonable to presume that this match probably isn't just from a common population pool? I should add that AncestryDNA has not turned up any similar matches with other Price descendants elsewhere in the U.S. (or, at least none that didn't have some other, more obvious, connection to one of my ancestral families).

I should also add that the first-mentioned (non-matching) patrilineal descendant also doesn't match second one. The first one also shows no matches with any other of his own paternal cousins, but we don't know that any have submitted samples (it's not a large descendant tree). It's tempting to suspect an NPE in his direct paternal line, but I'm concerned that might be interpreting the evidence to fit the theory.

Answer 1

Autosomal DNA tests are very good at reliably identifying close relatives, i.e. those who are second cousins or closer. But they are not good at identifying more distant relatives where the match is made of smaller segments and the total cM of all segments is not very high.

The reason is that once you get to a 3rd cousin, there starts to be a chance that the two share no DNA at all, and that likelihood of no DNA shared goes up the further the relationship gets.

What this basically means is that you cannot rely on DNA to "prove" a distant relationship.

Okay. So Ancestry DNA says you have 1 segment of 12 cM that match. The first problem as you say, is that there is a 50% that this is not a true match. The second problem is that even if this is a true match, you cannot in any way guarantee that the match is through the common ancestor you believe it is through if you only have that match and do not have any other DNA based evidence to support it.

In other words, how do you know that you're not related through some other ancestor? It is very possible that you are related multiple ways. Maybe 4th cousins through one ancestor, 6th cousins through another, and 8th cousins through 5 different sets of 7th great grandparents. There is no way to tell which ancestor the segment may be from.

So yes. Knowing (or believing) that you are related in one particular way to someone, and you have a small DNA match with them in no way means that the DNA is from that relationship. And you therefore cannot use that DNA match alone as evidence of that common ancestor. Doing so is definitely confirmation bias.

To gather more evidence, you must do a number of additional analysis. One thing you can do at Ancestry DNA is clustering (e.g. the Leeds Method) and ensure that you and your DNA relative cluster with people who also would be related to the common ancestor. That wouldn't be enough to prove the match through the common ancestor, but it could disprove it.

Better would be to analyze that segment itself. Upload it to GEDmatch and get your DNA match to do so as well. Then you can see exactly which chromosome and what positions on the chromosome you match. You should then look to see who you triangulate with on that segment. Do you know how any of them are related? Do those people have the possibility of coming from somewhere along the same ancestral line between you and your suspected common ancestor.

The only way to really narrow down on something that can resemble evidence to support the common ancestor with the segment is to do DNA painting. You will have to look at all of your segment matches near the area of your one match. You will need to have a reasonable number of fairly close DNA matches who you know how you're related to, e.g. 1st, 2nd and 3rd cousins, so that you can "paint" the ones who match and use the ones who don't to identify the lines that the segment doesn't belong to.

This will result in what Jim Bartlett in his article The MRCA Knothole calls "walking the segment back". If you can successfully walk the segment back 3 or 4 generations and it heads in the direction of your suspected ancestor, then you will not have disproved that the segment might be from the ancestor.

So you see, it is possible but will take additional DNA analysis to get to the point that small amounts of DNA can validly be treated as supporting evidence that a distant common ancestor might be the correct one.