Playing the odds
If there’s anything you learn fast when dealing with autosomal DNA (atDNA) — the kind of DNA where testing works across genders and helps you find cousins — it’s that you’re always playing the odds. The odds that someone you think you might match will agree to test. The odds that you and the other person are really related at all. The odds that, even if you are related, you’ll have enough DNA in common to show up as a match.
It’s a crap shoot.
Fun, yeah. But still a gamble.
And the more you get involved in these tests, the more you realize just how much of a gamble it is. Which is why everybody involved in genetic genealogy says the same thing: test as many cousins as you possibly can.
Take a look at this family chart. All of these folks descend from one known couple (who happen to be my 3nd great grandparents, Jesse and Nancy (Davis?) Fore).
Here’s the who’s who on the chart:
• A is a great great grandson of the common couple through their son George Washington Fore. His is a straight male line: the couple to their son, grandson, and great grandson.
• B, C and D are all full blood siblings to each other, and are the great great grandchildren of the common couple through their daughter Mary “Polly” (Fore) Johnson. So they’re third cousins to A. Theirs is a mixed line: the couple to their daughter to her daughter (the granddaughter) to her son (a great grandson).
• E and F are nieces of B, C and D. E is the daughter of an older sister of the trio and F is the daughter of a younger sister. They’re first cousins to each other, and third cousins once removed to A.
• G and H are full blood siblings to each other, first cousins once removed to B, C and D (whose father was the brother of their grandmother), second cousins to E and F. They’re also third cousins once removed to A.
When we match up the first cousins once removed B, C and D against G and H, we get strong matches all the way across the board both in terms of total DNA in common (the first number, in units called centimorgans or cM) and in terms of the longest common chunk of DNA (the second number, also in cM):
But look at the variations between the three siblings and the male first cousin once removed versus the female first cousin once removed! In every case, each of the siblings has a significantly stronger match with the female than with the male.
Now look at the match-up of the second cousins:
Cousin F is fairly closely matched between the two second cousins with only a few centimorgans difference between her DNA in common with the male and in common with the female. But look at the disparity in the results for cousin E! She’s a much stronger match for her male second cousin G than she is for her female second cousin H — and remember: G and H are brother and sister.
By the time we start looking at third cousins and beyond, the numbers can be all over the map:
By far the strongest match for A is not any of his third cousins but one of his third cousins once removed. She has more than half-again as much DNA in common with A as any of the older generation representatives. And her brother? He doesn’t have enough DNA in common with A even to show up as a match.
E, who is also a third cousin once removed, is next closest genetically to A, and she has about a third again as much DNA as either of her uncles or her aunt, who are closer to A in terms of generations than she is. F has enough DNA in common to match, and beats out her uncle B in terms of the longest common stretch with A.
See what I mean about playing the odds? If we’d only tested G, we wouldn’t have gotten this match at all. Testing G and H would have only improved our odds to 50-50. Testing all four second cousins, the odds go up again, but we’d still have failed to see the match in one out of four tests.
That’s why you test all the cousins you can. If you’re looking for that match, you want to do your best to stack the deck in your favor.