At first I thought there was a discrepancy between the prediction and the actual relationship, but after seeing screen grabs from Jenny's account, I realized that Ancestry has two different categories that each include first cousin; one is "Close Family - First Cousin" and the other is "First Cousin - Second Cousin". Given this opportunity, let's take a look at Ancestry's "First Cousin" categories.
I happen to have a known first cousin in my mother's AncestryDNA account, so I can compare the two categories here. Vi is predicted to be a First Cousin to my mother with a range of 1st Cousin to 2nd Cousin:
|My mother's first cousin at AncestryDNA|
With this explanation:
|Click to Enlarge|
|My mother's first cousin at 23andMe|
There is another category at AncestryDNA that is labeled as Close Family to First Cousin. It looks like this:
With this explanation:
|Click to Enlarge|
This is the category that the Jenny's match was placed, so in this case AncestryDNA's prediction was also correct.
AncestryDNA has not released their category guidelines, but from this I can deduce approximately what the percentage cut-off must be between the two categories (if they are using the traditional matching process that we are familiar with from 23andMe and Family Tree DNA's Family Finder). An aunt would share approximately 25% of her DNA with her niece, first cousins share approximately 12.5% of their DNA, first cousins once-removed share approximately 6.25% of their DNA and second cousins share approximately 3.125% of their DNA. These are variable numbers which are only an average, so some individuals, like my mom and her cousin Vi, will share more and some will share less than expected. AncestryDNA has apparently taken this into consideration since they correctly placed my mom's first cousin Vi in the lower First Cousin - Second Cousin category instead of the Close Family - First Cousin category. Without more comparisons, it is impossible to accurately guess AncestryDNA's upper end cut-off for this category, but at 23andMe my highest aunt/uncle relationship comparison is 27.88%, while my lowest is 21.14%, so they probably include at least 28% in this category and possibly much higher since the next closest genetic relationship level shares about 50% of their DNA - full siblings. AncestryDNA likely takes additional information into account in their predictions such as longest segment length and how many shared segments, but for general guidelines, the categories possibly look something like this:
Close Family - First Cousin Category 28% - 15% in common
First Cousin - Second Cousin Category 15% - 5% in common
If this is close to their actual thresholds, then the Close Family Category could include half-siblings, grandparents, aunts and uncles (as we have seen in Jenny's case) and possibly great grandparents/first cousins who are on the high end of expected shared DNA.
With all of this analysis, I don't want to get sidetracked from the joy of this story. Jenny is absolutely thrilled to find her paternal birth family and they seem to be thrilled to find her! She is very appreciative of AncestryDNA's part in her discovery and, thanks to that aunt who is a member of Ancestry.com, Jenny already has an extensive genealogy for half for her biological family!
Jenny shared her thoughts with me:
I need to give AncestryDNA some credit, they were 100 percent right on this one. Although I still stand by my opinion that their system needs changes, because there still is no way to know if there are errors...They need to release the raw data, not everyone is going to be as lucky as I was, but I also don't want to discourage people from submitting their DNA. The more DNA they receive, the better the results I would presume. They have a great system for the most part, but with no way to interpret it, we have to just have blind faith in Ancestry that our results are correct.
Personally, I am very happy to be able to report this positive flip side to my earlier story because I do not want to discourage anyone from participating in DNA testing. Although not my first choice, with a database that has exceeded 65,000 (judging from recent ID numbers), AncestryDNA is a viable place to search for relatives whether it be for adoption reunification or genealogical purposes. You can bet that we, as a community, will keep badgering the folks at AncestryDNA for our raw data and access to more detailed genetic information, but some will enjoy success with their matching system even as it currently is. In fact, because of their outstanding automated matching combined with their vast collection of already existing family trees, a number of individuals have already reported confirming many common ancestors with their matches at AncestryDNA and, in some cases, broken through decades-old brick walls. These successes could greatly aid in, perhaps, the loftiest dream of our community - to build a universal genetic family tree by "assigning" DNA segments to specific ancestors, but only if AncestryDNA decides to release the genetic data to their customers. Let's keep our fingers crossed that they will soon listen to us, because what a wondrous thing this could be!
[10/26/12: This test is now out of Beta, so you can order it here.]