Monday, November 5, 2018

How Accurate Is an Ancestry Quick and Dirty Tree?

Blaine Bettinger posted a YouTube video recently about how he uses a "Quick and Dirty Tree" using an Ancestry Member Tree in Building Quick & Dirty Trees to Identify Genetic Matches.



In the video, Blaine explains that this type of tree should be used only to provide Hints that might lead to a Common Ancestor with one of your AncestryDNA Matches.  Please watch the entire video for instructions on how to do this, and for the caveats and warnings that Blaine explains.

In order to see how this works, I used Unlinked Trees of ten of my DNA Matches that are 4th cousins or closer.  The process found common ancestors in four of the quick and dirty trees, and no common ancestors in the other six quick and dirty trees.  You do need to start the tree with a dead person with a name and, hopefully, birth and death years and places.  And perhaps their spouse.  And then Ancestry will usually suggest "Potential Father" and "Potential Mother" for you for the persons you enter into the tree.

I decided to make just one Quick And Dirty Tree and named it "DNA Skeleton Tree" so that I minimized how many Quick and Dirty Trees I use.  I made the tree Private and not Searchable.  The plan is to enter any number of DNA matches and their ancestry into this tree.

To add a new unconnected line, I can add a new unlinked name of my DNA Match to the existing tree and then go from there back in time finding ancestors.

But then I wondered:  "How Accurate is an Ancestry Quick and Dirty Tree using this method?"  

To find out, I thought "I will start a Quick and Dirty Tree with my own information and see what happens."

I entered myself in my "DNA Skeleton Tree" and I got no "Potential Ancestor" Hints - I think it was because I am still alive.  So I entered my father's and mother's names and birth and death information too.  That worked - it generated "Potential Father" and "Potential Mother" Hints for both of them:


I clicked on the "Potential Father" button for my father, and saw:


It found Fredrick W. Seaver (1876-1942) along with his spouse and children, and asked me "Do you think this is the father of Frederick Walton Seaver in your tree?"  The buttons are "Yes" "No" or "Maybe."  I chose "Yes" even though the name wasn't spelled correctly and the death place was incorrect.

Within 15 minutes I was able to take this Skeleton tree back to my paternal 4th great-grandparents (not all are shown on the image below).


Of the 8 2nd great-grandparents shown above for my Seaver line, only one of them is incorrect.  The mother of Juliet White should be Amy Frances Oatley (1826-1864), the first wife of Henry A. White.  The 2nd wife of Henry A. White, A Taft Winslow White, is shown on the chart above, and the system did not show potential parents for her.  My guess is that if I edit the profile for the mother of Juliet White and add the correct mother, that the system will provide potential parents for her.

As far as I could tell, all of the 14 3rd great-grandparents on the chart above had the correct names.

However, in the 4th great-grandfathers, there were several erroneous names.  One was a typo - "Nathaniel Gates" for "Nathan Gates" but the years and parents were correct.

However, as shown on the screen above, the parents of my 3rd great-grandfather John Richman (1788-1867) of Hilperton, Wiltshire are shown as William Richmond and Hannah Kettle of Yorkshire, which is, I believe, totally incorrect.   The same with the parents of John's wife, Ann Marshman.

How did that information get into this Hint system?  The simple answer is, I fear, that Ancestry has concocted a Big Tree (we've talked about this before - back when the We're Related mobile app was introduced) that has conclusion information in it gleaned from Ancestry Member Trees (or perhaps not?).

In an effort to see if the trees with the wrong parents for John Richman showed up in the Hints for John Richman, I clicked on his name in my Skeleton tree and saw his Facts page:

There are 21 Hints for John, so  I clicked on the "Hints" link and saw:


The "Ancestry Member Trees" Hint said there were 10 trees that they compiled information from to help create the profile.  So I clicked on the "Ancestry Member Trees" Hint and reviewed the 10 entries:

A funny thing happened.  None of the 10 Ancestry Member Trees on the list had William Richmond and Hannah Kettle as the parents of John Richman.  Two of the 10 trees listed parents who resided in Hilperton, Wiltshire in the 1788 time frame.

One of the Hints on the Hint list was a birth record for a John Richman baptized in 1786 in Yorkshire with the parents William Richman and Hannah Kettle.  Another Hint was for a John Richman baptized in 1774 in Somerset.

So to summarize the findings so far, I found one error in the 2nd great-grandparents, none in the 3rd greats (other than the unknown parents of the wrong 2nd great-grandmother) and at least four errors in the 4th great-grandparents.

I will do my mother's side of the family next and see if the errors I found in the "We're Related" mobile app are duplicated in this quick and dirty tree process.  My guess is that they will be.

Because I am including my tree in this DNA Skeleton tree for a number of my AncestryDNA Matches, I can find their relationship to me fairly easily IF I add one of my ancestors to their tree.

Of the ten trees I've tried to make so far from Unlinked Trees of my AncestryDNA Matches using this process, all of them had less than ten named persons in the tree, and some just had one named person in the tree.  For three of them, I was unable to find any records for the named person(s).   For three others, I was able to go back to some of the 4th great-grandparents and beyond, but found no common ancestors even though the matches were supposed to be 4th cousins.  Perhaps I didn't go back far enough, or perhaps some of the "Potential Parents" were wrong, just like in my own Skeleton tree.

My opinion at this point in time is that the methodology is "quick" and "dirty" (big surprise) and that a user of this method needs to be really careful in how s/he uses it.  Errors are probably inevitable back in the 2nd to 4th great-grandparents because of the way Ancestry has put together their unseen but omnipresent "Big Tree."  They really should let researchers provide records and proof arguments to correct the misleading and obviously wrong entries in their "Big Tree."

I'm wondering how successful other researchers have been using this Quick and Dirty Tree process to find common ancestors of their DNA Matches.  Please tell me in Comments or in email.

                    =======================================================

The URL for this post is:  https://www.geneamusings.com/2018/11/how-accurate-is-ancestry-quick-and.html

Copyright (c) 2018, Randall J. Seaver

Please comment on this post on the website by clicking the URL above and then the "Comments" link at the bottom of each post.  Share it on Twitter, Facebook, Google+ or Pinterest using the icons below.  Or contact me by email at randy.seaver@gmail.com.


7 comments:

Dona said...

Proof arguments!!! Oh yes, that would be super swell. But do I think it will ever happen? No, unfortunately, Ancestry doesn't seem to encourage its users to engage in proofing or even basic critical analysis of the hints they're given, which leads to garbage in their trees. And suggestions often aren't received well by tree owners.
And I wonder if there is any benefit from my checking "no" to the hints given to me. Their algorithms for suggesting hints let through absurd impossibilities such as offering an English birth/baptism record from the early 1800s for one of the people in my tree whose birth facts (already supported by other documents) were for Pittsburgh in the early 1900s. Makes the vetting of those hints a longer and harder job than it ought to be, IMHO.

Cferra said...

Here's my take on Blaine's "quick and dirty tree". It can be helpful to a lot of people and at the same time really difficult to do. Let me explain. Making them is easier when you deal with people who have a ton established trees. However, it can be hard to make Q and D trees in various situations. Like if your family goes to a different country fairly early on. It helps to also check where the tree comes from AND you can always go back and edit mistakes you make.

I was a little cautious about making the trees. But, after practicing I found they work as long as you follow along with established and sourced trees. It really just depends. I do agree that you need to be cautious about them. In the end, you can still go back and fix any problems.

Greenhill39 said...

I don't like Hints that refer back to my own sources!

I have a DNA match of 40 Centimorgans that is also a 20 Centimorgan match to my Mother and a 20 centimorgan match to my Paternal Aunt. The Match only had a six person unattached tree. Using the quick and dirty tree method I found a potential link to my mother's family about 5 generations back. The hints and Potential parents stalled out after three generations on some of the branches and I was not able to find a link to my fathers side. My own research on my fathers family only goes back four generations on my Irish branches. I will keep banging on the Q & D stalled branches but I do not want to spend too much time on somebody else's ancestors.


I do not plan on fixing or documenting this Q & D tree. I will slowly build out and down my primary tree with sources to tie into this new branch of my mother's family. I will leave the Q & D tree unsourced and hidden.

Linda Schreiber said...

I have found quick and dirty trees invaluable. I had been using this process a couple of years before Blaine mentioned it. For me, they were always down-and-dirty trees ;)
Of course they needed to be built with a lot of common sense and a pound of salt.
But they work. For me, about half the time. This has been particularly useful because I was in search of a biofather in an endogamous area. I had no large and well-established tree to work from.

Matches who are geographic outliers are gold.
And when you start with a match's five person tree, and 3 of them are private, but you have grandparents with names and dates, it is worth a shot. Lots of the time, I can build back, and find that their gggrandparents were siblings to mine. And that other matches, at the right DNA level, are shared matches, and they are descended from another sibling.
These trees are not really 'trees', and are not proof of anything. They are working explorations.
But they are solid and fairly confident 'hints'.
Combined with other DNA analyses, they can be very powerful.

Diane Gould Hall said...

I am currently in the process of building a few of these Q&D trees. Like you, Randy, and as Blaine suggested, they are private and secret. I don’t know how useful they will be in my research, but I’m willing to try anything that will help break down some of my brick walls. I am completely aware of the cautions and will proceed with those in mind. I'll get back to you if I turn up anything. I sure hope I do.
Thanks for the post.

Marian B. Wood said...

Quick and dirty trees are helpful for thinking out loud about hypotheses. Only once I prove a connection do I add it to one of my public and searchable trees. Before then, the hypothetical tree is private and not searchable, the same as yours. The "green hint" parents are wrong nearly half of the time, so I have lots of blanks waiting to be filled in as I accumulate more info. Q&D trees are not yet ready for prime time, but they're a good tool for testing ideas!

Clorinda said...

I haven't done too much DNA research yet but I've found doing "quick & dirty" trees useful for other types of research too. As long as we keep the limitations in mind, building these trees in Ancestry, Rootsfinder or offline will be useful