Genealogy from the perspective of a member of The Church of Jesus Christ of Latter-day Saints (Mormon, LDS)

Monday, October 12, 2015

How Accurate are the FamilySearch Family Tree Record Hints?

Not too long ago, FamilySearch.org implemented a feature in the Family Tree of offering "Record Hints." Here is a screenshot showing an individual in the Family Tree with some Record Hints indicated:


As you can see from this entry, my ancestor was supposedly born in Fletton, Huntingdon, England. However, this is not a "standardized" place entry. The standard place name is "Fletton, Huntingdonshire, England." If the place name is not standard, or at least standardized, then the Record Hints are likely less accurate. I standardized the entry so now it looks like this:


Now, I will check the Record Hints. It is important to verify the accuracy of the information you find already present in the Family Tree before attempting attach record hints because the information in the tree may be wrong but match the Record Hint. This means that someone used a wrong record to enter the information in the first place. The first suggested record is not my "Elizabeth Chattell."

This suggested Record Hint is for a family in Manchester, about 120 miles from Huntingdonshire. 


I marked this "Not a Match." I had already found a British Census record for my family in Huntingdonshire. Why did the program suggest a Record Hint for an inappropriate family? The answer to this question involves some of the most complex issues in existence facing family historians and those who develop the programs. My Elizabeth Chattell (or Chappell) was born in Huntingdonshire. She married a James Parkinson, who was also born in Huntingdonshire, about 120 miles from Manchester in Lancashire. My Elizabeth and her husband James, did have a daughter named Sarah born in 1833 also born in Huntingdonshire, not the girl in Manchester born in 1831. Here is the 1841 England, Wales and Scotland Census Transcription of the family from Findmypast.com:


At this point, there is another not-so-obvious issue. The location of the births of two of the children, including Sarah is "Cambridgeshire" not Huntingdonshire. Don't we have the same problem with this being yet another James and Elizabeth Parkinson family? It might help to know that Huntingdonshire is a non-metropolitan district of Cambridgeshire as well as one of the historic counties of England. Huntingdonshire lost its county status in 1974. See Wikipedia: Huntingdonshire. So now we have two questions: why does FamilySearch find a family in Lancashire for a match and why does Findmypast.com show the children born in the modern county of Cambridgeshire?

If I attach the FamilySearch.org Record Hint to my family, it is simply wrong. It is not a reference to my family at all. But on the other hand, if I attached the English 1841 Census record, it is correct even though the name of the birthplace is technically inaccurate (we record the name of the place as it existed at the time of the event). In both cases, to utilize the Record Hint or the search results, I have to know additional information. In the first case, I must recognize that Lancashire is a long way from Huntingdonshire. This fact alone is enough to call the record hint into question. In the second instance, I need to know the history of Cambridgeshire and Huntingdonshire.

Now, was the Record Hint for the Manchester, Lancashire family wrong? Is there something wrong with the program when it suggests this family? Not at all. But the answer does involve the need for the researcher to examine each Record Hint very carefully. But at the same time, this type of situation points out the need to study the information offered carefully in the context of all of the other possible records for any given family.

Assigning a percentage accuracy to a search engine's efforts is helpful in understanding the effectiveness of the process but it is also misleading, if the researcher relies on the stated accuracy rather than using his or her own judgment and looking at the records in the historical and geographical context. 

5 comments:

  1. I had an ancestor who has had the same wrong records attached to him twice now (by different users), and I realized it was because of the record hints, so I marked them as Not a Match. If I mark a record hint as "Not a Match," does it disappear for everyone or just for me?

    ReplyDelete
    Replies
    1. Hello RT and M, For everyone. Thanks for the comment.

      Delete
  2. People can undo the Not A Match hints, and attach them again. Your best remedy is to use the new User-to-user messaging function. Find the name of the person who attached the incorrect hint, and send the a message as to why it is wrong.

    ReplyDelete
  3. When I was doing the Beta I kept getting the wrong person at the top of the screen, unrelated to the 2 screens below (record and family tree person). It kept happening so I got out of the Beta.???

    Also, will there be a way to delete some of the many duplicates in the Sources from Hints on the person page? I thought we had to add each duplicate hint or it would mess up the Hinting system?

    Thanks for discussing Hints, I think it will be great when the bugs are fixed!

    ReplyDelete
    Replies
    1. You are correct in your statement that all correct hints should be attached, even if they appear to be or are duplicate records (same image). If the URL for the hint is an exact match FamilySearch folks want to know. If it's the same record with a different URL, this is the result of some images being filmed and/or recorded more than once. Discussion is going on to see if some of this can be eliminated but I doubt it's at the top of the list of to do's.

      If you mark as Not a Match a record that is for that person you are 'training the dog' incorrectly and it will not learn to find the right information.

      Delete