No announcement yet.

Evaluating a Find

  • Filter
  • Time
  • Show
Clear All
new posts

  • Evaluating a Find

    Having found serious errors in any type of record, I conclude that there is little certainty in genealogy. So, I have adopted a binary approach to evaluating an ancestoral find. Any item I find is just a clue with an equally likely chance of being correct or incorrect. For instance, if I find a record of what appears to be for an ancestor in the 1830 US Census, where the name, time period and location are feasible, how confident should I be that this record is actually for my ancestor? Assuming that each clue (name, time period, location) has a 0.5 probability of being correct, I can calculate the joint probability that this record is for my ancestor or

    1- (0.5 x 0.5 x 0.5) = 0.875.

    Now if I have a DNA match that ties into the same lineage, I can add that as another clue and improve the confidence I can have in the find. Adding the DNA match gives me a new level of confidence of being correct

    1- (0.5 x 0.5 x 0.5 x 0.5) = 0.9325.

    Comments invited.


  • #2
    One problem I can think of is the common occurrence of multiple people with the same name existing in the same time period and location; it's also quite possible for people to share all of those features and also be of the same direct lineage, yielding matching DNA results. When looking through genealogical records, I search for corroborating pieces of evidence such as marriages, number & names of children, death dates and places, etc. to ascertain someone's identity.


    • #3
      I live in Pittsburgh, and have ancestors who lived here with somewhat common names. However, some of them lived here when it was much smaller (such as 10-30,000 people as opposed to several hundred thousand). So in a strange way, sometimes the search for the further back people was easier in some ways, since there were much fewer people. It also helps to know how large a place the person was living at a specific time, and also depends on how rare or common the name was, as well as the same with occupation etc.


      • #4
        Only an Example

        Vinnie and Lincoln

        I agree with what both of you say. I too look for more clues than just the three mentioned, but at times that's all that is available. I used a simplistic example to demostrate the concept.

        Another refinement might be to assign different probabilities of corectness to records. You might say the probability a birth certificate is incorrect is 0.3 instead of 0.5, but that sure complicates the problem.

        Also, if I have a lot of clues (name, spouse, children, location, age, etc.) why bother with the calculation? You know you have it right.



        • #5
          Sherlock Holmes Said...

          Sherlock Holmes:-"Observe only the clues. Wherever they lead, that is the conclusion, however improbable..."

          (Until added, differently-pointing clues turn up.)

          So when I wrote a book about my family name, I described the references to it found in nonfiction or historical records, which was much more interesting than simply pursuing legalistic genealogy.