No announcement yet.

Margin of error?

  • Filter
  • Time
  • Show
Clear All
new posts

  • Margin of error?

    How seriously should I take the possiblity that my results fall within the margin of error?

    Similarly to many of you out there, I am researching a potential non-paternity event. I understand that in order to make the most of DNA testing, I will have to carefully design a series of tests from various individuals at each point in the hypothetical line of descent (i.e., test myself, test someone from the suspected actual patriline, and control against an observation from someone from the proported patriline).

    But yesterday I noticed something which may be significant. At , I discovered an individual, from whose few biographical details I currently know, I suspect may share a common ancestor in the last 100 years. The consistency of these details seem very impressive to me--specific given names, surname, dates of birth, and other specific genealogical details within the last 100 years.

    However, our FTDNA matches are not impressive at all. At 25 loci we are a G.D. of 8, and have different tested sub-clades (me R1b1c, my counterpart R1b1c7). My understanding of a G.D. of 8 at 25 loci is that we are barely assured of belonging to the same species. If it matters at all, we are no more than 1 repeat off at each location (see VCGWS and CDN2X).

    On the other hand, my counterpart's haplogroup does seem eerily consistent with my suspected non-paternity event (e.g., with a surname associated with the Ui Neill modal grouping).

    So my question is: Does it seem more likely that:

    a.) I am merely seeing what I want to see, OR

    b.) conicidentally, there is more than 1 patriline testing through FTDNA who were orphaned in the 1920's in a specific state by an individual with the same name and birth year as my ancestor, and whose ancestry supposedly hailed from the same counties over a period of 200 years? I do not believe that our surname (Mason) is particularly common nor uncommon.

    The only alternatives I see to coincidence are:
    a.) Incredibly fast mutation rates. I doubt it. I have heard that mutation may very with the age of the father. My father was born when my grandfather was 50, I was born when my father was 40. Still, a G.D. of 8 is huge.

    b.) An error in the written genealogy. I have researched this line for the last 20 years, and feel very comfortable with the information I have in the relevant time frame. I cannot realistically imagine a motivation for my counterparty to falsify his information, and, given the recent vintage of some very private events described, don't believe it would be information widespread in the public domain.

    d.) I somehow failed to properly observe the YDNA collection protocol, and in essence 'fouled' my sample. Is such a thing possible?

    c.) I fall into the statistical margin of error. Various websites I have visited suggest that the rate of false acceptance in criminal STR testing is 1/10th of 1%. I think most lay people would accept this as pretty darn accurate. At least for the purposes of pure genealogical research.

    So, ladies and gentlemen, your thoughts, please? You have proved extremely helpful in the past.


  • #2
    Originally posted by Clochaire
    How seriously should I take the possiblity that my results fall within the margin of error? :confused
    The two ysearch IDs you cited do not, as you noticed, almost certainly do not share a common paternal ancestor in the past few hundred years.

    While "margin of error" is not the best term to use (I think you mean "possibility of error"), I gather you wonder whether one or both DNA results might just be "wrong". The odds of that are very low, and I don't see any obvious mistakes in the ysearch entry. So I'd discount that, and the only way to know for sure is to test other relatives (brothers, uncles, fathers) but that is challenging in a NPE case.

    Given the details as I understand them, I don't think you can draw any conclusion except that you are not related paternally to this particular person. It would appear that what you found is mere coincidence.


    • #3
      Thanks, Vineviz, and 2 points

      #1: This is exactly what I would tell anyone else sending the same message as I did. Thanks. You are a valuable sounding board for ideas as I progress forward here. Please keep it up.

      My grandfather's history seems to be the "weak link" in an otherwise well-documented American lineage. I am entertaining several competing hypotheses for the gaps/inconsistencies in the written record for this period.

      Philosophically, I feel that even the most bizarre behaviour should be explanable by reference to social drives that we all experience (e.g., security, afiliation, etc.), and testable by reference to the historical and DNA record. At least within a 100-150 year timeframe.

      I think that's why I have been so insistent in pursuing this "R1b1c7" link. To date, the scenario surrounding my expectation of an "R1b1c7" SNP has been the one the satisfies most of the loose ends in my grandfather's history--except the reported FTDNA results. At the same time, I believe that FTDNA is the most independent source of information I have on a very controversial subject.

      I think FTDNA provides a great, reliable and even necessary service. But I've encountered quite a number of errors and stumbles in the written record during my search of 20+ years. So much so, in fact, that I'd hazard a guess that 2/3 to 4/5 of all published genealogical information is erroneous.

      #2 Given the above, I'm ashamed to admit that I have ordered another Y DNA test, just to cross reference the results. I really do think that the possibility of error is very small,but I can't think of another appropriate response. I continue to investigate the information from my expected match (i.e., "CDN2X") and the written record, but I still have a nagging doubt about my own results.

      I guess we'll see in about 3 months from now.

      Thanks again for your insights.