No announcement yet.

Is the average years per variant useful for estimating?

  • Filter
  • Time
  • Show
Clear All
new posts

  • Is the average years per variant useful for estimating?

    I keep running into the question about how to estimate how far back a variant occurred based average years per variant. While I'm sure the average is a correct statistic, it seems dubious to simply multiply it by the number of variants to get an estimate of a prior variant. The number of variants jumps all over the place between different testers, and if you bring in the distances for your matches each of them changes the result as they take their tests. It seems like we need a confidence level like we have for TiP reports or the Shared cM project. Am I missing something?

  • #2
    I concur with your misgivings! There are several competing lines of reasoning involved here. One is the "molecular clock" idea, very popular but rarely verifiable by reference to external standards, which in this case would be a suite of dated bones that happened to inform us about the minimum age of particular Y DNA variants (and thus a second line of reasoning that could be used to calibrate the first). Another component is your observation, entirely in line with research about the genetic control of mutation, that mutation rates should be expected to vary among individuals and lineages. Likewise, there is no reason to expect that each locus should have the same mutation rate. Further, as a random or pseudo-random process, the evolution of Y DNA profiles, even if it can be modeled on a population level using an "average" rate, should be an appropriate subject for the use of statistical methods to estimate a probability distribution (and thus confidence limits in some sense) for the variants already catalogued, even if that distribution might not be applicable to variants that haven't yet been discovered.

    Perhaps the way forward in light of the uncertainties is first to consider what would be "useful": in what genealogical or other context ? for what specific goals? for whom? In other words, what question or questions do we hope the estimated age of a variant will help us to understand?