GEN-MEDIEVAL-L Archives
Archiver > GEN-MEDIEVAL > 2007-08 > 1187197328
From: John Plant <>
Subject: Re: Calculating The Joint Probability Of False Paternity Events [FPE]
Date: Wed, 15 Aug 2007 18:02:08 +0100
References: <c48.16aaca1d.33f47869@aol.com>
In-Reply-To: <c48.16aaca1d.33f47869@aol.com>
wrote:
> <<In a message dated 8/15/2007 1:42:54 A.M. Pacific Daylight Time,
> writes:
>
> You want to talk about a *precise* calculation: I said "about 50%"; you
> said sixty something percent. I still say "about 50%">>
>
> --------------
> 50 is made-up, 60 is made-up, 2 to 5 is made-up
>
> Is that clear yet? These figures have no basis in reality whatsoever.
> Present the raw data.
>
Close.
Except that the "about 50%" comes from my Y-DNA study, with a small
sample set. This has to be contrasted with one of these, one of those,
one of another, one of something else, with hardly anything (if anything
at all) matching, which is what comes out more typically from a surname
Y-DNA study, until eventually one or two are found that match.
My modal Y-DNA results can be explained by a single family with "quite a
low" (about 2%) fpe rate. The *science* book "Human Evolutionary
Genetics" gives rates for various populations of <1% to 30%, with an
"urban myth" amongst geneticists of around 10% - that is my basis for
claiming that "about 2%" is "quite low". I have mentioned a model that
is meaningful in as far as I have sample data from volunteers (which I
have *already* presented in raw form at
http://www.plant-fhg.org.uk/dna.html) which has a modal, single-family
lump and a random other "one of this, one of that, etc." remnant. Both
of these features - the lump and the random remnant - are what is to be
expected from the simple "single family with quite low fpe" model that I
have mentioned.
My name distribution data accumulated with some effort and the efforts
of others (for more details, see
http://www.plant-fhg.org.uk/distrib.html ignoring the pictures there for
the benefit of others) is consistent with this, as I have summarized in
my Nomina 28 paper.
On the other hand, when none (or hardly any) of a sample set of
volunteers for a surname match, this can be explained by a higher fpe
rate or by "multiple surname origins". The latter is a term used by the
traditional surname gurus. I suspect that a lower-end supposition to the
fpe rates (the quoted 2% to 5% ball park figure, not of my making) may
come partly out of surname holders simply preferring to believe that
their name was "multi-origin" rather than believing that there has been
a lot of fpe. More properly however, they should be assessing also the
historical "name distribution" data for their surname which is
relatively hard work and which is ignored by most people, especially
overseas where "their perceived origins" often do not go back as far as
documentary medieval evidence in England unless they are claiming a line
of descent from the royal family, with absolutely no interest in we
lesser mortals.
Any closer?
This thread: