GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2008-11 > 1227907914


From: Al Aburto <>
Subject: Re: [DNA] TRMCA for R1b1
Date: Fri, 28 Nov 2008 13:31:54 -0800
References: <cc9.42b05535.3661a19b@aol.com><ea3bd9560811281206p23edf5f5lf85489417fb4f5fa@mail.gmail.com>
In-Reply-To: <ea3bd9560811281206p23edf5f5lf85489417fb4f5fa@mail.gmail.com>


David,
You use the 30,000 years for the first draft in the sense to put down
the paper in an attempt to indicate that the "Karafet et al paper is not
valid and should therefore be dismissed". You don't actually say those
words but your intent is clear regardless. You indicate a lack of
attention to confidence intervals yet when one actually looks at the
paper one sees in Table 2 that the CT clade is given an age of 68,500
+/- 6000 years ago. They did not actually assume that date but used the
date given in a paper by Hammer & Zagura 2002 as the reference.
Karafet's own study indicates a CT age of 70000 years without a
confidence interval given, but still it is in the range of the Hammer
Zagura 2002 study. Your arguments do not appear valid ... to me anyway
... just my 2 cents ...
Al

> David Faux wrote:
> When the paper was in first draft, the year given for the age of R1 was
> 30,000 years. Somehow by the final draft someone (I don't know how many of
> the authors) had decided upon another date. Well that is inspiring.
> Generally in a paper of this nature the first and last authors are those who
> will have the final say. I know there was disagreement, I have spoken to
> one or more of the authors.
>
> How people can accept Karafet's dating as almost sacrosanct when she admits
> that she is very hesitant about using the C/T proposed date for calibration,
> is way beyond me. Also folks chronically forget about confidence
> intervals. It is much more settling to just accept the number in the middle
> of it all, like the red dot as the "most likely estimate" in the ABDNA
> biogeographical test - but anyone with a stats 101 under their belt won't
> accept a single study's date with a novel dating procedure when the
> author(s) are not terribly confident and the confidence bands are very wide.
>


This thread: