GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2007-01 > 1169873543


From: Jonathan Day <>
Subject: Re: [DNA] Genetic distances
Date: Fri, 26 Jan 2007 20:52:23 -0800 (PST)
In-Reply-To: <F6186E5A-3056-49BC-A78D-E2B2DC252656@vizachero.com>


--- Vincent Vizachero <> wrote:

> Sharon,
>
> I think you'll make life easier on yourself if you
> stop trying to
> compare one man to a group. It is always better to
> compare just one
> person to another, not to a group. At the very
> least, condense the
> group down to a single representative haplotype.

That would seem to depend on exactly what it is Sharon
is trying to do, which is unclear to me.

Now, I was horrible at statistics (ok, worse than
horrible), but I seem to recall that multiple tests
suffer a loss of accuracy. However, you are absolutely
right that you can't readily test an individual
against a group.

If there are any statistics experts here, they can
probably tell you the correct methods to use for this
sort of stuff. Digging into my memory, I can say that
this sounds like the sort of thing that would require
what is called a "Two-Way Analysis of Variance", or
something similar, as that deals specifically with the
problem of how to test every possible permutation of
pairs of individuals without losing accuracy.

It has been far too long, and I never did learn enough
when I was in class, for me to be able to say much
beyond that. You'd need to look for someone who knows
this field better than I, and probably for some
free/open-source software that specializes in this
sort of statistical number-crunching.




____________________________________________________________________________________
Yahoo! Music Unlimited
Access over 1 million songs.
http://music.yahoo.com/unlimited


This thread: