Archiver > GENEALOGY-DNA > 2010-03 > 1269037242

From: Vincent Vizachero <>
Subject: Re: [DNA] OT How many races are there?
Date: Fri, 19 Mar 2010 18:20:42 -0400
References: <><BLU126-DS1116FCD0C524B9A8699569922C0@phx.gbl><><><BLU126-DS78BB86EB03D28062DDCC5922B0@phx.gbl><><BLU126-DS12A9F670ADE3CF93C9CEC5922B0@phx.gbl><><BLU126-DS1277532AD79C302AA7924B922B0@phx.gbl><><BLU126-DS11A44040F56C6978906383922A0@phx.gbl>
In-Reply-To: <BLU126-DS11A44040F56C6978906383922A0@phx.gbl>

There is a division at k=2, k=3, k=4, k=5, k=6, k=7, k=8 . . . .

When you run this sort of analysis, k is an input not an output. If
you choose k=4, then k=4.

The paper Ann linked clearly shows that you can make k equal whatever
you want (limited by the number of sampled populations, of course) and
you'll get a sensible output.

There is nothing special about k=4, except that it looks "right" to you.


On Mar 19, 2010, at 10:10 AM, Raymond Whritenour wrote:

> I think you make my point, Vincent. It's quite extraordinary that
> the division into races you call an arbitrary social construct
> appears, at k=4, exactly as we all supposedly imagined it was. And,
> the point is that we did NOT think there were 20 races, for a
> reason; and I would challenge you (or anyone) to show us where the
> racial boundaries occur at k=20, before peeking at a k=20 analysis!
> Anyone can tell you which k=4 group a non-mixed-race individual
> belongs to just by looking at him or her (or his or her
> photograph). NOBODY can tell you which k=20 group such an
> individual belongs to by the same method!

This thread: