GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2002-11 > 1036537560


From: "John F. Chandler" <>
Subject: Re: [DNA] Forensic Markers
Date: Tue, 5 Nov 2002 18:06 EST
In-Reply-To: guinness@indigo.ie message <a04310102b9ed30ffecca@[194.125.131.87]> of Tue, 5 Nov 2002 01:52:50 -0700


Patrick wrote:
> Yes, all that was possible, but the FSS (database) agreed that it was
> an 'adventitious' match at 7 loci to one of their 70,000 crime scene
> samples. The FSS said a 37m-to-one chance. It was in all the
> newspapers.
>
> They had double-checked at 7 loci, and there was a match between
> Easton and the burglar.

Let's see now. Are you saying that they compared the one crime-scene
sample against 70,000 different results in a database? If so, that
means we've been discussing the wrong measure of probability (or, at
least, I have, and so was the FSS). The chances of matching Easton
specifically would not be the proper quantity to examine, but rather the
chances of matching *somebody* on file purely by chance. This is
intermediate between the one extreme of the generalized birthday problem
and the other extreme of targeting Easton. In this case, the chances
would be 500 to 1. That's still unlikely, but not out of sight,
especially when you consider that dozens of samples per year (or even
hundreds?) would be going through this same process. By this measure, I
would have to agree that 7 loci are indeed too few. 10 ought to be
enough, but a larger nation (like the US) would probably need even more.

John Chandler

This thread: