GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2003-02 > 1044476100


From: "John F. Chandler" <>
Subject: Re: [DNA] mutations and paper trail
Date: Wed, 5 Feb 2003 15:15 EST
In-Reply-To: pcramwinckel@wanadoo.fr message <00d301c2cd30$0dbb53e0$ae1ffea9@vaio> of Wed, 5 Feb 2003 08:54:25 -0700


Pieter wrote:
> Based on latest results in my project( see previous message) I do think that
> Mutatons in Y DNA testing are more frequent that the famous rule of thumb 1
per
> 500 generations (per marker) makes us believe.

The above is, or perports to be, a statistical statement. However, the
reasoning behind it turns out to be what we in the business call BIASED
statistics. Any time you focus on one particular case, you magnify
the effect of the normal fluctuations. When the average rate of a
randomly-occurring event is, say, 1-in-500, and you examine the result
of 200 "tosses", you *know* that the answer will not be 0.4 (because,
by assumption, an event either happens or it doesn't). The most likely
result is 0, but 1 is not so very far behind, and 2 is not out of the
question. (The probabilities are 67%, 27%, and 5%, respectively.)

Bear in mind that there are hundreds of family DNA projects going on.
If you want to make a statistical statement, you have to collect the
statistics first and not rely on just one or two instances.

> But I would like to hear from particpants in this mailing list with proven
paper
> trail were the mutations did occur and therefore were higher than this rule
of
> thumb.

If your goal is to find some shared characteristic that indicates a
possible CAUSE of mutations, then you are doing exactly the right thing.
If, however, you want to prove that mutations occur more frequently
than the current best-guess average rate, you have to look at ALL the
cases, not just the ones that fit your theory.

John Chandler

This thread: