GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2008-09 > 1222194064


From: "Ken Nordtvedt" <>
Subject: Re: [DNA] What shall R1b1c call themselves now?
Date: Tue, 23 Sep 2008 12:21:04 -0600
References: <200809231808.m8NI8slJ016880@mail.rootsweb.com>


----- Original Message -----
From: "Elizabeth O'Donoghue" <>
To: <>
Sent: Tuesday, September 23, 2008 12:08 PM
Subject: Re: [DNA] What shall R1b1c call themselves now?


> Vince, maybe I'm going on too much here, but when counting the allele
> value
> and number of mutations in a database, it's surely not ever an equal
> amount
> over and under the modal. In charts I've seen, it's usually at least
> somewhat lopsided. So how can you 'expect an equal and opposite pull
> somewhere else' consistently to keep the ASD in equilibrium? It would be
> nice to have, but the real world is rarely so accommodating. I think I've
> getting into overload here, but I appreciate your and everyone else's
> comments on the topic.
>
> Elizabeth

Elizabeth, That's what the statistical confidence interval is all about ---
it tells you how much deviation you should expect from the expected value
(average) for the variance due to the fact that mutations are not Swiss
trains, but happen randomly (I was going to suggest how the trains run in
another country, but there is another lady tuned in who would chew me out.
No jest is permitted on this site.).

Forget about variance for a whole sample population, and look at the TMRCA
estimate for just two haplotypes. There is a wide distribution of possible
outcomes (confidence interval), because we know that two haplotypes exactly
G generations apart will not have a precise number of mutational
differences. We can calculate the most likely number, but sometimes more
will happen and sometimes less. Same with the occurrence of up versus down
mutations. We can calculate the expected net balance but also the typical
size of imbalance.

So when we say Variance = M G
that is only the expected or average value of outcomes. Sometimes variance
will be more and sometimes less, and we can calculate the size of the
statistical deviations above and below the expected or average.





This thread: