GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2009-12 > 1260920217


From: (John Chandler)
Subject: Re: [DNA] R-U152 and R-L21 on the European Continent
Date: Tue, 15 Dec 2009 18:36:57 -0500
References: <200912120748.nBC7mtLt011441@mail.rootsweb.com><REME20091212155909@alum.mit.edu> <004301ca7c17$a753f5f0$6400a8c0@Ken1>
In-Reply-To: <004301ca7c17$a753f5f0$6400a8c0@Ken1> (knordtvedt@bresnan.net)


Ken wrote:

> [[ But it is expected value (mean-value estimator) which is needed. When 50
> marker haplotypes are used, for example, to make interclade variance age
> estimates, 50 independent age estimates have been made and are combined
> (hopefully in an optimal or near optimal manner) to form a single estimate
> superior to the individual estimates.

Ah, but the estimates are not independent. Their sum is constrained
because it is only the relative mutation rates that are estimated
from the collection of haplotype data. If you are considering only
a subset of the marker set, the constraint is lessened, but it is still
there, and the individual marker rates are thus anticorrelated with all
of the other individual rates.

John Chandler


This thread: