GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2006-09 > 1159662349


From: (John Chandler)
Subject: Re: [DNA] Which mtDNA test for insight into ethnic origins?
Date: Sat, 30 Sep 2006 20:25:49 -0400 (EDT)
References: <20060930090653.43247.qmail@web30502.mail.mud.yahoo.com><451E7639.9050203@scs.uiuc.edu>
In-Reply-To: <451E7639.9050203@scs.uiuc.edu> (message from Doug McDonald onSat, 30 Sep 2006 08:50:49 -0500)


Doug wrote:
> HOWEVER: note that it is imprecise. If it says 10% NA it
> REALLY means that it could, with fairly good odds, be
> anywhere from 0 to 20%. If it says 15% , then it is very
> unlikely to actually be zero.

"Very unlikely" is a big overstatement. The standard deviation for
these numbers is about 15 percentage points. Thus, if it says 15%,
the likelihood of being zero is about 1/6. You could certainly call
1/6 "unlikely", but not "very unlikely".

> Thus, it can with very high probability unambiguously detect
> a single pure NA grandparent if all other ancestors are
> European

Again, "very" is uncalled-for. I wouldn't call it an unambiguous
detection if the measured value came in under 10%, and yet the
likelihood is again about 1/6 that the above situation would
yield a reading of 10% or below.

Bottom line: this test can *probably* detect a grandparent of
a different ethnic category, but it's by no means certain. A
great grandparent would be very uncertain indeed.

John Chandler

This thread: