Archiver > GENEALOGY-DNA > 2011-09 > 1316042901

From: Mike W <>
Subject: Re: [DNA] Interclade: G=11.5 SigmaG=7.5
Date: Wed, 14 Sep 2011 18:28:21 -0500
References: <><BE04498836564CAD9858C5C6B2A6356E@kenPC><A50FBB589C5A418FA4181F87A3595BF5@kenPC><AFFC72827C6F4308A6346EC70B83D8D6@kenPC><><E531D26FA7B34A54B360D74B8B58C262@kenPC>
In-Reply-To: <E531D26FA7B34A54B360D74B8B58C262@kenPC>


You mentioned "downweighting" faster STRs. Is the following calculation akin
to what you mean?

I have a calculation that I call "normalized" variance. The intent is NOT to
calculate TMRCAs and therefore I don't need to apply mutation rates. The
intent is to look at relative variance between subclades of a clade to gauge
relative ages. In this case, I calculate the variance of each indivual STR
for all of P312. I save them and then use them as the divisors against
individual STR variances for subclades or subgroups of P312. I add up these
ratios of subclade variance to (P312) clade variance and compare between
subclades. I've "normalized" so that each individual STR "experiment" has no
more weight than the other.

Is this rational? or is there a better way to total the individual STR

Mike W

---------- Forwarded message ------- From: Kenneth Nordtvedt

.... It is erroneous in any event to include the faster STRs without
downweighting, because their variance of variance begins to grow
quadratically in age rather than linear in age sooner.

This thread: