GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2010-03 > 1267810144


From: "Ken Nordtvedt" <>
Subject: [DNA] TMRCA Confidence Intervals
Date: Fri, 5 Mar 2010 10:29:04 -0700


A couple days ago I reported finding a relationship between the sigma (statistical confidence interval) for variance-based TMRCA age estimates of individual clades and the difference between the TMRCA and coalescence age estimates. This eliminates the unknown tree-dependent factor from the issue. The relationship is:

SigmaG = SquareRoot { [ G - G* ] / M }

G = Sum i [ r(i) - rf ]^2 / NM is TMRCA when summed over STRs
G* = Sum i [ r(i) - <r> ]^2 / NM is coalescence age when summed over STRs
M is sum of STR mutation rates
rf is founder's assumed STR repeat value
<r> is arithmetical average of STR repeat value

For clades in the 3000 to 4000 years age area, the 2SigmaG values seem to be running about 20 to 22 percent of the G age estimates.

I am modifying my Generations4 excel spreadsheet to add G estimate to the already present G* estimate for clades, so that SigmaG can be estimated.
It will come out as Generations5.

The SigmaG* is being worked on.


This thread: