GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2010-02 > 1265961517


From: "Lancaster-Boon" <>
Subject: [DNA] unbiased estimations versus margin of error
Date: Fri, 12 Feb 2010 08:58:37 +0100


Dear Anatole

Maybe it also helps, or not...

>Andrew: In the real world my response to the above situation is to say we
need more data...

Anatole: Oh, yes. Always. However, sometimes you do not have such a luxury.
For example, with excavated haplotypes. So, you try to minimize an error,
and put a properly calculated margin of error. Often it still provides you
with a useful information.

Maybe it sounds too obvious to remark, but on the other hand maybe you think
people disagree with you about this, because you mention it relatively
often. So maybe it halps to say that I think everyone agrees with you about
this. We have to work with the data we get.

I think the biggest discussion right now is about how big we should
understand the margin of error to be. We can not give up just because the
data is bad, but we also should not ignore that when the data is bad, as it
usually is, our margin of error goes up.

I think in this context your remark to David Ewing is relevant:

Anatole: In fact, you do not see "back mutations", period. When they are
back, you just cannot see them. How do you know looking at "17" that it was
18 and 17 back again?

Again you are right. We can not take any proper account of things like back
mutations which are invisible to us. All we can do is keep in mind that they
are maybe there and try to make unbiased estimates.

However the potential for such things is real. Ignoring that they are
probably inside your data and messing things like mutation counts up might
not make you method biased, but it might change your understanding of what
your margin of error is.

You might be making the best estimate possible, but of course it can be
wrong. And one of the ways it can obviously be wrong is in the "mutation
count".

Best Regards
Andrew


This thread: