> Richard Bernstein ... uses the DNA tests and the Neiman Monte Carlo
> simulation to support his paternity belief in his book.
He doesn't just use them for support. It's more than that. As I said
earlier, he calls them two of three "pillars" of paternity proof,
accompanying what he says is the third pillar, historical evidence. Jan
Lewis said much the same in her 2001 WMQ piece. And, as I say -- or, as
I argue -- this statistical "pillar" is not merely flawed; it
contributes nothing.
Also, to clarify concerning the statistical study, it actually involved
two parts: first, Monte Carlo simulation, which is a common
computer-based statistical technique, and second, something called
Bayesian analysis. The latter, sometimes controversial among statistics
practitioners, is a way of applying other knowledge to statistical
results. In this case, it amounts to applying the main threads of the
pre-existing historical proof, such as it is, to strengthen the
conclusion asserted from the Monte Carlo work. People who remember when
"begging the question" meant something other than "raising the question"
will recognize this as begging the question, or maybe just circular
reasoning.
My essay at TJscience.org offers a lot more about the statistical
study's uselessness as paternity evidence. As long as I'm bloviating
about it, I'll add this: for years I wondered, and occasionally asked,
whether statistical experts would be vetting the statistical study.
Gradually it became apparent that this was not going to happen. This
study professing to be scientific was going to remain sequestered in a
humanities journal, unread by scientists. So I called in two friends --
a biostatistician who is a fellow of the American Statistical
Association, and a particle-accelerator physicist who is a fellow of the
American Physical Society and who routinely conducts computer
simulations. My analysis is more than just technical, but its main
technical judgments come from these two scientists -- and might well
interest anyone who has faith that statistical science has even
contributed to paternity proof, not to speak of being one of three
"pillars" of any such proof.
Thanks.
Steve Corneliussen
______________________________________
To subscribe, change options, or unsubscribe please see the instructions at
http://listlva.lib.va.us/archives/va-hist.html
|