Xref: utzoo sci.bio:1465 sci.misc:2326
Path: utzoo!attcan!uunet!husc6!linus!gatech!ncar!boulder!pell
From: pell@boulder.Colorado.EDU (Anthony Pelletier)
Newsgroups: sci.bio,sci.misc
Subject: Re: Strange results in Nature article (fallout...)
Summary: The report is in, and nothing resolved.
Keywords: skepticism debunking
Message-ID: <2554@boulder.Colorado.EDU>
Date: 5 Aug 88 20:54:24 GMT
References: <1911@aecom.YU.EDU> <6445@megaron.arizona.edu> <492@metapsy.UUCP> <1930@aecom.YU.EDU> <6086@spool.cs.wisc.edu> <156@oasis.icl.stc.co.uk>
Sender: forys@boulder.Colorado.EDU
Reply-To: pell@boulder.Colorado.EDU (Anthony Pelletier)
Organization: University of Colorado, Boulder
Lines: 68

Well, I just got through reading the Randi et al. report and Benveniste
rebuttle.  Unfortunately, I don't think they really clear anything up.
That comes as little surprise to those uf us who have been insisting that
repeating (or failing to repeat) the observations in the hands of other
scientists is the only way to resolve things.

It was clear tha Benveniste was doing some things wrong and that future
attempts to test the experiments should should include some alterations.
It also seems that the investigators biases were even worse than anyone
would have imagined and, if one were to invoke the same skepticism as
being put to the original results, their claims can hardly be accepted at
face value.

The main thing that troubles me reguarding the results is that some
statistical considerations were ignored.  Benveniste makes reference
to a form of statistical analysis used at INSERM with which I am unfamiliar.
But, the investigating team asserts, and the data seem to suport, that sampling
error was not considered.  The standard error given is far too low for the
experiments the way they are described.  And, along these lines, samples
were sometimes counted several times only if the first count gave unexpected
results.  This could introduce alot of bias and error.

Unfortunately, this legitimate complaint that needs to be addressed
can be obscured by the circus of the rest of the investigation.
If I were convinced, or wanted to be convinced, of the results, I could
find ample reason to toss out the conclusions of the committee.

The biases of the investigators are most apparent from the 4th experiment done
in the presence of the investigators.  The first three were done "open," in
the fashion most frequently done in the lab.  Walter Stewart, one of the
investigators, challenged the experimentor by taking tubes, which had been
diluted in his presence, and distributing them randomly in the mico-titer
plates, so that there was no pattern to be discerned and only he knew the code.
He apparently thought this would show up the experimentor.  In fact, the
results obtained were about the best the lab ever got, showing unusually high
activity and the characteristic sinusoidal fluctuation.  Stewart then declared
the test, of his own design, valueless.  He implyed that the experimentor
must have set up the dilutions in a way that the result was guarenteed,
presumably by sneaking exactly the right amount of anti-IGE into each tube.

The crux of their argument, apart from the their analysis of the experimental
design, which needs to be considered, is a set of experiments set up
in a very amusing blind, involving secret codes wrapped in aluminum foil,
sealed in an envelop and taped to the ceiling (this is science?--oh well,
I guess, if you send a magician, you get a magic show).  The main flaw
with this experiment is that all of the investigators were trusted
implicitly.   They all knew the code and Stewart, while knowing the code,
set up the titer plates.  Given Stewart's behaviour in the case above,
I do not think his impartiality is above reproach.
Stewart, insisted upon complete quiet while he was observing the experimentors
doing the counting etc., so that he would not be distracted and miss some
cheating.  However, while Stewart was setting up the plates, he allowed Randi
to perform magic tricks for the technician who was to be observing
him (Stewart).  Even assuming that he was honest, the results, which failed to
reproduce the phenominon, are suspect because of the circus atmosphere under
which they were obtained.

All in all, this was a remarkably unqualified team to send.  The main thing
that they were qualified to detect, namely fraud, was not evident.
With that out of the way, perhaps some qualified scientists can test the
results.
I, for one, would love to see some qualified people come up with a
reasonable explanation of this.  If the mainstream science community
relies on this kind of circus to refute the findings, the homeopaths
will have ample reason to ignore what we are saying.  It will do nothing
but add further credence to their practices, without good reason.

-tony