On Thu, 16 Nov 2000 14:56:12 GMT, [EMAIL PROTECTED] wrote:
< snip >
> Occam's razor would say that undercount pickups (due to manual
> "discovery" of chad-issue ballots) in statistically greater
> proportion than the overall breakdown of the county
> is due to vote tampering by unknown persons of one political group.
Informed statisticians would suggest that you need to describe the
whole problem. Counties are divided into precincts, and precincts can
be homogeneous even when their counties are not. Some of the problems
are localized.
Please be sure you have separated out the arithmetic errors and
transcription errors in the 10s and 100s columns.
> Combinations of the "elderly effect", position in ballot column
> (.i.e. distance from voter), variations in awl, etc. violate my
> personal principle of checking the simple solutions first.
>
I thought that the mechanical was the simple. "elderly effect" was
not a serious offering, I don't believe. Certain mechanical problems
with machines can be tested after the fact, which makes those
explanations "simpler" to my way of thinking than the untestable
ones.
I don't have great faith in politicians. I have my doubts about how
honest a statewide recount would be, when you might not have
politicians and media staring over the shoulders at every counting
station. But in the media I have seen, there has been very little
suggestion of fraud or manipulation (and most of that has been
pre-election stuff, and bad press for Republicans). It does not look
like the counting has offered much opportunity, though I would want to
ban all amateur magicians.
< Another application of Occam's razor: if the people in your circles
are cheaters of a certain ilk, then you would think of that sort of
solution as simple. >
--
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
