On Tue, 14 Nov 2000 22:02:09 GMT, [EMAIL PROTECTED] (J. Williams)
wrote:
> In today's local paper here on the Space Coast of Florida, an
> elementary school teacher divided her 4th grade language arts class of
> varied abilities into 3 distinct groups of 11 students. Each group
> was asked to vote using the butterfly ballot now being questioned.
> One group was asked to vote for Gore, the second for Bush, and lastly
> for Buchanan. Without exception all the kids marked the ballots
> correctly. A couple of days ago, the newspaper published another
> similar study of 77 elementary school kids again with the same
> results. Interestingly, the paper endorsed V.P. Gore and supports a
- well, it is an unfortunately flawed introduction to the science.
Isn't it so much nicer when examples actually *illustrate* the
problems that every expert will recognize? instead of failing to show
them, because of inadequacy of the experiment? The Canadian trial was
large enough to show the point, just barely.
The first set. There were 33 subjects, but only 11 were in the
condition ("vote for Gore") where an error is created by the layout --
from several examples, it shows at a 5% rate, among people who (say)
may have problems with vision, etc., and are not primed for a quiz on
alertness. 5% gives about a 50-50 chance of seeing it in the test.
Oh, 5% error, that would imply that more than 10,000 Gore-votes were
omitted. (The margin for not-having an automatic recount was about
2900.) So maybe the rate is less than 5%.
--
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/
=================================================================
