In article <[EMAIL PROTECTED]>,
Robert J. MacG. Dawson <[EMAIL PROTECTED]> wrote:

>       I think what Radford Neal is saying is that *some* dimpled ballots were
>dimpled by a process other than an attempt to vote, and these will
>divide 50-50. This should be easily determined by comparing the number
>of dimples at null locations on the card with the number that are
>interpretable as Bush or Gore votes. 
>       
>       I think that what Virgil is saying is that *most* dimpled ballots were
>dimpled by attempts to vote and these shoud divide proportionally to the
>valid votes.
>
>       These are not incompatible hypotheses. Moreover, both are probably
>true.  
>
>       As the votes are being added to look for a majority, an admixture of
>ballots of the first type will not affect the outcome; so if the two
>types cannot be distinguished it makes sense (introduces no bias and
>increases information content) to count them all.

That's true, but it isn't relevant to the claim made that one can
conclude that there is no bias in the recount from the fact that the
split in the newly-counted ballots is slightly closer to 50-50 (ie,
less favourable to Gore) than in the original count.  It could be that
it should be shifted by a larger amount toward 50-50, and that instead
it's only slightly shifted because the recount is biased toward Gore.

However, as I said in my original post, one certainly can't conclude
that the count IS biased on this basis.  One really needs some sort of
independent check.

----------------------------------------------------------------------------
Radford M. Neal                                       [EMAIL PROTECTED]
Dept. of Statistics and Dept. of Computer Science [EMAIL PROTECTED]
University of Toronto                     http://www.cs.utoronto.ca/~radford
----------------------------------------------------------------------------


=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
                  http://jse.stat.ncsu.edu/
=================================================================

Reply via email to