This is an interesting case which will not be
appreciated by many. The early call for Gore was
due to a statistical fluctuation in the exit
polls. It was an "error" only in the statistical
sense. Exit poll values are fed into calculations
which assume mean values. I suspect that in
future, analysts will pay more attention to
confidence limits.
Paige Miller wrote:
>
> "J. Williams" wrote:
> >
> > It appeared at first it was Gore winning Florida. No, wait. It's
> > Bush. He's the next President. No, now it's too close to call. All
> > of it based on the exit poll consortium used by the networks. This
> > appears to be the reason all the networks got it wrong. To save
> > money, the networks pool their resources and receive one set of
> > predictions. Now, the talking heads are blaming (pick one): bad data,
> > goofy statisticians, precinct fraud, and on and on. As I have
> > indicated in earlier years, these telephone polls and "exit"
> > interviews are very troublesome --- particularly in close races. How
> > many people are willing to detail their "secret" ballot to a stranger
> > whether on the phone or in person? I think there are many who either
> > refuse or ignore such requests. I know I would. It is no one's
> > business how I voted. Precinct history is not etched in stone either.
> > Sampling and the consequent polls have taken a hit. I suppose I can
> > understand how the general public and various courts are concerned
> > about how the Census should be conducted vis a vis sampling instead of
> > actual head counts. Cynicism about sampling and polling will be alive
> > and well after this election---regardless of who wins.
>
> How do you explain how the networks got the other 49 states correct?
>
> --
> Paige Miller
> Eastman Kodak Company
> [EMAIL PROTECTED]
>
> "It's nothing until I call it!" -- Bill Klem, NL Umpire
> "Those black-eyed peas tasted all right to me" -- Dixie Chicks
>
--
Bob Wheeler --- (Reply to: [EMAIL PROTECTED])
ECHIP, Inc.
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
