Hello!

I came across an article a couple of years ago about a researcher who
had used statistical models to FIRST predict that the 2000 U.S.
Presidential election would end in a 'statistical-tie'. Looking for
information on this and other statistical models that can be used for
market research.

More specifically, we need a statistical model that could factor in
the behavior or potential response of end users who do not use the
Interent (and therefore would not respond to our online polls) based
on the responses of Internet Users.

Please let me know if you need any clarifications.

Thanks
Pat79

The best predictors are not statistical models, but rather futures
markets.  In last year's election, the UK-based futures on U.S.
elections called almost every race.  As I recall, the accuracy was
100% on the U.S. Senate races.

ron

P.S.  This accuracy led to the Pentagon sponsored "Terrorist
Futures"----while academically sound, the program was killed
politically just days after its initiation.


P.P.S.  In 1963, the stock market started plunging *before* the first
newscasts that President Kennedy had been shot.  Markets seem to know
everything first.