Poll-based Election Model Gives Kerry a 99% Probability of Winning

[IChing]

The Election Model by TruthIsAll is a comprehensive and unique election analysis model which cuts through the fog of polling. Right now, the model shows that Kerry has a 99% probability of winning the election. That is the number derived from the National poll projection model and the State Poll simulation model. They confirm one another. The methodology is explained in summary and detail on the home page. The graphics are plentiful and illuminating. The polling data input (national and state) are presented in detail, and the analysis as well. This model has been bactested with the 2000 final state voting numbers and was correct to within .04% of the Gore and Bush final percentages.

this is from the democrats.com link to the poll is http://www.geocities.com/electionmodel/

[Lorien]

1. I'm assuming that's OUR TruthIsAll?

(a regular poster on DU). From his keyboard to God's ears....

[Cooley Hurd]

5. That's our TIA...

Her optimism continues to fuel my own...:thumbsup:

[truthisfreedom]

2. interesting, but probably not the right forum. you might want to post

this in the 2004 election forum instead.

[AndyTiedye]

3. I Want To Believe....

...I would sleep a lot better.

But everything about this election is different.
It has already become painfully obvious that ordinary
rules do not apply to this regime.

This is old, but it has accurately foreshadowed the difficulty
we have had picking up undecided voters. It also tempers my
former optimism about undecided voters staying home, and
suggests that most who decide at the end will go for Booosh.
http://www.zogby.com/Undecideds3.htm
We need to step up our GOTV efforts to counteract any late shift
of undecideds to the devil they know.

[HereSince1628]

4. Models build on assumptions....did this one include election theft?

As a population biologist I am a believer in modelling. I am not putting down the method...but assumptions about placing boundaries on what relationships should be in the model can be wrong (and usually are to some extent).

The ability of a model to correctly project data (that wasn't used to build the model) is a good thing, but only if the conditions in reality that the model explores remain within the limits which allow the model's interpolative mechanisms to function reliably. Which is to say extrapolation is a risky method (models _project_ within limits of known relationships which isn't extrapolation)

I doubt that a subroutine on election theft is included in the model.