Reply #21: More nonsense...

[Imajika]

Without even taking more than a moment to look at your latest junk number crunching "research", I can already see serious problems.

"Taking an average of the final 18 national RV polls
(registered voters) and assuming that Kerry would get 67-75%
of the undecided/other vote (as was fully expected, since
undecideds have historically voted for the challenger), he
was projected to win"

Why not use likely voters? Likely voter poll results have always been considered more reliable than using registered voters? From what I recall seeing just prior to the election, a combined average of likely voter polls showed Bush and Kerry either in a deadheat or with Bush ahead slightly.

You can't, with any certainty, assume that Kerry would get 67-75% of th undecided vote. You can try to build a model with those figures, but if the results don't match your predictions you'd have to be willing to go back and determine whether that assumption was just wrong. Many pollsters do not necessarily agree that undecideds would break for the challenger in a Presidential contest at the rate your suggesting.

Your problem is that your "work" is completely unobjective and it renders all of the results worthless. You can post this blather till the cows come home but, outside of Internet message forums where preaching to the choir is the norm, no one is ever going to take any of it seriously.

Imajika