Burnt Orange Report

A Burnt Orange Flashback; and, let's do it right this time

By Jim Dallas

I'm once again starting to ponder quantitative election models. My last foray into this field was a couple of years ago; I ended up producing a rather-flawed (mathematically speaking) model that ended up producing a pretty good forecast, although that's probably just a coincidence.

My 2003 prediction got 48 of the 50 states called correctly; but one of the errors was a doozy. Calling West Virginia for the Democrats looks downright embarassing in retrospect:

Still, 48 out of 50 ain't bad for a model which was created by an innumerate slacker like me. In fact, this was far more accurate than my "expert" subjective/qualitative/bullshit/whatever projection I made a week before the election:

I'm intrigued by the folks over at Pollyvote, who did a really, really good job.

I'm going to go back and re-think the whole thing, now that I've got a few years to play with computers.