Monday, November 03, 2008

Pulls the win, but doesn't cover the spread

So, it's time for some more electoral predictorization. Let's talk about the point spread on the election.

I think the worst case scenario looks like this:

and the best case looks like this:

That's a pretty wide point spread there, at 68 points; though not as wide as some of the liberal nutbars out there are thinking. They actually believe that half the south and midwest are going Obama... because they are deluded, and believe their own BS.

Significantly though, we're only talking about a 5 state spread here: Nevada, Colorado, Ohio, Pennsylvania, and Virginia.

Also significantly, all are within six points on polling; and if you believe DJ Drummond over at Wizbang (and I do. His analysis is spot on), all are actually within the margin of error on the most trustworthy polling data available... which is clearly not very trustworthy.

The polling companies have been making some arbitrary, and very unusual and ahistoric decisions on weighting factors; weighting democrats likely proportion of voters as 12% to 16% higher than republicans.

Historically speaking this is completely incorrect. Typically the actual spread is 3-4% and in the last 50 years has never been more than 7%.

So what it comes down to is, I think that anything in the 6 point range is currently a tossup, and anything 3 points or less for Obama is likely to break for McCain.

I made this prediction six weeks ago, and I'm sticking to it: