What if we just look at Rasmussen?

I've made a couple little simulations of the Mittens vs. Bronco Bama election, but they're both toys. They're just simple models treating each state as completely independent and not doing anything, really, to the polls. One pretended the polls were systematically skewed 2 points against Romney and even cherry-picked the best recent poll (http://rpubs.com/gzt/mittenssimulation ), one didn't add a bias or cherry-pick (http://rpubs.com/gzt/lazyelection). I'm going to make another toy that wonders what would happen if we just used the most recent Rasmussen. Not every state has had Rasmussen polls, unfortunately, so I will try the best I can. Most of the ones that don't have Rasmussen, though, don't have other pollers.

It looks like Mittens wins 47.28 percent of the time if we only use Rasmussen's latest result.
plot of chunk unnamed-chunk-2

Bronco wins more often than not (as of 11/5, 52.72 percent of the time). However, Rasmussen is definitely predicting a “tighter” election. Of course, Rasmussen is also supposed to be “biased” - what happens if we nudge the results a little to the left? Just 0.5%…
plot of chunk unnamed-chunk-3

Mittens' results degrade to winning only 37.67 percent of the time. Of course, again, this is only a simplistic scenario for fun.

Code is available if you really want it, but it's mostly like the previous pages.