A new site to aggregate polls, using recursive Bayesian filters, and a round-up of recent polls and their effect on the Senate, Presidential, and Governor races. Along with an interactive web applet.
http://stochasticdemocracy.blogspot.com/
*********Cross Posted at StochasticDemocracy**********
We'll start with the Electoral College today, where Democrats have the most good news:
I've incorporated the polls of the last two days. Consequentially, the Delta tabs on the applet at the top of the site show the change in the electoral map since Wednesday.
A histogram, showing the distribution of possible election outcomes and their probability
A cartogram, where state's are sized proportionally to their electoral votes. Shaded according to Obama's chance of winning
A map showing changes in polling since Wednesday. Shaded according to changes in Obama's chance of winning
The new swing states*, in order from most to least Obama friendly(And with probability of Obama winning in Parenthesis), are Missouri(76%) ,North Carolina(74%),Florida(68%), Indiana(53%), and North Dakota(53%)
After this, Obama's next best states are Montana (23%), Georgia(12%), and West Virginia(10%) [Other state's probabilities can be seen on the applet]
While Florida and Ohio seemed to be tightening on Wednesday, the model seems to have decided that it was just noise in the face of strong polling for Obama.
At the same time, Obama's probability of winning Indiana has shot up from 20% to 53% in the face of multiple polls showing Obama ahead. Perhaps a consequence of the Powell endorsement? (Indiana is a particularly interesting state from a modeling point of view, expect a post on that soon)
A favorable poll was released in Montana showing Obama ahead, but it has a low sample size and a large number of undecideds, so my model is skeptical. But it would be great to see more polling.
Lastly, a poll shows McCain surprisingly surging in Arkansas with a 15 point lead.
Nate, who predicted a dead-even race, must be a bit embarrassed...(In his defense, I had a thankfully unpublished prediction of McCain +5)
So onto the Senate:
Probability of maintaining control of the Senate(>=50): >99.99%
Probability of improving our position in the Senate(>51): >99.99%
Probability of a Filibuster-Proof Majority(>=60): 25.9%
Probability of a Lieberman-Free Filibuster-Proof Majority(>=61): 6.13%
A map of this year's Senate races shaded according to The Democrat's chance of winning
The competitive seats** are, from most to least Obama Friendly:
North Carolina(89%) - Hagan(D) vs Dole
Oregon (78%) - Merkley(D) vs Smith
Minnesota(73%) - Franken(D) vs Coleman
Alaska (65%) - Begich(D) vs Stevens
Georgia (34%) - Martin(D) vs Chambliss
Mississippi (20%) - Musgrove(D) vs Wicker
Kentucky (15%) - Lunsford(D) vs McConnell
Today was a bit of a bad day in the Senate. New polls in Kentucky have cut Lunsford's chances of victory in half.
Finally, Governor races:
There are only two remotely competitive races this year, and Democrats surged ahead today in both of them
North Carolina (93%) - Perdue(D) vs McCrory
Washington (85%) - Gregoire(D) vs Rossi
*defined as states where the probability of an Democratic Victory is between 20% and 80%.
**defined as states where probability of a Democratic Victory is between 10% and 90%
*********Cross Posted at StochasticDemocracy**********