[SSS Two]

How does the fact that each state contest is not an independent variable fit into your analysis? Many of the states are highly correlated.

If McCain wins California, he doesn't need it -- the election is a blowout. Yes, there's a 12% chance McCain win California, but if he wins California, he'll also win Pennsylvania, Michigan etc. Because the correlation is high, I'm not sure taking 12% of 55 ev's is very meaningful.

[Political Junkie Too]

My first thought is to say that this is not modeling the election, it is modeling InTrade's betting on the election. At InTrade, each state is an independent auction. Right now, many of the states are in the "safe" category, with probabilities in the extreme ends of the range (<=15% or >=85%).

Taking 12% of California's 55 EV's is the basic calculation, but the probability of winning is based on calculating the S-curve over 20,000 random simulations and seeing where the 270-point is. If McCain wins California, that will raise his probability of winning, but he will only win California 12% of the time in the simulations.

-PJ