CoffeeJanitor
Member
for anyone wondering why the big jumpIn other exciting news...
http://election.princeton.edu
Probability of Clinton win: random drift 92%, Bayesian 95%
He changed the standard deviation for the model from 6% (average from the 50s to now) to 3% (average from 94 till now)
Basically elections are narrower and have tended to move less recently so he switched his model to account for that--prob of thing moving the other way is low