I am not "predicting" anything. That is, I am not making guesses about turnout, or historical trends from past elections, or crossover impacts from one race to another.
Nor am I doing any statistical analysis. I am letting the pollsters and pundits do the statistical analysis. I am taking the RESULTS of their analysis and converting it to a probabilitistic analysis of the leader actually winning, using their outputs as my inputs.
I am leaving the "special sauce" to others and simply reporting the impacts of the numbers that THEY produce. This is about interpreting what the pundits are saying.
-PJ
You are making an implicit assumption that the central limit theorem applies and that the covariance of the distributions are zero, whether you realize it or not.
My prediction - the Repubs will pick up a large chunk of seats and will soundly have control of the House after the election - BUT...
I also predict there will be some fairly close races in traditionally dRAT areas that will see incredible election fraud in a last-ditch effort to hold on. We have already seen blatant efforts in some races (Reid out in NV for example).
I believe some areas will have some of the worst voter fraud/ballot fraud ever seen.
And the Obama justice dept. will just ignore it (unless it is a Republican candidate being accused).