Re: 2012 Presidential Election Part II -- Charlotte, a National Treasure or sede vaca

It can certainly happen, but what's bizarre on the right is they are creating an alternate reality where they assume the best possible circumstances (low turnout, late swing, selective demographic participation) for their desired outcome. The real polling methodologies assume the future will be

Here is the difference. Take a very simple political system in which in every four elections one is a Dem wave, one is a Republican wave and two are in the middle. Real polling will tend to underestimate the wave elections, and so they will tend to be slightly wrong for the two waves and pretty close for the two middling elections. The right's "unskewed" (cough, fair and balanced?) polling will tend to be wildly wrong on the Dem wave, slightly wrong on the two middling, and exactly right on the Republican wave.

In terms of relative error:

Actuals: D + M + M + R = Sum; Expected value of error

Real poll: 2 + 1 + 1 + 2 = 6; 1.50

GOP poll: 5 + 2 + 2 + 0 = 9; 2.25

The expected value of the accuracy of the GOP polls will be terrible, but because once every four cycles they will be dead on correct, they will trumpet themselves as accurate (and probably make accusations of voter fraud the other times -- after all, "we were exactly right last time so how could we be so off this time?!"

Originally posted by

**FlagDUDE08**View Post*pretty much*like the past, particularly the recent past, in those circumstances.Here is the difference. Take a very simple political system in which in every four elections one is a Dem wave, one is a Republican wave and two are in the middle. Real polling will tend to underestimate the wave elections, and so they will tend to be slightly wrong for the two waves and pretty close for the two middling elections. The right's "unskewed" (cough, fair and balanced?) polling will tend to be wildly wrong on the Dem wave, slightly wrong on the two middling, and exactly right on the Republican wave.

In terms of relative error:

Actuals: D + M + M + R = Sum; Expected value of error

Real poll: 2 + 1 + 1 + 2 = 6; 1.50

GOP poll: 5 + 2 + 2 + 0 = 9; 2.25

The expected value of the accuracy of the GOP polls will be terrible, but because once every four cycles they will be dead on correct, they will trumpet themselves as accurate (and probably make accusations of voter fraud the other times -- after all, "we were exactly right last time so how could we be so off this time?!"

## Comment