WHAT NUMBER DID YOU HAVE IN MIND?
Over on Democratic Underground, the morelocks are getting all excited about the "odds" of the exit polls being wrong. Here's part of what DU member "TruthIsAll" says.

Those are the odds that Kerry's EXIT poll percentage would EXCEED his ACTUAL reported vote percentage by MORE THAN THE
MARGIN OF ERROR in 16 out of 51 States by chance alone. That is exactly what occurred on Nov. 2.

I know that is hard to fathom. But here is the data. And here is the calculation, based on the number of individuals polled in each state and the corresponding Margin of Error (MOE).

The chances of a given state falling outside the MOE = 1/20 =.05. The calculation for the probability that 16 out of 51 states would fall outside the MOE is a simple one which uses the BINOMIAL DISTRIBUTION FUNCTION:

The Probability (P) that at least 16 out of 51 states would deviate beyond the exit poll MOE is:

P = 1-BINOMDIST(16,51,0.05,TRUE)

This returns P= 0.0000000218559% or 1 out of 4,575,415,347!

He then lists the state by state data that he uses.

Sounds pretty damn impressive doesn't it? So does Geordi La Forge but his scientific techno-babble is mostly fiction too.

And who is "TruthIsAll"? Well, you may remember him from the pre-election crowd trying to predict the election. He had a problem though. His inputs and assumptions were bad. He assumed Kerry would get a huge percentage of the undecideds and he ignored pollsters he didn't like (complaining about "corporate" polls). His results reflected his biases.

TruthIsAll's complicated and thorough election simulation predicted a Kerry win, 51.8% to 48.2%, with an electoral victory of 337-201. He pegged the probability of Kerry winning at 99.8%, which is a virtual certainty. And his statistics work with the numbers was probably right on. The problem is the garbage data he used.

And he's right about the odds of polling being outside the margin of error - kinda. If the exit polls were correctly specified, the odds are damn low that 16 states would be screwed up outside the margin of error. But the margin of error concept assumes a truly random sample from your universe. The only way to be sure the exit polls were true random samples, is to sample EVERY precinct and blindly select the voters to be sampled - and every voter so selected would have to both consent to the survey and accurately answer the questions. TruthIsAll assumes that the exit polling was a true, respresentative sample of the national vote, but that the national vote itself was not.

For example, take Florida. TruthIsAll believes the Exit Polling that said Kerry would win with 50.51% of the vote (You have to love that .01% he got in there to show his accuracy). Thus, he has to discount the poll taken in Florida on November 2 (we call it the "election") as well as the majority of polls taken in Florida throughout October, which showed Bush leading by as many as 8%. And despite those polls (positives a Bush lead, negatives are Kerry, here they are 9,5,4,-2,8,3,4,3,-1,3,-1,1,3,0,-3,2,8,3,2,0,-1.6,4,5,4,-3,8,4,-5,1,0 and Kerry did not poll at higher than 50% ever before the election, and at 50% only a handful of times), TruthIsAll calculated - using "statistics" a probability that Kerry would win Florida at 86.49%. No one watching the polls throughout October could have possibly thought Kerry had a greater than 50% chance at Florida.

I calculated a probability of a Bush win in Florida at 63.9%. In Ohio at 53.6%. TruthIsAll pegged Ohio at 86.49% for Kerry. His calculations weren't bad, but his assumptions were. While TruthIsAll predicted Kerry would win with 51.8% of the vote. I predicted Bush with 50.5% (he got 50.9%). TruthIsAll said the electoral college would go 337-201 for Kerry. I said 296-242 for Bush in my composite (I missed Wisconsin where Kerry won by less than 1/2 of a percent), and my election simulation said 285-253 (1 point off!).

Check out Sam Wang and Stuart Theil. Sam Wang's a professor up at Princeton. He didn't ignore data he didn't like and his simulation nailed it. Bush 286, Kerry 252. He also did some assuming (like TruthIsAll) about how "undecideds" would break and got a result similar to TruthIsAll. Kerry 311-227. Stuart Thiel predicted a Bush win of 278-260 (though I don't remember him having these numbers on Nov. 2 - but I could be wrong).

So, TruthIsAll opened up his spreadsheet once again, input bad data and came up with conclusions to establish something for which there is no evidence. That the exit polls were correct and there was wholesale, widespread, historic vote fraud. And the leftfringe is eating it up.

Eventually, I plan to more fully document the performance of the FederalReview Composite Poll and other prognosticators on the web. I think I did pretty well, as did many others.

And thanks to Stinging-Nettle for spotting and linking the DU thread.

MEANINGLESS "NEWS"
When you don't have any idea what to report, commission a poll and report that as "news". So, we find out that "six in 10 Americans say there should be a mandatory retirement age for Supreme Court justices."

In the same poll, six in 10 Americans say they don't know what job William Rehnquist holds.

So, the news here is that a bunch of people unacquainted with any clue except the board game, have an opinion about something they know nothing about. Good thing there wasn't an essay question on the separation of powers or checks and balances, or then we'd really know why we live in a representative democracy.