Finally picking back up again from my last post on last week's "elections"... (I've been delaying because frankly I get sick to my stomach just trying to start. That's the truth.)
I hope you'll download and read a paper called The Unexplained Exit Poll Discrepancy, released on Thursday by Prof. Steven Freeman of the University of Pennsylvania. It's worth seeing in its entirety. A lot of other folks have been posting about it. Rightly so.
What jumped out at a lot of people on the night of the election was how the "errors" in the exit polls consistently occured in the same direction.
The thing about genuine errors, extremes, and anomalies in results... is that they're random.
The chance that a flipped coin will land "heads" four times in a row is only 1 in 16 -- but you're just as likely to see it land "tails" four times in a row. And if it's an honest coin, flipped fairly, over time, you will. Very basic math will tell you exactly how likely a given outcome is.
But even without the math, we have a sense of this in our daily lives. If you were betting another guy a dollar a flip, and the coin came up tails ten times in a row (about a 1 in 1000 chance) common sense would tell you the coin was weighted.
And if somebody told you it wasn't -- that it was just an error or pure random chance, never mind, keep emptying your wallet -- you'd start to wonder about their motives.
Common sense. Not a conspiracy theory. Just what you're seeing, right in front of you.
Without getting into all the state-by-state details -- I'll let Prof. Freeman tend to the numbers -- what happened last Tuesday, where a wide variety of extremely accurate exit polls suddenly turned out to be at the extremes or even beyond their margin of error, was exceedingly unlikely -- even if the benefits of these errors had been evenly distributed.
But they weren't evenly distributed. They favored Bush. Over and over and over. That's the coin flipping. And flipping. And still coming up heads. Heads in Florida. Heads in Ohio. Heads in a bunch of other swing states (even while the exit polls remained relatively accurate elsewhere). Almost everywhere the election was close, the coin just kept coming up heads.
How bad was it?
According to Dr. Freeman's analysis... 1 in 250,000,000.
One in a quarter of a billion.
| 