Polls: Why Rasmussen Should Get A Prize
Simply put, they nailed it, calling 11 out of 12 Senate races...
...and calling 12 out of 13 gubernatorial races:
Honorable Mention: Larry Sabato's Crystal Ball
Seems like a lot of pollsters got the Minnesota governor's race wrong.
Check out the debunking of Rove/FOX's claim that polls (like reality) have a democratic bias of 6 to 8 points.
From Greenwald on the Senate:
MD 5% (against Dem.)
MI 0%
MO 1% (against Dem.)
MT 1% (for Dem.)
NJ 3% (against Dem.)
OH 1% (against Dem.)
PA 5% (against Dem.)
RI 2% (for Dem.)
TN 1% (against Dem.)
VA 0%
WA 7% (against Dem.)
Add a 2 point bias in favor of Lieberman against the real Dem, and that's only 2 races showing a pro-Dem prediction error of a total of 3 points (1% and 2% respectively) all the rest were either skewed (usually slightly) in favor of GOP or it was dead on, for a total GOP bias of 23 points. That's a net 20 point GOP advantage for an average of about 2% GOP bias per race (excluding CT).
But just try telling Brit Hume that. FOX=science is for suckers.
But reality has a well-known liberal bias, so that makes up for it.