Since once again polling data has become the center piece of a thread I thought I would again show evidence that the polls tend to skew towards Democrats.
From
Pollingreport.com
2004 Presidential election results
Bush 51% Kerry 48%
Here are the "Pollster Vote Projections" i.e. their best guess as to who would end up with what.
Bush first Kerry second
Zogby 49.4 - 49.1
Battle ground: Tarrance(R) 51.2 - 47.8
Battle ground: Lake(D) 48.6 - 50.7
TIPP 50.1 - 48
Harris 49 - 48
Democracy Corps(D) 48.7 - 49.5
Gallup 49 - 49
Pew 51 - 48
So out of 8 polls: 3 were close to the real outcome. 2 picked the right winner, but had the numbers wrong. And 3 either had a tie of went for Kerry.
Notice that the two polls with the D next to them were run by Democrats and both predicted that Kerry would win... interesting huh?
Even more damning, of the 8 polls only 2 put Bush's total higher than the 50.75% that he actually got, while the other 6 put Bush's number lower than actual result. While on the other side 4 of them over stated Kerry's total. Only 1 under stated Kerry's actual vote.
So we have 2 polls that leaned towards Bush (Tarrance, Pew) 1 poll that was almost dead on (TIPP) and the other 5 leaned towards Kerry by 2% or more, one was 3% wrong and one was 4.5% wrong (which happens to be out of that polls MoE. That one was also run by a Democrat hmmmm)
You can talk about margin of error all you want, but when over half the polls create an error in one direction and only 2 go the other way there are definite problems with how they determine outcome and who really is a 'likely' voter.
The numbers are not as easy to run on Congressional preference because they don't have the nice "Final Tally" section of that.
Congress numbers
Actual results
Republicans 55 million Dems 52 million or
R 49% D 46.5% = R+2.5%
All of these are from late October of that year.
Once again Republican first Democrat second
George Washington U. 47 ? 44 = R+3
CNN/USA Today/ Gallup 47- 48 = D+1
NBC/Wall Street Journal 43 ? 44 = D+1
Newsweek 49 ? 44 =R+5
Democracy Corps (D) 45 ? 48 =D+3
AP 46- 47 = D+1
CBS/NY Times 39 ? 45 =D+6
7 polls. One of them (Newsweek) was semi close, had the R number right and the D number low. One of them (GW) had the winner right, with lower numbers for both sides. FIVE of them predicted that the Democrats would take a majority of votes for congress. 5 out of 7 were WRONG by more than 3% (MoE in all of there is around 3%)
Notice that NONE of these polls over stated the Republican number, but 3 of them over stated the Democrat number, which is HUGE because all of them have over 3% undecided or not sure. CNN and Democracy Corps had 5% not sure and still over stated the Dem vote by 2%. That means that every undecided and 2% of the people who said they would vote Democrat actually voted Republican.
CBS-NY Times gave Democrats a 6% point edge, a 9% point error from the actual vote. (and you wonder why we bitch about them the most?)
Of course no one would ever accuse them of being biased right?
Now please throw out your ?vote stealing? argument. Since 1% of the vote is nearly a million votes I would love to hear how Republicans stole over 2 millions votes without anyone getting caught.
What does this mean for this year? Well since most tracking polls show about a 10 point edge for the Democrats when we subtract the 4 point average error from 2004 we end up with a 6 point Democrat edge, which is going to hurt like hell on election day.
Ps. what a long ass post