A thought and some questions about the exit polls.

[Goldeneye]

I don't have the screen shots saved anywhere, but these are TIAs calculations based on the screen shots from CNN before the votes started to be mixed in.

National Exit Poll

13,047 respondents,
randomly selected,
1.0% MOE

HORIZONTAL WEIGHTED
PARTY ID
MIX Bush Kerry Nader Bush Kerry Nader
Dem 38% 9% 90% 1% 3.4% 34.2% 0.4%
Rep 35% 92% 7% 0% 32.2% 2.5% 0.0%
Ind 27% 45% 52% 2% 12.2% 14.0% 0.5%
100% 47.77% 50.69% 0.92%

Probability: Poll(47.77%) to vote (50.73%): 0.00000000329447

********* 1 in 303,538,508 ************

Prob = 1 - NORMDIST (.5073,.4777, .01/1.96, TRUE)


These are the final CNN results

party B K N
Democrat (37%) 11% 89% 0%
Republican (37%) 93% 6% 0%
Independent (26%) 48% 49% 1%

which you can see at this link:
http://www.cnn.com/ELECTION/2004/pages/results/states/US/P/00/epolls.0.html#1064

Based on the above data, it seems the appropriate argument for Edison Mitofsky to make is that they oversampled Democrats and undersampled Republicans. Something they should easily have been able to control. The reported exit polls have democrats and republicans voting at the same rate...which is historically very wierd. From what I understand, democrats have voted at a higher rate than republicans for a long time.

It seems funny to me to for the Edison Mitofsky to claim that Kerry voters were more eager to be polled than Bush voters. Kerry lost only 1 point from Dem voters between the pre-adjusted polls and the final polls. He also lost 1 point from pre-adjusted polls to final polls from Republicans. Would Republicans who were voting for Kerry be as eager to talk to exit pollsters as Kerry voting Dems?

The independents lost 3 points between the pre-adjusted and final polls. Would independent voters who decided to vote for Kerry be that much more eager to talk to exit pollers than dem voters who were voting for Kerry? Seems funny to me.


The biggest change in the results is the number of republican and democratic voters. If you were going to adjust the exit poll results to show a Bush win, where would you change numbers? The easiest thing to do would be to change the number of Republican vs. Democratic voters. Obviously the independents would be another target, because no one knew before how they were going to vote (beyond the 3 point margin of error for pre-election polls), and because their votes were set to determine who would win.

Its easy for them to claim that Kerry voters were more eager to talk to exit pollsters than Bush voters, but until back up these claims by providing the raw exit poll data, we should continue to question the final vote results. As other people have pointed out, the Mitofsky report assumes the exit poll results were wrong and the vote count was correct. There is little reason to believe this is true. We have a huge list of voting irregularities:

-electronic voting machines that had Kerry votes being switched to Bush votes
-partisan election officials
-poll tapes in Volusia county that showed Kerry votes being eliminated
-hackable voting machines
-stickers placed over Kerry votes and replaced with a vote for Bush
-stolen election in 2000
-poorly conducted recounts
-videos of republicans saying the election is over because they'll do the counting
-huge swings in exit polls
-stonewalling republican congressional leaders blocking paper trails
-not to mention the most corrupt, power-hungry administration ever, that had the money, connections, and reason to do it.
-............

Is it so hard to believe that the exit polls should be off, with this kind of activity going on?

So, the two things that had to happen in order for Bush to win, (high republican turnout, and a lots of independents voting for Bush) happened. Funny the exit polls didn't show it, but the final results did. So what I'm wondering is have exit polls in the past had as much trouble with independent voters? What did pre-election polls tell us about independent voters? How many newly registered voters were republican vs. democrat? How many total registered voters were democrats vs. republicans? If there were more registered Democrats, it seem logical that a larger percent of the voters were democrats than republicans. This would explain why there were more Kerry voters polled...because there were more Kerry voters.

[papau]

1. true - and obvious - but per our media - not a permitted answer

:-(

[Igel]

2. "Oversampled" is probably not the word you want.

I think M.'s methodology prohibits that oversampling (or very nearly prohibits it). But it can't stop the targets of the sampling method from absconding, either after having said 'no comment' or not. (Select every nth voter; if the voter selected says no, mark him/her down as non-respondent; wait for the next nth voter, do *not* simply take the next one that comes along.)

I think the claim is that 37% of people targeted said "no comment", and 10% just wandered off before they could be asked to participate. In other words, 47% of the people were never asked "bush or Kerry".

I think the way he deals with non-respondents may have made the errors worse, but I can't be sure. And may be wrong.

[NewYorkerfromMass]

3. I have one question about the Kerry voters overrepresenting

per http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=102x1164449

The sample weighting should have accounted for them.
IOW, an overstatement of Kerry support would be offset by throwing out the over share of Democratic respondents.
The only way the Kerry overstatement can be believed is if the margin was entirely in the independents who were "more eager" to state Kerry support.
Is this correct? Has anyone done that analysis?

[Goldeneye]

4. I don't know, the more I think about it the more confused I get.

What you said about the independents sounds about right. Combine that with them saying republicans and dems voted at the same rate, and there you have the differenc.

Another question:

Is this statement from CNN true?

"CNN did not air those inaccurate results or post them on its Web site, and CNN's projections of winners on election night were accurate."

I thought they showed Kerry winning in Ohio...and then the results were adjusted, and Bush was winning...