An interesting new question on exit polls to investigate:

[s-cubed]

From Cannonfire, Joseph Cannon's excellent blog:

A reader just asked me a few simple questions -- damned good questions.
Are there exit poll results for the races for governor, senator, representative, whatever? Did those poll results consistently undervalue the Republican vote, or is that phenomenon restricted to the presidential contest?
Mind you -- if the exits predicted a margin of victory differing from the actuals in races for lesser office, we should not discount the possibility of vote fraud in those instances as well. Just ask Tom Feeney. He knows all about that sort of thing.


www.cannonfire.blogspot.com

[Calvinist Basset]

1. That *IS* a good question!

Has anyone been looking at these figures?

[Darkhawk32]

2. I think there was a thread that had some Senatorial races....

and mentioned that they were dead on right.

[BlueDog2u]

3. Yes, there is a post from a few days ago

which links to "uncorrected" data on three or four local races. One of the states was Colorado; I forget where the others were. I would like to see a larger data set, but based on the data presented from on that site, there was no discrepancy in the local races. In the same states, the President data was off by several points.

[Heewack]

4. I doubt the national media would be interested...

... in those races. And since they paid for the exit polls I doubt they would have footed the bill for the hundreds of races down the ballot.


I think the whole focus on exit polls has pretty much run its course since Mitofsky has said the early numbers released were incomplete.

[clydefrand]

5. I heard our leader? * make a snide comment

on Thursday, I think, about the the accuracy of the exit polls in Ohio. Wished then that I could have gotten to his neck!

[Digit]

6. Anyone have a link?

[politmuse1]

7. Yes, there definitely was a discrepancy

I don't have time to look up articles this morning (remind me tomorrow if others don't find the references), but I read numerous times that the exit polls and the vote matched for lower offices (I believe in Florida in particular, not sure), and were completely off for the presidency.

This election was fixed on a level greater than anyone will ever acknowledge -- as sure as you're reading these lines right now. If we keep focusing on anything less (irregularities here and there, no matter how bad), we'll never prove anything. Our only true proof of the theft is in the exit poll discrepancies ... too many of them to list. That's statistical evidence that doesn't lie. It's science. It can't happen by chance. The polls were right. The election is wrong.

[dargondogon]

8. Exit polls are a joke

I'm persuaded the U.S. exit polls are deliberately unreliable and sloppily prepared and have been for years. I don't think you can prove anything about the presidential election from these polls, and I think that's intentional. I don't think we'll ever learn the truth from Mitofsky because he's committed to defending his reputation with belittlement, condescension and dismissiveness, not to defending the U.S. electoral system.

He's not the sort of person who will admit an error, and considering that the existence of errors is obvious, he won't answer questions about what happened aside from directing questioners to his Web site.

His exit polls have failed in two elections now to allow an orderly projection of the winner, and in the most recent election the exit polls were irrelevant because the networks awaited actual counts in the battleground state and even in crimson red states such as Virginia. We'll never know the meaning of those early numbers that showed Kerry winning. Mitofsky won't explain them.

He insists that the purpose of his polls is not to audit an election. So he can't project a winner, and he can't help audit the vote. His work is pointless, his career wasted, aside from the fact that it put food on his family.

[jkd]

9. Pointless

All this comparison of exit polls to actual vote is meaningless. Either those exit polls or the actual vote was incorrect. The only state with a completed recount confirmed the actual vote.

[jkd]

10. This might help, if you really think these polls mean anything.

Ignatzmouse gave some interesting data relative to your question. It could be found in the archives. The presidential and senate races were the only one's that exceeded the MOE in North Carolina.