Does anyone know about Warren Mitofsky's official position

on the * shift from the exit polling to the actual vote? This is the man that is used as the expert proving fraudulent elections. Arnebeck promotes him in his lawsuit but doesn't seem to have an affidavit from him about his opinion. Has anyone seen his opinion published anywhere?

TC

"derrr, I guess we just screwed up. My guess is that we were oversampling Kerry voters, who were very excited to be exit polled. Republicans are way busier than democrats anyway...no time to stand around and be exit polled. So I guess we'll get it right next time, or maybe not. We'll see. I'm optimistic that next time the GOP will tell me about their plans ahead of time and I will get the exit polls right. For now, I think we should definately hold off on any kind of investigations. Please just wait til' bush is innaugurated. Pretty Please?"

the actual vote so easily why they were so arrogant or sloppy as to not make the exit polling fit.

TC

Maybe they didn't know how many adjustments they would need before hand.
I still think its weird that they adjusted the exit polls to match the final outcome. Do they normally do that?

if there is any official statement from Warren Mitofsky. Arnebeck sets him up as an expert in the suit but I wonder if he would agree with the suit at all. Setting him up as an expert in the suit without having him on your side would be a fatal mistake. All the other side has to do is produce testimony from "your" expert to explain away your suit.

TC

Maybe he is and maybe we'll see his results is Arnebeck's suit is ever hear.

I can't seem to find it.

He refiled on 12/17; the pdf you link to is the original suit bounced by Moyers on 12/15 or 12/16.

concerning his politics. Has been for decades.

At times he's held off "calling" a state for nearly 24 hours after the polls close. Just because his early stats say one thing doesn't mean he believes them. He can't quantify his sampling error in many instances, and has to make sure that he can factor in appropriately the people that don't respond to the questionnaire, that he has all the age/gender variables set properly, and that he's properly gauged turnout among the various groups. He doesn't have that info until he sees early voting results; in some cases the results from a very small number of precincts can cause him to revise some weightings.

Margins of error are useful only when you know that you have a random sample. Mitofsky tries for one, but knows he usually doesn't get one. He always adjusts for sampling error.

He's also aware that absentee and early voting patterns aren't the same as the patterns on election day, and has been hard put to properly account for them--in a lot of telephone polls over half those sampled refuse to answer, and that's not counting those who don't pick up the phone. Supposedly he has a handle on sampling error from these sources, but "handles" can be slippery.

Mitofsky's seldom miscalled a state. But note "seldom"; it happens. He's also chewed out network people that have called states based on his numbers when he hasn't authorized it, because a lot of fuzzy-thinking people don't make this distinction.

David Moore's "Superpollsters" summarizes a lot of Mitofsky's methodology as of the early 1990s, long before any * weirdness in 2000/2/4.

I think his official position is that he called all the states correctly this time (that may be "within the margin of error"), but I'm not 100% about this. But note that when he calls a state and when the networks call the state based on his data are two entirely different things.

when they were "fixed" the "new" Bush numbers came up? I think we need to know what political party Mitofsky donates to. Just a thought. OR could the KKRove backroom employees have brought the servers down, changed the poll numbers and noone knew? I know NOTHING about computers so if this is not possible...ignore me. :7

I don't believe Mitofsky has responded.

So Mitofsky can't release them without consent of the commissioning entities.

That is how I read his response to Conyers' letter.

He might just be blaming the media, though, a novel concept.

sites in, I think, March.

Search archives for link.

But his afternoon numbers were nowhere near the margin of error as far as I know.

it is that your numbers accurately reflect what you've measured.

But you can't know that what you've measured accurately reflects what there is to be measured.

Until he quantifies his sampling error (non-randomness in sample, disproportionate measuring of one or more groups), the MOE is a statistical nicety.

that the exit poll says 52 for A and 48 for C. If the MOE is 2% then the MOE Range for 52% is
52 + 2 = 54% top of range for A
52 - 2 = 50% bottom of the range for A

48 + 2 = 50% top of range for C
48 - 2 = 46% bottom of range for C

This means that the exit poll really means that:
A's exit poll is 50 - 54
C's exit poll is 46 - 50%

this could be read as a tie in the exit poll and no clear preference.

What is not explained in NH where there was such a large deviation in the poll and after selected hand recounts, it showed clearly that the exit poll was wrong.