Reply #122: thanks, that somewhat clears the air

[OnTheOtherHand]

I hope you can understand my frustration at being bludgeoned with facts that I don't believe to be factual. I agree that that exit poll was especially incredible, which is why I was rubbing my eyes at your reference to Venezuela. Venezeula actually had two vote count audits that year, which help to support the substantial accuracy of the official returns. (Not long ago I read a statistical paper that tried to argue for extensive fraud; it was quite poor.)

I do think that exit polls are less accurate than you seem to think -- I won't generalize about "people." Exit polls do, as Freeman says, have the advantage that the respondents know if they voted (or tried to vote) and presumably know who they voted (or tried to vote) for. On the other hand^TM, considering that voters have substantial discretion about whether to approach (or to be approached by) the interviewers, and that voters can see the interviewers, one would expect exit polls to be subject to different social effects than telephone polls. (Presumably 'secret ballot' exit polls should be less subject to those effects than actual in-person interviews, and it seems that they are.)

Election Defense Alliance has presented really strong evidence about exit poll bias, which surely is not something the organization set out to do. Check out pages 38-40 or so (in the Scribd pagination) of their report here. Here is a money quotation from page 39:

By comparing the tables above, it is seen that Republican voters were undersampled in all 10 polling places, by ratios ranging from 1.16 to 1 at Taft High School to 2.5 to 1 at Topanga.

If I told you that Republican voters were undersampled by ratios like that in the 2004 national exit poll, you'd be entitled to incredulity. Yet this is EDA describing their own exit poll, in which they did a lot of work to get the most reliable results possible. By their own account, they were way off. Now, they think that they found a way to compensate for that bias, and I think they are wrong, for reasons I'm glad to elaborate upon. But if people aren't willing to subject exit poll results to rigorous scrutiny in the wake of this report, I think they are whistling past the graveyard. And I do mean scrutiny -- not that exit poll results should be assumed to be wrong, any more than they should be assumed to be right.

Sometimes you could stand to focus less on trying to "ascertain (my) motive" and more on the meaning of the words! Just saying. Yes, I think what you will find is that the U.S. government and certain NGOs have supported exit polls as an election transparency measure; the Carter Center has opposed the use of exit polls as a transparency measure; international organizations tend to downplay or disregard their results. (Check me on this, but my recollection is that the OSCE observer report on Ukraine didn't mention the exit polls at all. Even Lugar's official statement didn't mention them, although he alluded to them in press conference Q&A.) The bag is much more mixed than you apparently have been led to believe.

Since you seem obsessed with the idea that he did win it...

Frankly, I think it would be fairer to say that you seem obsessed with finding some reason for evading a factual discussion. But maybe we could set aside the imputations. I'm game if you are.

I don't think the Conyers report proves that Kerry did or would or should have won; the report itself doesn't claim so; you will find that the dominant opinion of political scientists, regardless of their political leanings, is that it proved no such thing. We could be wrong, but it isn't that none of us read Conyers or RFK Jr. Moreover -- and it really bothers me that so few people understand this point -- a lot of what the Conyers report describes would not have affected the exit poll results. Kerry probably lost at least 10,000 net votes in Franklin County due to lousy machine allocation, but the people who were turned away are not likely to have filled out exit poll questionnaires. Using the Ohio exit poll results to buttress the Conyers report is simply backwards.