« The Freeman Paper | Main | Fraud in Florida? »

November 18, 2004

Exit polls: breaking news

Two late breaking updates that I did not want to get lost in the very long previous post on the Unexplained Poll Discrepancy paper, although both are highly relevant.

The first is news from the USA Today's Mark Memmott that the National Election Pool (NEP) intends to withhold exit poll results until later in the day in future elections.  Money quote:

Sheldon Gawiser, chairman of the polling consortium's steering committee and NBC's director of elections, said Wednesday that in future elections, no data will be sent to the networks and AP until at least 4 p.m. ET. The "first wave" of data that bloggers posted this year, he said, was just too raw to be valuable to "people who don't know what they're dealing with."

Second, and even more relevant to what we've been discussing, blogger Mayflower Hill posts an exclusive interview with Warren Mitofsky.  Using the method of analysis MP anticipated, Mitofsky explains that his data show no evidence of fraud involving electronic voting machines.  Money quote:

One possibility he was able to rule out, though, is touch screen voting machines that don't leave any paper trail being used to defraud the election. To prove this, he broke down precincts based on the type of voting machine that was used and compared the voting returns from those precincts with his own exit polls. None of the precincts with touch screen computers that don't leave paper trails, or any other type of machine for that matter, had vote returns that deviated from his exit poll numbers once the average 1.9% non-response bias was taken into account.

The interview includes much that confirms items in the previous post.  Read it all.  (And if you're here today via Kaus or Today's Paper's, welcome, and please don't miss the previous post!).

Related Entries - Exit Polls

Posted by Mark Blumenthal on November 18, 2004 at 12:11 AM in Exit Polls | Permalink


The idea that touch-screens were used to commit fraud is old, old hat, that everyone accepts as debunked. The discrepancies that have been found were in counties and precincts using optical scan machines to count hand-marked ballots. Why does everyone continue to try to debunk something that no one is pushing?

Posted by: Nick Simmonds | Nov 18, 2004 2:09:28 AM

BREAKING... And speaking of Exit Polls and our friendly neighbourhood Mystery Pollster...

...This, hot of the press of Slate's widely read, "Today's Papers."

Damn you, Shafer! USAT says the consortium running exit polls won't send numbers to subscribers next time until at least 4p.m. EST. The early hours data that leaked this year, said a consortium official, was just too raw to be valuable to "people who don't know what they're dealing with." Mark Blumenthal, better known as Mysterypollster, said the delay will result in "better numbers" that will be "leaked immediately.

Posted by: Rory | Nov 18, 2004 8:28:59 AM

Isn't Mitofsky still depending on a nonresponse bias he can't explain? Could that deviation be because of altered tallies in precincts not polled?

"Black Box Voting began to compare the special printouts given in the FOIA request with the signed polling tapes from election night." . . . "Some polling place tapes didn't match. In fact, in one location, precinct 215, an African-American precinct, the votes were off by hundreds, in favor of George W. Bush and other Republicans."

Posted by: Catastrophile | Nov 18, 2004 1:14:57 PM

What the hell is "non-response bias"? Is that Bush voters not wanting to talk to pollsters? If they didn't want to talk to pollsters then how do the pollsters know they were Bush supporters?

Or are they just guessing and adjusting their figures by 1.9%?

Posted by: aaa | Nov 18, 2004 8:21:58 PM

I'm curious what you and your readers have to say about the announcement today by a UC Berkeley research team lead by sociology professor Michael Hout. Their statistical analysis of touchscreen machines in three heavily Democratic counties in Florida found the likelihood of GW Bush gaining as many votes as he did was less than 1 in 1000. They started out trying to debunk e-voting fraud theories, and now they've uncovered a new one. You'll find the report at http://ucdata.berkeley.edu/ and a good news story on it by Wired News at http://www.wired.com/news/evote/0,2645,65757,00.html?tw=wn_tophead_1


Posted by: dan tynan | Nov 18, 2004 9:57:31 PM

Republican attempts to suppress the Democratic vote are well-documented, and were successful in some places in Ohio. As for fraud, it may or may not have taken place, and on a small or decisive scale - but why the assumption that it's so very unlikely on the face of it? Has human nature suddenly changed? Should we throw out Shakespeare? And why the assumption that it must be a "vast" conspiracy? How many hackers would it take to break into central tabulators?

I ask as a journalist and blogger who observed Bush and his surrogates lying repeatedly during the last month of the campaign about the box he wore on his back into the debates. To my astonishment, the press largely ignored or actively misreported the story, though anyone who cared to look could see a solid rectangular box and cord clearly outlined under the president's jacket. A NASA imaging analyst and spyware experts said it looked like an inductive audio receiver. Anyone who followed the story throughout October witnessed repeated instances of the president and his men lying about the object. The press's suppression of this story, both the fact of the box, and the fact of the lies, was remarkable. Nervous political writers began by smirking that it was a "tinfoil hat" theory, though there was nothing science-fiction-like about the story. When it could no longer be laughed off -- by the press as much as by the president's men -- the media simply lapsed into silence -- leaving readers and viewers misinformed or ignorant. The collective decision of the press seems to have been to leave unwritten this bit of history.

I realize that the question of whether Bush and his surrogates cheated and lied about the device on his back has has nothing to do with whether or not they arranged for cheating in the election. But the eagerness of the press NOT to be appropriately skeptical of the Bush administration's honesty and NOT to report on the plain evidence of their own eyes (so much in contrast with press suspicion of Clinton's every move) -- these make me certain that we can expect the mainstream media to attempt to squelch rather than pursue the far more-difficult -to -report investigations of the vote count's reliability. In short, much of the media seems to think its job is to head off questions and investigations, and to do only a sketchy job of presenting such questions when they can't be entirely ignored. It's astonishing but there it is, and one reason why I'm grateful to mysterypollster and other dedicated truth seekers.

I also want to point out this 2003 story from the UK Guardian, by Andrew Gumbel, about the Georgia elections of 2002 and the very disturbing electronic voting issues it raises.


It begins:

"Something very odd happened in the mid-term elections in Georgia last November. On the eve of the vote, opinion polls showed Roy Barnes, the incumbent Democratic governor, leading by between nine and 11 points. In a somewhat closer, keenly watched Senate race, polls indicated that Max Cleland, the popular Democrat up for re-election, was ahead by two to five points against his Republican challenger, Saxby Chambliss."

Posted by: ibw | Nov 19, 2004 3:22:55 AM

The only thing Mitofsky offers to rebut questions of possible fraud is speculation. Mitofsky refuses to consider possible machine irregularities.

Why is he interested only in the possible error of exit polls?

It's interesting that he was unable to determine why there was a discrepancy in the final data. I'm willing to accept a speculative explanation, such as that Dems talk to exit pollsters while Repubs don't. However, when it comes to making such assertions, I expect there to be some evidence behind them, or else I begin to wonder why this expert is only considering the errors in exit polls.

It simply makes no sense for scientists to wade into the controversy, declare exit polls faulty, simply because the numbers don't add up, without properly considering the possibility of machine error.

And how could Mitofsky be so certain of no machine error? Isn't his credibility shot after the Cal-Berkeley study on voting in Florida's touchscreen counties?

I see a lot of bloggers and reporters doing their best to squelch the fraud theories, but they haven't been squelched yet. Mitofsky checked a few touchscreen precincts to see if there were discrepancies? How is that supposed to debunk the theories? Which precincts did he check?

I think Cal-Berkeley through a huge wrench into everything the Mystery Pollster and Mitofsky have been saying.

Posted by: Upstate NY | Nov 19, 2004 10:51:17 AM

The comments to this entry are closed.