Early figures from the exit poll operation on election night showed several states with vote totals highly favorable to Sen. John Kerry, leading many to expect a Kerry victory. Indeed, the estimated popular vote was 51% to 48% in Kerry's favor, though the final results turned out to be a mirror of these figures -- President George W. Bush by 51% to 48%.
The resulting brouhaha over the erroneous exit poll results has produced numerous wild interpretations of what really happened. Two of the most interesting are discussed below.
Conspiracy #1: Exit Poll Errors Were a Ploy to Depress Republican Turnout
This is the view of Dick Morris, who writes "The Political Life" for The Hill, a newspaper "for and about the U.S. Congress." He was a political consultant for President Bill Clinton during Clinton's re-election effort in 1996, though in general Morris worked more for Republicans than Democrats.
His argument is that "exit polls are almost never wrong." The fact that the 2004 exit polls were wrong in so many states leads Morris to speculate "more than honest error was at play here." As he writes, "Dark minds will suspect that these polls were deliberately manipulated to dampen Bush turnout in the Central, Mountain, and Pacific Time zones by conveying the impression that the president's candidacy was a lost cause." After repeating his thesis a couple of times, he finally admits he has such a dark mind, and concludes, "This was no mere mistake. Exit polls cannot be as wrong across the board as they were on election night. I suspect foul play."
For this charge to be true, the organization in charge of the exit polls, Edison/Mitofsky, would have faced monumental problems of coordination to pull off the ruse. On election night, and on every primary/caucus night during the early part of the year, I was at 34 West Main Street in Somerville, N.J., where Edison/Mitofsky make their exit poll calculations and decisions. I was CNN's contact point at the organization, and I can vouch personally that so many people are involved at all levels of data collection and analysis, it would have been impossible for any deliberate bias to be incorporated into the system without it being widely known.
Besides, what would have been the motivation? Early on in the evening, Lenski and Mitofsky warned all of the networks that their samples were biased and that early returns could well be overstating the Kerry vote. Why would they have devised an intentionally biased system and then notified all the networks early on that the bias was there? That doesn't make sense to me. In fact, Edison/Mitofsky were so concerned about the bias, they waited until they got actual vote returns in their sample precincts before calling the winner in any of the questionable states. And not one state was called incorrectly.
So much for this conspiracy.
Conspiracy # 2: Kerry Won. The Exit Polls Were Correct; the Vote Was Wrong.
This conspiracy apparently has not yet gone away. On AAPORNET, the listserv for members of the American Association for Public Opinion Research, there are still questions being raised about the validity of the vote count. On Thursday after the election, Greg Palast of TomPaine.com wrote that "Kerry Won." His argument was that many votes in Ohio were simply discarded, most of which he hypothesizes came from areas where Kerry was doing well. Combined with the outstanding provisional votes, the discarded votes, he argues, would have been enough to overcome Kerry's deficit -- which is somewhat more than 100,000 votes. Had Ohio gone for Kerry, Bush would have lost his re-election bid.
A more general criticism is that in states where there was only electronic voting and no ability to have a recount, the exit polls were much more favorable to Kerry than were the reported vote counts. In states where there was a "paper trail" of some sort -- either paper ballots, or perhaps electronic voting that produced a paper copy -- the exit polls and the actual vote count were quite similar. The implied conclusion: Kerry votes were stolen in states where there is no way to check the results, but not in the states where there was a paper trail.
It's still too early to verify or dismiss such charges, and I would not want to prejudge any legitimate challenge to vote fraud by either political party. But I can say from my own personal experience living in New Hampshire and studying its politics for years, that the early exit poll -- not the vote count -- had to have been wrong. In my voting location in Durham, and I believe in all or virtually all of the other precincts in the state, there is a paper trail of the vote.
The early exit poll in New Hampshire showed Kerry up by as much as 16 percentage points, though he won by just 50.2% to 48.8% -- a difference of just 1.4 percentage points. Had Kerry actually won by double digits, it would have sent shock waves all over the country. All of the pre-election polls showed a close race, with Kerry up by no more than a couple of points. Bush won the state in 2000 by one point, and there was nothing that happened in the state that would have suggested a major turnaround in the electorate, producing a landslide for Kerry. Moreover, the state is mostly Republican: It has about 40,000 more registered Republicans than Democrats (a difference of about 4.3 percentage points ), and all the federal offices in the state are held by Republicans -- the two House members and the two U.S. senators.
By the way, the exit poll also showed a double-digit lead for the Democratic gubernatorial candidate, John Lynch, who was challenging the incumbent Republican Gov. Craig Benson. Lynch eventually won with a two-point margin, close to what the pre-election polls had shown, but much smaller than what the initial exit poll showed. Again, it was clearly the exit poll that was wrong.
So, what really happened to the exit polls?
The Edison/Mitofsky Explanation
Warren Mitofsky and Joe Lenski, co-directors of Edison/Mitofsky, say that the Kerry voters apparently were much more willing to participate in the exit poll than were Bush voters. The interviewers at each of the sample voting locations are instructed to choose voters as they exit from the voting booth -- every third, or fifth, or eighth voter -- some sequence of this sort that is determined ahead of time. Many voters simply refuse to participate in the poll. If the refusers are disproportionately for one candidate or another, then the poll will be biased.
Later, any such biases will be adjusted once the poll results are compared with the actual vote in those same precincts. Because this correction includes the reported vote, the "adjusted" exit poll results cannot be used to check whether the actual vote count is accurate. The exit poll models assume an accurate vote count.
Still, there are checks. The models compare results with previous elections, with the pre-election polls, and with party affiliation. And based on those checks, Mitofsky and Lenski dismiss the idea that in fact Kerry won -- that the exit polls were right and the vote count wrong.
Still, some people fall in love with their own conspiracy theories and no amount of evidence will undermine that sentiment.