Fearing a repeat of 2004 when exit polls appeared to indicate possible fraud in the presidential election, citizen watchdogs will be fanning out across the country for an unprecedented exit poll of their own that organizers say could rival the scope of the “official” exit poll conducted by a consortium of mainstream media outlets.
But disagreements still persist among pollsters and statistical researchers regarding whether exit polls — intended solely for the media to analyze election-night returns — should in fact be used to gauge the integrity of an election.
Steve Freeman, a professor of research methods at the University of Pennsylvania, founded ElectionIntegrity.org and wrote a book, Was the 2004 Presidential Election Stolen?, in part about the 2004 exit poll discrepancy.
Together with the Election Defense Alliance, Freeman and others are organizing this year’s “citizen exit poll.” He wouldn’t reveal any of the details of where they plan to survey voters lest anyone try to game their findings. They will, however, reveal any and all details following Election Day, something Edison Media Research and Mitofsky International, the firms hired by the National Election Pool since 2004, have never done.
“We’re going to try and determine in select precincts around the country whether or not votes are in fact being counted as cast, how survey results match up with official election results,” Freeman said.
Why would they do this?
As if there wasn’t enough to worry about this election — what with voter purges, barriers to the ballot box, court battles over voting rights and faulty voting machines — there’s also widespread skepticism that the exit polls conducted by the NEP will report their actual results.
The National Election Pool comprises ABC News, The Associated Press, CBS News, CNN, Fox News and NBC News. Calls to Edison-Mitofsky by Miller-McCune.com to inquire about the exit poll planned for this year were forwarded to a Fox News communications person, who did not return the call.
In 2004, early exit poll data leaked on the Internet showed a substantial “red shift” in 42 states of roughly seven percentage points — 10 million votes — from Sen. John Kerry to President George W. Bush. But the exit polls by the NEP were corrected after the official returns came in, which fueled conspiracy theories.
“If in fact people cast their ballots how they said they did in confidential questionnaires upon leaving the polling place, George Bush would not have won by 3 million votes but rather would have lost by 6 million votes,” Freeman said. “Rather than winning Ohio by 120,000 votes, survey results indicate he lost by half a million votes.”
Edison-Mitofsky in 2004 said the exit polls were just plain wrong. They oversampled women and undersampled rural areas, and Bush voters weren’t as enthusiastic about speaking to pollsters on their way out of the voting booth, they said.
Some election integrity watchdogs, on the other hand, said the exit polls were likely correct and the election was wrong. They point to a history of exacting exit polls that seem to divert from the actual election results only beginning in the mid-’90s, about the same time the country began widespread adoption of electronic voting equipment. The recently released film, Stealing America: Vote by Vote, documents this story.
Four years later, the debate over whether exit polls are a justifiable way to gauge the accuracy of an election is far from settled.
Even in this year’s primary, the exit polls reported a discrepancy compared with the actual results. Mark Blumenthal, a respected polling analyst who blogs at Pollster.com, reported that exit polls overestimated Obama’s lead by an average of 7 percentage points in 18 or 20 states, based on leaked early results.
Pollster John Zogby told Miller-McCune.com that he does not believe exit polls should be used to basically audit an election.
Why is that? Exit polls by their very nature do not draw from a random sample of the voting population. They rely rather on a random sample among a select number of precincts.
In addition, the sample size is often too small to draw broad conclusions, Zogby said. If the exit polls were off by something like 10 percentage points or more, as they were in Ukraine and Georgia in recent years, that would be greater proof of election malfeasance, Zogby said. In Ohio, for instance, exit polls were off by roughly 2 percentage points.
Zogby said that post-election polling by telephone and Internet that Zogby International and others will conduct in the day or two following the election are far better measuring sticks of whether votes were counted accurately.
“I don’t think exit polls should be used to predict winners,” Zogby said. “The real value is they give us the data we need on who voted and why they voted the way they did. Exit polls can’t be used as a reliable indicator on a close election.”
Jonathan Katz at the Caltech/MIT Voting Technology Project sides with Zogby. The group put out a report shortly after the 2004 election explaining why exit polls should not be used to audit an election. (Freeman and others called their analysis a “travesty.”)
“It (using exit polls to audit an election) is odd for a whole bunch of reasons,” Katz said. “About a third of people will vote absentee, and exit polls will only be representative of people who vote at the poll on Election Day.” But that’s just part of a broader issue, Katz said.
“Exit polls are not like the polls repeated daily in the press,” Katz said. “Those are representative of voting households. In order for a poll to be valid, we need to make an inference about a population. That’s not what goes on in an exit poll. When you conduct an exit poll, an analyst decides on some selection of precincts or polling places to poll. They may or may not be a random sample of a state. But even if they were, while they may be representative of a precinct, they would not be representative of voters. Why is that? Polling places are small geographic areas, and we know that people in small areas are more similar than not.”
Katz said post-election polls together with recounting actual paper ballots and double-checking electronic voting equipment is a better way to audit an election.
But with the abundance of reports of voting machine malfunctions and the refusal of such states as Pennsylvania to make paper ballots readily available, even standard audit methods might not catch the mistakes. Until our election system is reformed, citizens who care deeply about democracy will be standing outside polling places across the country to try and determine if votes are being counted as people say they are cast.
Are you on Facebook? Click here to become our fan.