Experts Explain Problem With Exit Polls

For a brief moment in time last Tuesday, the numbers were in favor of Democratic presidential candidate John F. Kerry.

That’s because exit polls—data taken from interviews with voters as they left the voting booths—erroneously calculated that Kerry would sweep up enough electoral votes to beat President Bush and win the presidency.

But when that prediction proved wrong, many were left wondering why the polls were off beam.

“It was a phenomenon of overreaction,” said William Schneider, a CNN senior political analyst who spent election night interpreting exit poll numbers. “Yes, there were some problems, but nothing particularly unusual.”

Schneider, who was an associate professor of government at Harvard from 1972 to 1979, said that the significance of exit polls was overplayed.

He said that exit polls, as a predictor, represent only a portion of the electorate and their results do not necessarily reflect the preferences of an entire state.

In a report obtained by the New York Times on Friday, officials with the National Election Pool (NEP)—the group in charge of administering the exit polls—attributed the inaccurate predictions to factors outside of their control. The NEP officials said that in some cases, pollsters were unable to get close enough to talk to voters, and that in several precincts, Democrats were more willing to talk to surveyors than Republicans, resulting in numbers that favored a Kerry win.

The NEP report stressed that in the end, exit polls achieved what they were designed to do: assist news networks in deciding which states to color red and blue in the race for electoral votes.

“If you look at broadcasts from election night nobody made the wrong call,” said Bradlee Professor of Government and the Press Thomas E. Patterson, who teaches at the Kennedy School of Government.

Patterson pointed to the internet as a major factor in causing the numbers confusion, specifically weblogs, or “blogs,” that fed early exit poll data to millions of people without warning viewers of the polls’ limitations. Beginning at noon on Election Day, numbers from one-third of the precincts were leaked to blogs.

“I don’t think exit polls failed,” Patterson said. “I think the bloggers misused them.”

“Partial exit poll results are always unreliable,” said Schneider.

Since 1980, exit polls have been the primary means of predicting election winners. But after the 2000 election, when exit polls caused the media to vacillate between calling Florida for President Bush, or for former Vice President Al Gore ’69, networks have been more cautious about declaring the winner based on only exit poll results.

CNN, according to Schneider, relied on four sources to predict results this year: exit polls, pre-election public opinion polls, polls in specific precincts and finally, the vote count itself.

“None of those is always reliable,” he said. “Even actual vote counting has errors, as we saw in Florida in 2000.”

He said that aside from being a predictor, exit polls are equally valuable in helping analysts figure out why voters vote the way they do.

“We can look at a number of voting patterns and analyze, for example, the religion factor, the culture factor and moral values of voters to give us a clear picture of where they are coming from,” Schneider said.

The exit polls also had a palpable effect on the campaigns, something that was a surprise to many analysts.

“I’ve never seen it happen where it affected campaigns,” Schneider said. “Both Republicans and Democrats tried to use them to make last-minute campaign adjustments.”

In the future, said Schneider, the media will be even more wary of using exit polls.

“Caution is the bottom line,” he said.