Political snapshots proved sharp
— -- The pollsters got it right, after all.
For weeks, analysts such as Steven Schier, a political scientist at Carleton College in Minnesota, had questioned whether Election Day results would expose deep flaws in the daily drumbeat of surveys showing Democrat Barack Obama consistently ahead of Republican John McCain. Would the public opinion researchers stumble, pundits such as CNN's David Gergen wondered, over the challenge of measuring support for an African-American presidential nominee?
Would there be a "Bradley effect," in which people lie to pollsters about their willingness to support a black candidate? Would there be a repeat of New Hampshire's primary, where polls suggested Obama would win but voters delivered the state to Hillary Rodham Clinton?
In a word, no.
"I think the polls did a very good job this year," said Scott Keeter, director of survey research for the Pew Research Center in Washington, whose final estimate of 52% for Obama and 46% for McCain proved accurate.
"Both at the state level and the national level, I think it was a remarkably positive picture," said Frank Newport, editor in chief of The Gallup Poll.
Most major polls came close to the outcome. Rasmussen Reports joined Pew in calling it on the money, while the Battleground Poll, Zogby International and Gallup, which surveys with USA TODAY, came within a few percentage points. Gallup's final estimate of likely voters had Obama leading, 55%-44%.
The much-debated Bradley effect is named for Tom Bradley, an African American who ran for governor of California in 1982 after serving as Los Angeles mayor. Pre-election polls and surveys of voters leaving polling places suggested he would win, but when the votes were counted, he narrowly lost. Social scientists have debated ever since whether the Bradley effect was real or the product of flawed polling.
"There were no pollsters who were worried about the Bradley effect," pollster John Zogby said. Scott Rasmussen of Rasmussen Reports said, "If people aren't going to vote for an African American, they'll tell us."
In August, Harvard researcher Daniel Hopkins published a study of 133 elections from 1989 to 2006 that featured candidates of different races. He concluded that black candidates once underperformed their poll numbers, but that stopped after 1996.
"I hope we can lay that one to bed," said Gary Langer, director of polling for ABC News.
Another potential polling pitfall that didn't pan out: Whether results can be skewed by the failure to survey younger voters who have abandoned regular telephone lines and use only cellphones. "We did two studies of people who use cellphones only vs. those who use land lines, and we found that while there were some slight differences in demographics, there weren't differences in political attitude in behavior that could not be handled by weighting," Zogby said.
Among the hundreds of major polls published during the campaign, some look out of step in retrospect, including an AP poll published Oct. 22 that said Obama led McCain by 1 point, 44% to 43%. That can probably be chalked up to "random variation — the old bell curve," Rasmussen said. "There will always be a couple that are extreme in one direction or the other."
Rasmussen said his company's polls were consistent in the weeks leading up to the election. Was he anxious on Election Day? "We felt very good about it, but you're never sure until the real numbers come in," he said.