Pollsters look back at where they goofed

The Baltimore Sun

The pollsters have some explaining to do.

Widespread forecasts that Barack Obama would defeat Hillary Clinton in New Hampshire's first-in-the-nation Democratic presidential primary proved faulty Tuesday night, with Clinton, a New York senator, winning by 3 percentage points.

Polls released in the days before the vote showed the Illinois senator with leads ranging from 5 percentage points in a survey by Boston's Suffolk University to 13 points found by USA Today/Gallup.

It was a "fiasco," in the words of Peyton M. Craighill, ABC News' deputy polling director.

"The problem with the polling performance in New Hampshire is exactly the same as it was some 60 years ago with the 'shocking win' election of Harry S Truman," said pollster Peter Hart. "The pollsters concluded their polling before the voters made up their mind for the final time."

Democrat Truman defeated his Republican presidential challenger, Thomas Dewey, in 1948 and confounded the pollsters who had Dewey ahead before the election. A victorious Truman famously waved a copy of the Chicago Tribune with the mistaken headline: "Dewey Defeats Truman."

With a compressed five-day campaign between the Iowa caucuses and New Hampshire, polls that ended a day or two before the primary missed the last-minute rush to Clinton among women. That shift may have been sparked by coverage of the former first lady choking back tears Jan. 7 in Portsmouth as she described how she weathered the rigors of campaigning, pollsters say.

"Everybody saw it," said New Hampshire pollster Dick Bennett, whose American Research Group showed Obama leading Clinton by 10 percentage points. For the first time in the campaign, women could understand Clinton's motivations for seeking the presidency, Bennett said. "It drove women right back to her."

Clinton, her voice cracking with emotion after one question, answered that politics for her was very "personal" and not a "game." Clinton said she persevered because running "was the right thing to do for our kids' futures."

Bennett said that his last poll, which concluded about 7 p.m. the day before the primary - hours after the Portsmouth episode appeared on television - showed a movement of women back to Clinton.

"At the time, I didn't think it would be enough to do it," he said. "Maybe the polls were correct up to the point when the interviewing stopped."

ABC's Craighill, in an e-mail posted on the network's Web site, said "it is simply unprecedented for so many polls to have been wrong."

Pollster.com, a Web site that compiles U.S. poll results, said Obama's lead over Clinton averaged 6.3 percentage points.

The confounding results probably weren't the result of flaws in poll sampling methods because "a large number of independent polls conducted totally separately from one another all showed similar results, that Obama was ahead," said Frank Newport, editor in chief of the Gallup Poll.

More likely "there was an actual shift in voter intentions between the times we interview them and when they went into the booth," Newport said.

The short time between the Iowa caucuses and the vote in New Hampshire was "especially a problem" for pollsters, said Peter Brown, assistant director of Quinnipiac University's polling institute in Hamden, Conn. "Either some of the polls were wrong or they were accurate on Saturday and Sunday when they were taken, but not on Tuesday when people voted," Brown said in a telephone interview.

"And that's the point: Polls are a snapshot in time."

Copyright © 2019, The Baltimore Sun, a Baltimore Sun Media Group publication | Place an Ad