Stunned by N.H., pollsters regroup to seek answers

Stunned by N.H., pollsters regroup to seek answers
Many theories, no consensus
Email|Print| Text size – + By Brian C. Mooney
Globe Staff / January 10, 2008

It will be cold comfort for the fraternity of pollsters embarrassed by Hillary Clinton's surprise victory over Barack Obama in Tuesday's Democratic primary, but New Hampshire's ornery electorate has been messing with the survey-takers for years.
more stories like this

* What the N.H. primary results taught us
* Late surge of sympathy may have aided Clinton
* Pundits were the big losers in New Hampshire primary
* In unofficial tally, 525,870 voters cast their ballots in N.H. primary
* NH exit polls explain surprises, suggest new strategies
*

On Monday and Tuesday, seven independent polls showed Obama ahead by an average of 8.3 percentage points, according to the RealClearPolitics website. When the votes were counted, Clinton won by 2.6 percent.

What happened? The pollsters have theories, but no one is sure.

"It was a total shock," said Scott W. Rasmussen of the independent Rasmussen Reports, whose final three-day tracking poll - a total of 1,200 likely voters through primary eve - showed Obama ahead by seven points, down from 10 the day before.

"I can't remember a time when the entire polling industry showed a similar result and it was wrong," he said.

Rasmussen said his sample of women voters was smaller, and not quite as enthusiastic about Clinton as the primary day exit polls seemed to indicate. He had other theories: Clinton's better field operation might have been more effective; the huge turnout may have swept up more casual voters who wouldn't have passed through "a pretty tight screen" Rasmussen used to determine likely voters.

"But nothing accounts for a 10-point swing," Rasmussen said.

John Zogby of Zogby International, which recorded the biggest percentage margin before the primary - Obama by 13 points - offered several possible explanations on his website and promised additional study and explanation. Zogby conducted polling for the Reuters news service and the C-SPAN cable network.

In New Hampshire, he theorized online, several factors came into play Tuesday: a large percentage of late-deciding voters; the compressed period of five, rather than the usual eight days, between the Iowa caucuses and New Hampshire primary; and an apparent decision by many voters to not "endorse the Obama train without checking the engine."

His final night of a three-day sample showed Clinton bouncing back at the end, but that was more than offset by Obama's swelling margins in samples the prior two nights, he said.

Zogby's polling showed the volatility of the race: In the five days after Iowa, his New Hampshire tracks flipped from Clinton up by six points to Obama leading by 13 points.

Other pollsters missing the mark with results showing Obama leading Clinton were American Research Group (a nine-point lead), Suffolk University for WHDH-TV (five points), the University of New Hampshire for CNN/WMUR-TV (nine points), Marist College (eight points), and CBS News (seven points).

The previous Suffolk poll had Obama up by one point; the UNH poll had the race tied two days before the vote. UNH also conducts polls for the Globe.

Polling combines art and science, and survey-takers always note that the results are within a margin of error of plus or minus several percentage points.

While the wayward polling this season is dramatic, it happens with regularity in New Hampshire, where voters aren't shy about expressing their contrariness - electing dark horses or upstarts, and smacking pollsters upside the head.

In 1964, Henry Cabot Lodge, ambassador to Vietnam and a write-in candidate, rocked the Republican Party by beating Barry Goldwater and Nelson Rockefeller in the Granite State presidential primary.

In 1968, antiwar candidate Eugene McCarthy made a surprising showing in the Democratic primary. That drew Robert F. Kennedy into the race and led President Lyndon B. Johnson to announce that he wouldn't seek reelection

In 1984, Gary Hart shocked the Democratic Party, beating former vice president Walter Mondale by 10 percentage points in the New Hampshire primary. Three polls, just weeks before the vote, showed him in third place, an average of 27 points behind Mondale.

That prompted the Globe's Martin F. Nolan to muse at the time that New Hampshire voters "are historically fickle, and if Jack the Ripper had been the runner-up in the Iowa caucuses, he might have been destined for the hero's role of upsetting the front-runner.

"Do Granite State voters enjoy fooling pollsters that much?" Nolan asked. "Is massive mendacity part of the local color?"
© Copyright 2008 Globe Newspaper Company.