No single reason for faulty New Hampshire pre-vote polls

NEW YORK (AFP) — Experts on Wednesday said there was no single reason why opinion polls predicting a clear victory for Democrat Barack Obama on the eve of the New Hampshire presidential primaries were so wrong.

Polls predicted an Obama victory over his fellow senator Hillary Clinton by a margin of five to 13 percentage points -- but when voters cast ballots on Tuesday, Clinton won 39 percent against 36 for Obama.

"It is simply unprecedented for so many polls to have been so wrong," said Gary Langer, director of polling at ABC News, who has been conducting political polls for more than 15 years. "We need to know why."

Pollsters accurately predicted Senator John McCain's victory among the Republican presidential hopefuls. But they are scratching their heads over how Obama's supposed clear advantage evaporated in just 24 hours.

Was it Clinton's tear-choking incident on the eve of the primaries?

Were voters wary about choosing an African-American candidate like Obama?

Did New Hampshire independent voters, who can vote for either Democrats or Republicans, throw the poll figures?

Or was the wave of enthusiasm for Obama artificially amplified by the news media after his victory the week before in Iowa?

Each factor may have cost Obama a point or two, and each expert has an explanation.

"New Hampshire is a murderous place to poll," said Maurice Carroll, director of the Quinnipiac University Polling Institute.

"The people who did this are good pollsters, they worked hard and they came up with what they thought were reliable numbers," Carroll said.

"But two things: you don't know who you're polling because you walk into a polling place in New Hampshire and say 'I'll think I'll be a Democrat. No, you know, I think I'll be a Republican.' Well how do you poll something like that?"

Some believe that Obama suffered the "Bradley effect" at the hands of New Hampshire's largely white electorate, where African-American candidates have strong support in the opinion surveys ahead of the vote, but lose to white candidates at the ballot box.

The reference is to Tom Bradley, a black popular mayor of Los Angeles who ran for governor of California in 1982 but lost despite opinion polls giving him a wide margin of victory ahead of the vote.

But David Bositis, a polling expert with the Joint Center for Political and Economic Studies, said the New Hampshire election results suggest a discrepancy not with Obama's poll numbers, but Clinton's.

"If I were grading the polling, the grade for Obama would be A-plus," said Bositis, whose Washington, DC-based think tank deals with issues affecting the African-American community.

"The errors entirely came from the estimates for Hillary Clinton, not Obama," he said.

"The pre-election estimates for her were 30 percent, and she ended up with 39 percent. What needs to be explained is, how did Hillary improve her numbers?"

Some analysts said Clinton's surge could have been motivated by last-minute sympathy among women following an incident in a New Hampshire coffee shop on the eve of the primary, when Clinton grew misty-eyed and soft-voiced when asked how she managed to keep going every day.

Women turned out to vote in greater numbers than men in New Hampshire, according to an MSNBC Democratic exit poll. Fox News found that women voting Democratic preferred Clinton to Obama 47 to 34 percent.

The coffee shop incident "certainly humanized Hillary Clinton, which she desperately needed in some regards," said Costas Panagopoulos, director of the Center for Electoral Politics and Democracy in New York's Fordham University.

"But I don't think it was purely that incident that made this happen for her," he quickly added.

One possibility, Panagopoulos told AFP, is the loss of the Iowa caucuses may have helped "in reminding her supporters and voters who were on the fence that she could lose and encouraging them to show up."

In the end "there may be no smoking gun," said ABC's Langer.

"Those polls may have been accurate, but done in by a superior get-out-the-vote effort, or by very late deciders whose motivations may or may not ever be known," he said.