washingtonpost.com
The Real Story In New Hampshire

By Andrew Cline
Saturday, January 19, 2008

The polls showed Illinois Sen. Barack Obama leading New York Sen. Hillary Clinton by big margins in the weeks before the New Hampshire primary. But on the day of the primary, Clinton won by three percentage points. Why the discrepancy? Why, New Hampshire voters must be racist, of course.

New Hampshire is 95 percent white, some have pointed out, so that explains Obama's loss. The data, however, show otherwise.

The sole explanation offered for the racism argument is that New Hampshire voters did not want their secret racism exposed, so they told pollsters they would vote for Obama. This is known as the Bradley effect. In 1982, Los Angeles Mayor Tom Bradley ran for governor, and polls had him with a large lead. But on Election Day, Bradley, who was black, lost to his white opponent. It appeared that white voters had lied to the pollsters.

For the Bradley effect to have caused the New Hampshire Democratic primary results, however, New Hampshire voters would have to have decided on the weekend of Dec. 15-16 to begin hiding their latent racism by lying to pollsters. From Dec. 18, 2006, to Dec. 18, 2007, Hillary Clinton led in 53 of the 59 polls conducted in New Hampshire.

Beginning in mid-December, Democratic and Democratic-leaning independents began telling pollsters that they supported or were leaning toward Obama. It defies explanation to assert that white New Hampshire voters suddenly decided to start hiding their racism on that weekend after publicly supporting Clinton for a year.

Furthermore, the Bradley effect is defined by a big discrepancy between polled support for the black candidate and that candidate's actual support on Election Day. That didn't happen in New Hampshire. Obama averaged 38.3 percent in the polls and got 36.4 percent of the vote -- a difference of fewer than two percentage points and well within the margin of error. Obama's support in the exit polls was also the same as his support in the voting.

The discrepancy was in Clinton's support. She polled around 30 percent but got 39 percent of the vote. Obama's support was not so much overpolled as Clinton's was underpolled.

The Bradley effect did not happen in the New Hampshire Democratic primary. But two questions remain unanswered: Why did Obama surge in the polls in mid-December? And why did the polls not catch Clinton's surge on primary day?

The answer to the first question is easy: Oprah.

On Dec. 9, Oprah Winfrey stumped for Barack Obama in Manchester. She drew 8,500 people and made huge news for days. Two days later, Obama beat Clinton in a New Hampshire poll for the first time since July.

Until mid-December, women consistently said that they preferred Clinton to Obama by a big margin. By early January, that had shifted, and the polls showed Obama and Clinton statistically tied among women. Oprah drew the attention of New Hampshire women to Obama, they liked what they heard and some began telling pollsters that they were leaning toward him.

On primary day, however, almost half of all women who voted in the Democratic primary voted for Clinton.

Many factors were behind the shift to Clinton: There were the much-discussed moments in which Clinton expressed vulnerability and doubt -- her answers to the debate question about her likeability and the question in the Portsmouth coffee shop about the strain of the campaign. John Edwards, Chris Matthews and even Obama seemed to gang up on her in the final days.

Another factor, much less reported, is that the Clinton campaign sent out a mailer a few days before the primary portraying Obama as a less-than-solid supporter of abortion rights. Consider also that Clinton's support came largely from lower-middle-class families, who are routinely underpolled. Samples are weighted to account for that, but that technique does not always work.

Even two days before the primary, nearly half of Democratic voters remained undecided. Many of them told pollsters they were "leaning" toward a candidate, but they had not made up their minds. Obama had more "leaners" than Clinton, somewhat inflating his numbers. In the last day or so, women "leaners" broke from Obama and the other male candidates and joined undecided women in voting for Clinton in big numbers.

Whatever the polls said, some well-informed New Hampshire Democrats privately predicted that Clinton would win or come close. They picked up what the polls did not -- that many Democratic women really wanted to vote for Clinton but felt it was their duty as informed voters to check out all the candidates. Those informed Democrats also said Clinton had an operational advantage that could bump her final numbers. Clinton's operation did a masterful job getting her supporters to the polls. That probably accounted for some of her large margin of victory in New Hampshire's two largest cities.

All of the evidence points to one conclusion: Large numbers of women, many of whom wanted in their guts to vote for Clinton, looked closely at Obama, Edwards and Richardson, then decided late to vote for Clinton, either because she was a woman or because, as many said, they thought she was the best candidate. That, not race, is the story of her surprise victory in the New Hampshire primary.

Andrew Cline is editorial page editor of theNew Hampshire Union Leader.

View all comments that have been posted about this article.

© 2008 The Washington Post Company