Eugene Robinson
Washington Post
Friday, January 11, 2008; A17
 

Pollsters and pundits were quick to discount race and the so-called Bradley effect as factors in Barack Obama  's narrow loss to Hillary Clinton  in the New Hampshire  primary. Given that the same pollsters and pundits (okay, me, too) were so wrong about the outcome, I think we ought to take a closer look.

The phenomenon is named after the late Tom Bradley  , who in 1982 seemed certain to become the first black governor of California . Polls showed Bradley, the mayor of Los Angeles, with a double-digit lead over his white opponent, George Deukmejian. But Bradley lost.

Subsequently, several high-profile races involving black candidates followed the same pattern in which apparent leads somehow evaporated on Election Day. The polls said David Dinkins  would beat Rudy Giuliani  by more than 10 points in the 1989 New York  mayoral race; Dinkins ended up winning with 50 percent of the vote to Giuliani's 48 percent. That same year, the polls gave Douglas Wilder an 11-point lead over Marshall Coleman in the Virginia  governor's race; Wilder squeaked into office by less than half a percentage point. In 1990, the polls said Harvey Gantt would handily defeat incumbent North Carolina  senator Jesse Helms; Gantt lost, and it wasn't even close.

Was it that voters told pollsters they intended to vote for African American candidates and then, in the privacy of the voting booth, chose white candidates instead? Not really. In each of these instances, preelection polls were quite accurate in predicting the black candidate's vote. What happened was that the polls greatly underestimated the vote for the white candidates. Unusually large numbers of self-described undecided voters ended up making the same decision.

Fast-forward to Tuesday in New Hampshire, and in broad terms that's what happened to Obama: His vote was roughly as predicted by election-eve polls, but Clinton's was dramatically bigger than expected. There are so many caveats, however, that it's impossible to diagnose the Bradley effect with any confidence.

For one thing, the phenomenon has been absent in other recent high-profile races in which black and white candidates competed. In Harold Ford's unsuccessful 2006 bid for a U.S. Senate  seat from Tennessee  , for example, most of the polls actually underestimated his vote -- and overestimated the vote for Ford's white opponent, Bob Corker , who won by just three percentage points. Ford was hurt by a racist television ad, to be sure, but my point is that at least the polls were close to the mark.

The Pew Research Center looked at this and other recent black-white statewide contests and concluded that "fewer people are making judgments about candidates based solely, or even mostly, on race itself."

The other big caveat is the evidence that gender, not race, was probably the most important facet of identity Tuesday. Clinton, obviously, is the first woman to have a realistic chance of being elected president. Women voted in unusually large numbers -- they outnumbered men at the polling places 57 to 43 percent -- and they went heavily for Clinton over Obama.

Clinton's much-covered display of emotion may have been the turning point, but I'm not sure it was more decisive than her extensive grass-roots organization or her energetic get-out-the-vote operation.

Still, there are a couple of anomalies. The exit poll done for the television networks indicated that nearly four out of 10 Democratic voters made their decisions in the three days before the primary. But the exit poll also indicated that those last-minute deciders broke equally for Clinton and Obama -- which pretty clearly was not the case.

Well into the evening, even the Clinton campaign expected Obama to win. Opinion polling isn't an exact science, but it's extremely rare for so many experienced pollsters to be so wrong. When you try to think of precedents, you keep coming back to races such as, well, Tom Bradley's and Doug Wilder's.

There are many reasons why New Hampshire voters might choose Clinton over Obama, or Obama over Clinton, or John Edwards over either of them. What happened in New Hampshire was weird, but it's not possible to conclude that racism played any role in Clinton's big upset. The dynamic of two potential "firsts" -- first female president, first black president -- means that history may be an unreliable guide. These are, indeed, uncharted waters.

We'll have plenty of chances in the coming weeks to measure preelection polls against actual results -- including in states with much more racial diversity than New Hampshire. But after Tuesday's big surprise, embarrassed pollsters and pundits had better be vigilant for signs that the Bradley effect, unseen in recent years, has crept back.

eugenerobinson@washpost.com