Will whites vote for Barack Obama?

Obama has attracted white votes in Democratic primaries, but how would he fare with white America if he were the party's nominee?

By Alex Koppelman

Read more: Racial Issues, Politics, African-Americans, Race, News, Barack Obama, 2008 election, Alex Koppelman

Salon photo composite/Reuters image

Jan. 24, 2008 | Discussions of race have become a constant in the discourse about the upcoming Democratic primary in South Carolina, and not just because Hillary Clinton and Barack Obama have been sparring over the issue. Saturday's primary will be the first Democratic contest in a state with a large black population; African Americans, in fact, could well comprise a majority of the voters who show up at polling places. Blacks make up 29 percent of the state's population but provided 63 percent of John Kerry's South Carolina vote total in the 2004 general election. If they make up a similar percentage of Saturday's electorate, and if current polling resembles the final outcome, Obama could conceivably win South Carolina before a single white vote is counted.

But Obama will have to win plenty of white votes nationwide if he wants to be the Democratic nominee, not to mention win the general election in November. In Iowa and New Hampshire, he's already proven himself capable of attracting certain kinds of white voters, namely the young, affluent and educated voters so well-represented in the Democratic party's base. But can he also do well among the white electorate at large if he becomes the nominee? And if he does become the nominee, given the controversy over polling in New Hampshire, should we trust any future poll numbers that purport to show him doing well, especially in key battleground states that could be decided by narrow margins? The answers to both questions, at least according to most observers, are "yes" and "yes." Most experts believe that whites now respond honestly when asked whether they'll vote for a black candidate. And when it comes to willingness to vote for and elect African American candidates in a bi-racial election, the United States seems far ahead of where it was just a decade ago. But this may have as much to do with a change in African-American candidates as with a dramatic change in voter attitudes.

After Hillary Clinton's surprise victory in New Hampshire's primary, which seemed to contradict polls taken just days earlier, many pundits and pollsters reached for race as an explanation. They referred to the "Bradley Effect," the possibility that polls -- even those restricted to Democratic voters in a Northeastern state -- could be skewed simply by the presence of an African-American candidate in the race and whites' reluctance to appear racist by telling pollsters they would not vote for him.

The Bradley effect is named for Tom Bradley, the former mayor of Los Angeles. In 1982, Bradley, an African-American, ran for governor of California; pre-election polls gave him a clear lead, but when it came to Election Day, Bradley lost a close race. A similar phenomenon was observed the next year in Chicago, where Harold Washington, also an African-American, eked out a victory in a mayoral election despite pre-election polling that had Washington walking away with the race. In the late 1980s and early 1990s, five other biracial elections featured similar disparities between polls and the actual vote. Social scientists and pollsters theorized that this might have to do with something called "social desirability bias." When called by pollsters, especially African-American pollsters, this hypothesis goes, whites who do not want to vote for an African-American candidate will feel embarrassed about being perceived as racist if they express that sentiment, and they will lie. Then, when they head to the voting booth, their real preferences are exposed.

With the rise of Obama this election cycle, the return of discussion of the Bradley effect may have been inevitable. But the shock of Hillary Clinton's victory in the New Hampshire primary earlier this month, when polls had shown her trailing badly in the days just before the vote, jump-started the debate about the issue. Commentators looked at a host of potential factors in the failure of polling to accurately predict the race and dismissed many of them -- for example, they said the turnout models looked accurate, since the polls had called the Republican side of the equation correctly, and exit polls purportedly showed that there was no last-minute swing to Clinton -- before landing on the Bradley effect as one possible explanation.

Andrew Kohut, the president of the Pew Research Center, was one of the experts most willing to posit the Bradley effect as a cause for the discrepancy in New Hampshire. But what Kohut proposed was something of a corollary; rather than asserting that poll respondents had lied to their questioners, Kohut said that the problem could be a lack of response altogether. In an Op-Ed piece for the New York Times, Kohut wrote, "Poorer, less well-educated white people refuse surveys more often than affluent, better-educated whites. Polls generally adjust their samples for this tendency. But here's the problem: these whites who do not respond to surveys tend to have more unfavorable views of blacks than respondents who do the interviews."

In a recent interview with Salon, Kohut reasserted the hypothesis from his Times Op-Ed. "What occurred to me, and this still rings true, is that a lot of the other logical explanations could be, if not ruled out, then they're not so persuasive," Kohut said. "So it could be that we had a systematic problem based on an old issue. What led me to that possibility is the profile of [Clinton's] support. The only thing that I didn't have in that piece, and I have looked at since, is if you look at the gap between what Gallup is showing in terms of socioeconomic status and the exit poll, the biggest differences are among poor people. Poor people were supporting Obama, and Hillary less, at much higher rates in the pre-election poll than in the exit poll. That's where the big gap is."

Still, Kohut cautions against jumping to any conclusions, saying he wasn't asserting the Bradley effect as an absolute in his Op-Ed, and that those interested in explaining what happened in New Hampshire will have to wait for other results to come in. "We're going to have a whole bunch of primaries on February 5, and if the polls have a problem, then we can explore along these lines or other lines. If, on the other hand, the polls do reasonably well on the Democratic side, then we'll chalk it up to the irascibility of New Hampshire voters," Kohut said.

Additional experts who spoke with Salon uniformly emphasized that they consider Kohut one of the leading lights of the profession, but were at best hesitant about agreeing with his position on this question. Gary Langer, the director of polling at ABC News, was one of those. He dismissed the idea of the Bradley effect, saying there has never been enough science to prove its existence, whether in one New England state two weeks ago or two decades ago anywhere else in America.

"The argument of a specific 'Bradley effect,'" insisted Langer, "still looks to me to like a theory in search of data ... I don't see why this effect would be limited, before now, to a handful of elections 15 to 25 years ago. And I don't know how to understand its absence in so many other black-white races -- five [Senate and governors'] races in 2006 alone, as I note -- in which pre-election polling was dead on."

"Newton's Law of Gravity doesn't just work on Thursdays," Langer said. "You want an effect to be clearly established as an effect through analysis of empirical data, and maybe in more than one election. And to call it an effect you want it to be a consistent effect, or to explain its inconsistency."

Next page: "The African-American candidates ... have run de-racialized campaigns, and that I think has made all the difference"