The issue of race is the longest-lasting cleavage in American politics. It is also perhaps the least understood. The open exploitation of racist sentiment by vote-hungry politicians was for centuries a durable American tradition. More recently, race has assumed a subtle, often unspoken form during campaign season, as Republicans have sought white votes by slyly associating their Democratic opponents with controversial black figures like Jesse Jackson and Al Sharpton, or with topics--welfare, crime, federal funding for "midnight basketball"--that many voters identify with African Americans.
Now, with Barack Obama inching closer to the Democratic nomination, race looms yet again as a central factor in American politics. Already, race has played a key part in the Democratic primary, almost certainly hurting Obama among swaths of voters in states like New Jersey, Ohio, and, most recently, Pennsylvania. If he manages to win the nomination anyway--and it appears he will--race seems likely to play an even larger role in the general election.
What role, exactly, will that be? No one knows for sure, but the field of political psychology offers some clues. In recent years, scholars have been combining experimental findings with survey data to explain how race has remained a factor in American elections--even when politicians earnestly deny that it plays any part at all. In 2001, Princeton political scientist Tali Mendelberg summarized this research in a pathbreaking book, The Race Card. Her provocative analysis is hotly debated and far from conclusive; political psychology, after all, is not a hard science. Still, her ideas and those of other academics help to shed light on what has happened so far in the primaries and what might unfold once Obama wraps up the nomination. Their findings suggest that racism remains deeply embedded within the psyche of the American electorate--so deep that many voters may not even be aware of their own feelings on the subject. Yet, while political psychology offers a sobering sense of the difficulties that lie ahead for Obama, it also offers something else: lessons for how the country's first viable black presidential candidate might overcome the obstacles he faces.
If you were born before 1970 or if you read public-opinion polls, then you cannot doubt the profound transformation wrought by the civil rights era. In 1944, the National Opinion Research Center asked whether "Negroes should have as good a chance as white people to get any kind of job, or [whether] white people should have the first chance at any kind of job"--and 55 percent still thought white people should have the "first chance." By 1972, only 3 percent thought so. But some academics--noting the bitterness of battles over busing, affirmative action, and aid to cities, as well as the evolution of the GOP into a virtually all-white party--reasoned that racial prejudice remained, even if it was no longer overtly expressed. They believed it had simply changed form. Their challenge was to define and to demonstrate the existence of this new racism.
Many social scientists had long rejected the possibility that humans might harbor unconscious attitudes different from their conscious behavior. But, in trying to explain the persistence of racial prejudice, political psychologists were forced to hypothesize different levels of awareness and motivation. On the highest level was public moral reflection guided by social norms--which led to Trent Lott being pilloried when he famously said in 2002 that, if Dixiecrat Strom Thurmond had been elected president, the country could have avoided "all these problems." Beneath this, however, was a realm of knee-jerk opinion that might contradict a person's moral reflections; and still beneath that were unconscious attitudes, which, like a person's knee-jerk opinions, were often at odds with his or her public moral reflections. If racial prejudice persisted, it was on these deeper levels.
Political psychologists devised new tests to uncover these sentiments. First, they crafted survey questions aimed at unearthing what they called "symbolic racism," "modern racism," and, most recently, "racial resentments," which ascribe to blacks as a group certain negative attributes or undeserved advantages. For example, researchers asked respondents whether they agreed or disagreed with statements such as "It's really a matter of some people not trying hard enough; if blacks would only try harder they could be just as well off as whites" or "Over the past few years, blacks have gotten more economically than they deserve."
Experimenters then inserted questions like these into the American National Election Studies (ANES), extensive biennial surveys funded by the National Science Foundation. The answers revealed a degree of racial resentment that wasn't apparent from more explicit questions about racial bias. In 1986, for instance, 59 percent of respondents agreed that blacks were not trying hard enough (only 27 percent disagreed), while 67 percent thought blacks should work "their way up ... without any special favors." Psychologists David Sears and Donald Kinder, as well as others, found that this racial resentment was the single most important factor--more important than even conservative ideology or political partisanship--in explaining strong opposition to a host of government programs that either directly or indirectly benefited minorities. Of course, that doesn't mean there couldn't be principled conservative opposition to government-guaranteed equal employment or urban aid. But, according to the political psychologists, racial resentment played the largest role in fueling public skepticism.
The answers also revealed which groups within society continued to harbor racial resentment. With the help of Harvard doctoral student Scott Winship, I looked at the levels of racial resentment in ANES data from 1988, 1992, and 2000 (the questions were omitted in 1996). What Winship and I found was that resentment was highest among males rather than females, the middle class rather than the wealthy or poor, those lacking a college degree, those who worked in skilled or semi-skilled blue collar jobs or as laborers, and residents of small towns in the Midwest and South. Does that profile sound familiar? It's more or less a description of the white working-class voters who have spurned Obama and with whom John Kerry and Al Gore had trouble. The only groups that didn't evince racial animosity toward blacks were voters with post-graduate degrees and, of course, African Americans. Hispanics were nearly as prejudiced as whites, and a group labeled "other" that includes Asian Americans was even more so--a partial explanation, perhaps, for why Obama fared so poorly among these groups in California. Clearly, racial resentment persisted--just in a more nuanced form.
In fact, the structure of this modern racism was even more complicated than the ANES data suggested. In a study published in 1995, four psychologists from Indiana University recounted taking a group of subjects who had earlier taken the racism test (the questions had been interspersed among scores of other questions) and giving it to them again. This time, however, a black experimenter conducted some of the tests and a white experimenter the others. The psychologists discovered that, when the interviewer was black, white respondents scored substantially lower on the racism scale than before. This meant that gut-level reactions could be easily influenced by moral reflection and social norms. What psychologists needed was a method of measuring prejudice that elicited immediate emotional reactions rather than the products of deliberation.
Toward that end, they devised tests that measured racial attitudes without subjects knowing what was happening or being able to adjust their responses to social norms. In a study that appeared in 1989, University of Wisconsin psychologist Patricia Devine flashed words on a screen faster than her subjects could recognize them. Some of the words, like "blacks," were associated with African Americans; others were neutral. She then asked subjects whether a person's actions in a deliberately ambiguous story about a customer wanting his money back signified hostility or not. After words associated primarily with African Americans were flashed, the subjects rated the person's actions decidedly more "hostile" than after predominately neutral words were flashed. This suggested to Devine that terms associated with blacks were priming unconscious stereotypes about aggressiveness or hostility.
Another kind of test--known as an implicit association test--used the time it took to complete word association exercises to unmask stereotypes. Psychologists would ask subjects to associate positive and negative adjectives with African American and European American faces by pressing different keys on a computer. At each interval in the experiment, subjects would be told which kind of adjectives to pair with which subject. If a subject regularly took longer to pair positive words with a black face than he did negative words, that indicated unconscious racial bias.
Using data from more than 15,000 self-selected subjects who took the test on a website, psychologists Anthony Greenwald, Mahzarin Banaji, and Brian Nosek found that the same sorts of respondents who had registered higher on the racial resentment scale were more inclined to associate negative adjectives with an African American face. For instance, subjects who had not graduated from college displayed more prejudice than those who had. Men also were more prejudiced than women.
In addition, according to questions they answered before taking the test, there was a sharp disparity between what subjects said they believed and what the test showed. For instance, only 32 percent of high school graduates said they favored whites over African Americans, but in the test 64 percent did. This disparity suggests that, in answering questions about what they believed, subjects opted for prevailing norms over private sentiments. They did not want to appear racist, even though, at some level, they were.
But the problem with implicit association tests--or tests that use subliminal cues--is deciding what they mean in the real world. Do the unconscious racial feelings they uncover affect the way whites view policy, parties, and politicians? And, if so, how decisively?
In elections over the last three decades, Republican politicians have repeatedly used ads, push polls, and surrogates to appeal to white voters' racial fears and resentments. These ranged from George H.W. Bush's Willie Horton ad in 1988 to the Republican National Committee's infamous "Harold, call me" ad in the 2006 Tennessee Senate race between Republican Bob Corker and African American Harold Ford, which appealed to long-standing fears about black sexuality.*
In The Race Card, Mendelberg argues that political ads can indeed awaken underlying racial stereotypes and attitudes, just like a black face flashing subliminally on a screen. She recounts the story of the Horton ads, which first appeared in June but became widespread in October. While the press initially presented them as being about crime, Horton's picture also appeared--stirring unconscious racial associations. Using the ANES survey, Mendelberg drew connections between the ad's prominence and support for Bush against Michael Dukakis. From June to October, she shows, there was a mild correlation between how high a voter ranked on the racial resentment scale and his or her support for Bush. After October 1, when the ads were more widely aired and discussed, the correlation became much stronger.
In a similar vein, Nicholas Valentino, Vincent Hutchings, and Ismail White from the University of Michigan ran a series of experiments in 2000 using a hypothetical George W. Bush campaign ad that promised to cut taxes, reduce wasteful spending, and reform the health care system. There were three versions of the ad, all with the same text: One showed neutral images of laboratory workers and medical x-rays; another showed a black person counting money and a black mother and child in an office just as the narrator announced Bush's opposition to spending "tax dollars on wasteful government programs"; a third showed white people in the same roles. After watching the ads, subjects were asked to fill out questionnaires that included measures of racial resentment and their preference between Bush and Gore. The findings were similar to Mendelberg's. After seeing the neutral ad or the ad picturing undeserving whites, subjects who scored higher on the racial resentment test were no more likely to support Bush than subjects who scored lower. But, when subjects saw the ad picturing undeserving blacks, there was a strong correlation between how high they scored on the racism scale and their support for Bush. "Overall," the political psychologists wrote, "these results suggest that racial cues make racial concerns more accessible in memory, subsequently boosting the impact of these concerns on candidate evaluations."
Mendelberg's most controversial claim is that these ads work best when the appeal is implicit. If the appeal is explicit, she argues--that is, if politicians actually say that blacks are undeserving--then they lose support because they have violated the norm against racism. Although voters will respond unconsciously to an implicit appeal that they don't perceive as racist, they will recoil for reasons of conscience or social disapproval to an appeal that either is, or is seen as, racist. Mendelberg asserts that Bush actually lost support to Dukakis in the closing weeks of the 1988 campaign because, on October 21, Jesse Jackson denounced the Horton ad as racist and Dukakis's running mate Lloyd Bentsen followed suit two days later. That made explicit what had previously been implicit.
But other psychologists have questioned Mendelberg's theory of implicit and explicit racial communications. Two political scientists, Gregory Huber and John Lapinski, tried to test Mendelberg's theory by comparing the effects of two anti-welfare ads: one using a visual image of a black recipient and the other adding explicit language about "too many welfare recipients, especially blacks," taking "advantage of our tax dollars." They found that the explicit message did not produce a "significantly more liberal policy opinion than the implicit message." Karen Kaufmann, a political scientist at the University of Maryland, is also skeptical about Mendelberg's theory. "My own work in the context of local politics suggests that fairly explicit racial appeals succeed quite often," she says.
Some distinctions might help preserve what is valid in Mendelberg's argument. First, one has to distinguish between kinds of explicit messages. If a message obviously violates norms against old-fashioned, pre-civil rights racism--as Lott's did--then it is likely to backfire; but if it leaves any room for disagreement about whether it is racist, then it may not. Second, a lot depends on which voters candidates are appealing to. For example, researchers have shown that white women are more likely than white men to react negatively to racist appeals. George Allen learned this lesson in his 2006 Senate race when he lost support among female voters in northern Virginia after uttering a racial slur against an Indian-American.
But there are clearly limits to how much charging racism can help Democrats. During the 1988 election, Dukakis did surge in the last weeks, but it was at best only partly attributable to denunciations of the Horton ad as racist--at the time, most political analysts attributed it to Dukakis's belated adoption of a populist economic appeal. And Dukakis's recovery in those last weeks by no means made up for the ground he lost in the first weeks of October when the Horton ad dominated the airwaves. Even Mendelberg acknowledges that the Horton ad "helped George Bush win the election." Similarly, in the 2006 Tennessee Senate race, Ford was able to make up some ground after the RNC's* "Harold, call me" ad was denounced as racist, but he was not able to undo the damage the ad initially caused.
So making an implicit racial appeal explicit can help an embattled Democrat among some groups, some of the time--but it is hardly a surefire defense against the race card. That leaves one other option: changing the subject. In a campaign where a large proportion of voters would score high on a racial resentment test, a politician's best hope of countering the race card may be to simply emphasize a more important issue. Mendelberg herself acknowledges that "an effective defense against implicitly racial appeals requires an issue that trumps race in the considerations of white voters. In the nineteenth century, this issue was primarily sectionalism--northern whites' resentment of southern whites' secession. In the twentieth century, this issue was primarily economic prosperity and cherished social welfare programs." Bill Clinton used this approach successfully in 1992--a year when the Los Angeles riots had inflamed racial sentiments--to win back white working-class voters who had left the party in prior decades.
What, then, can the political psychology of race tell us about the current primaries and the coming general election? Clearly, Obama gained some votes in the early primaries from college-educated Democrats who liked the idea of an African American candidate transcending the historic conflict over race. And, if he had not been running against a popular female candidate, he might have won more support among white women. But Obama also lost voters to racial prejudice.
One indication is the exit polls. The percentage of voters who backed Hillary Clinton (or, earlier, John Edwards) while saying that the "race of the candidates" was "important" in deciding their vote is a fair proxy for the percentage of primary voters who were disinclined to support Obama because he is black. That number topped 9 percent in New Jersey; in Ohio and Pennsylvania, two crucial swing states, it was more than 11 percent. And that's among Democratic primary voters, who are, on average, more liberal than the Democrats who vote in general elections.
Obama's connection with the Reverend Jeremiah Wright, which exploded into the news after the Ohio primary, may do lasting damage to his candidacy by undermining his attempt to transcend race. Wright's words tie Obama to the stereotype of the angry, hostile--and also unpatriotic--black who is seen as hating both whites and white America. Wright turns Obama into a "black candidate" like Jackson or Sharpton. And, as a black candidate, Obama falls prey to a set of stereotypes about black politicians.
Some of these have to do with abilities. A 1995 study found that voters believe black politicians "lack competence on major issues." Other stereotypes relate to ideology. Several studies have shown that if subjects compare a black and white candidate with roughly equal political positions, they will nevertheless see the black candidate as more liberal. Obama is already vulnerable to charges of inexperience, and, after Wright surfaced, he fell prey to an ideological stereotype as well. Whereas he benefited in the initial primaries and caucuses from being seen as middleof-the-road or even conservative, his strongest support has recently come from more liberal voters. In Pennsylvania, he defeated Clinton among voters who classified themselves as "very liberal" by 55 to 45 percent, but he lost "somewhat conservative" voters by 53 to 47 percent and moderates by 60 to 40 percent. In a national Pew poll, Obama's support among "very liberal" voters jumped seven points between January and May, while his support among "moderates" dropped by two points. Since Obama's actual policies are, on the whole, no more liberal than Clinton's (his health care plan, for instance, is inarguably more conservative), these trends strongly suggest that some voters are stereotyping him because of his race.
If Obama wins the Democratic nomination, he should be able to inherit the white women who backed Hillary Clinton. As political psychologists have shown, these voters should be largely amenable to his candidacy. He should also continue to enjoy an advantage among white professionals. But Obama is likely to continue having trouble with white working-class voters in the Midwest--voters who tend to score high on racial resentment and implicit association tests and who, arguably, decided the 2004 election with their votes in Ohio. Obama will also have trouble with Latinos and Asians, groups that score high on both indexes, and that can be important in states like California. It's not hard to quantify Obama's problem: If 9 to 12 percent of Democratic primary voters in swing states have been reluctant to support him because he is black, one can assume that, in the general election, 15 to 20 percent of Democrats or Democratic-leaning Independents may not support him for the same reason.
Can Obama surmount these obstacles? If the strong version of Mendelberg's thesis is correct, then the very fact that Obama is African American will undercut any appeals to racial fears or resentments. And, if elections were held in the manner of the Iowa caucus, where voters publicly debate their positions and where Obama won substantial white workingclass support, then Mendelberg's stronger thesis might well prove true. But elections are held in the privacy of a voting booth, where a voter can give voice to fears and resentments without danger of being heard. Obama may be able to sway some white voters to his side by drawing attention to race, but probably not enough to fully compensate for the disadvantage he faces.
If addressing racial resentments directly is not the answer, what is? As Mendelberg also suggests, it's changing the subject--doing what the Republicans of the 1870s and the Democrats of the 1990s did. This year, that means diverting voters' attention from the politics of race to the plight of the economy and the continuing quagmire in Iraq.
In the end, the lesson of political psychology for Democrats is not to avoid nominating black candidates. It is simply to understand that America's racial history continues to influence the calculations of voters--sometimes near the forefronts of their minds, sometimes in the deep recesses of their unconscious. For liberals, acknowledging these obstacles is the first step to blunting them. If Obama can focus the election on the economy and Iraq, he could very well win in spite of the angry words of Reverend Wright and 200 years of both old- fashioned racism and newfangled racial resentment. If he can't, he is likely to suffer the same fate as Michael Dukakis--and this time it won't take a Willie Horton commercial.
John B. Judis is a senior editor at The New Republic.
* Clarification (5/14/08). This article was changed to make clear that the "Harold, Call Me Ad" in the 2006 Tennessee Senatorial race did not come directly from the campaign of his opponent, Bob Corker, but was produced by the Republican National Committee on its behalf.