How Open Are We?

U.Va. professors investigate the truth about race and gender in American politics.

By Dan Morrell
This is a table for Nick Winter's research

There were early signs that Americans were excited by this year’s presidential election: the packed high school gymnasiums-turned-caucus-sites in Des Moines, the masses waiting in the early morning for New Hampshire primary polling stations to open, the fleet of copiers put to use as throngs of voters threatened ballot shortages from Connecticut to Colorado on Super Tuesday. This enthusiasm was constant through the hard-fought primary battles and palpable among Lynn Sanders’ students this spring. “It was almost impossible to talk about our boring political science material,” says Sanders, associate professor of American government, with a chuckle. “We talked about the primaries half the time.” (This, of course, in a political science class, is notably relevant.)

And the faculty was just as energized. A serious campaign for president that included both a female candidate in Hillary Clinton and an African-American candidate in Barack Obama has no precedent. It’s the poli-sci equivalent of the English department stumbling upon an unpublished Frost poem or astronomy finding a new galaxy.

“Usually, presidential general elections are reasonably predictable,” says Assistant Professor of Politics Nick Winter. “We’ll know within a handful of points how things will play out.” Who votes for whom, who is undecided and where they come from—these quantities are somewhat known when you have historical constants of white male contestants. But with this election, America’s internal battles with race and gender materialized in the most public of ways, throwing a wrench into that predictable system and rendering carefully laid blueprints woefully inadequate.

As the race progressed through the spring, the talking heads and the columnists began to offer some new data on how the country was reacting—polls on how race affected voting, surveys on the likelihood of voting for a woman—revealing some old national wounds that still haven’t fully healed. And while the numbers gave a quick answer to some pressing questions, U.Va. professors are doing research on the mechanics behind the stats, which tell a much more complex—and perhaps more complete—story.

Shades of Gray

To anyone who questioned what role, if any, race would play in this year’s election, exit polls in the Pennsylvania Democratic primary in April made it clearer: One out of five whites surveyed plainly said a candidate’s race was a factor. “And those,” says Vesla Weaver (Government, English Language and Literature ’01), assistant professor of politics, “are just the ones willing to admit it.”

So for those who considered race an issue in their vote, what is the effect? It may depend not only on the thick line between black and white but also the gradient of skin color separating the two.

A few years ago, Weaver measured the response of white voters to lighter- and darker-skinned black candidates using a series of online tests with imaginary candidates. In head-to-head match-ups pitting black candidates (both dark- and light-skinned) against white candidates—with controls for ideology and candidate and respondent characteristics—Weaver found that whites generally preferred the white candidate. However, when respondents did choose a black candidate over the white candidate, they preferred the darker one. “When a black ran against a white, the norm of social equality is triggered,” says Weaver. Because the race is more emphasized in the darker candidate, she argues, the white respondents think they are being asked about race and will vote to show they have no biases. But, using the same formulation, when Weaver ran a lighter-skinned black candidate against a darker-skinned black candidate, the lighter candidate was preferred 40 percent more. “When you are viewing two black candidates, the norm is not triggered—you’re not making a racial decision.”

Her research on skin shade includes work as a co-author of Unstable Boundaries: Skin Color, Immigration, and Multiracialism in the American Racial Order, a book project underway that, in part, details the disparities between the outcomes of darker- and lighter-skinned blacks and Hispanics, including lower incomes, high incarceration rates and higher execution rates for the dark-skinned. Even when controlling for background similarities like parental involvement, the dark-skinned minorities average a few months more on prison sentences and take home nearly $10,000 a year less than their light-skinned counterparts.

This disparity carries over to politics as well. “We know that throughout history, light-skinned blacks have been much more likely to be elected to office and much more likely to hold the first high-powered appointments—with the notable exceptions of [U.S. Supreme Court Justice] Clarence Thomas and [former Chicago Mayor] Harold Washington,” says Weaver. And while Weaver doesn’t put Obama, who is mulitracial, in the light-skinned category, she sees differences. “His whole appearance is less afro-centric,” says Weaver. And not just skin color, but the pitch of his voice, his manner of speaking and the kinkiness of his hair.

The disparity among the various skin gradients isn’t talked about. “We don’t attend to this in our policies, our speeches or our political discourse,” says Weaver. Affirmative action, for instance, applies no matter the skin tone. And no one is proposing extra policies to address the disparities suffered by those with darker skin. “We just don’t have the language to talk about it.”

The Communal Truth

The language of prejudice, however, has long had a voice. And those Pennsylvania voters who said race was a factor were not alone. As an Inez, Ky., voter told New Yorker writer George Packer, “I won’t vote for a colored man. He’ll put too many coloreds in jobs.”

But would this person have expressed this opinion if Packer were a black reporter? And what would happen if a black man with a clipboard were polling those Pennsylvania voters who said that race was a factor in their voting?

Associate Professor Lynn Sanders says that the responses of whites answering to white interviewers has been shown to be very different than those of whites being interviewed by blacks, with the opinions expressed shifting to conform to a more moderate stance to agree with the perceived beliefs of the black interviewer.

Here, the definition of truth becomes tricky. If a white person shares prejudice toward blacks in an interview with another white person and then tells a black interviewer that he has no prejudices, which response is truer? And if he would have pulled a lever in private for a white candidate but instead were coaxed into voting for a black candidate by political pressures, is his vote still legitimate?

“Why is it the case that the secret ballot is truer than the public vote?” asks Sanders, who is working on a book about the history of using survey research to study politics. “If people in a more public setting are pushed or drawn or provoked into thinking a different way than in the voting booth, it’s not necessarily less true.” These kinds of pressures, she argues, are completely appropriate—legitimate social pressures provoked by the Civil Rights movement that push people to live up to egalitarian ideals and the ambitions of colorblind society.

The issue came into sharp focus in the Democratic campaigns’ differing results in caucus states versus primaries states.

Primaries offer the secrecy of a curtained voting booth that can ostensibly hide prejudices or just the appearance of them. But in caucuses the voting is public, with people literally amassing in groups based on who they are supporting, with a count of hands taken to tally the votes. This year’s results between the two different methods were very different. Columbia journalism professor Tom Edsall, writing at, noted that “in the 14 states that picked some or all of their delegates through caucus systems this year, Obama won 400 delegates to Clinton’s 193, a 207-delegate advantage that more than accounts for his overall delegate lead.”

“When it seems easier to vote for Obama in a caucus rather than a secret ballot, it’s quite arguably a legitimate pressure,” says Sanders. (The pressure for Clinton was not the same: Without an antisexist norm in American politics that is equivalent to the antiracist norm, Clinton reaped no benefit in the public setting.) And it’s not a lie, Sanders argues, when someone buries a prejudice in front of a crowd that they would reveal in a voting booth or to a white interviewer. In effect, this is what is supposed to happen. “When people who’ve been living among their own race or living in a segregated setting for a long time start to encounter new social and political experiences, we would hope that they begin to think in different ways. And one of the things that is going to provoke them to think in different ways is to be around different people—and that is not necessarily bad.”

Invisible Gender

In March, former vice presidential candidate and Clinton finance committee member Geraldine Ferraro told the Torrance, Calif., Daily Breeze that “sexism is a bigger problem” than racism. “It’s OK to be sexist in some people’s minds,” Ferraro told the paper. “It’s not OK to be racist.” The statement followed a much-discussed op-ed by feminist icon Gloria Steinem in The New York Times in January, in which Steinem argued that gender is “probably the most restricting force in American life, whether the question is who must be in the kitchen or who could be in the White House.” So began the great 2008 media battle of race versus gender.

This whole construct was misleading, says Nick Winter, whose Dangerous Frames: How Ideas About Race and Gender Shape Public Opinion (University of Chicago Press) appeared earlier this year. For one thing, this sort of adversarial view assumes that only Obama has race and only Clinton has gender. And it also assumes a simplistic view of gender that ignores a deeper, more complex view of candidates based on gender stereotypes. “Gender is not just about having either male or female candidates,” he says.

For instance, some of the public stereotypes associated with Obama, Winter says, are more in line with traditional gender roles for females, while Clinton projected more masculine characteristics. “She does shots in a bar; she’s the girl who wears pantsuits and hangs with the guys,” says Winter. “Obama advocates talking to our enemies, collaborating, working together, listening to all viewpoints—these are all stereotypically feminine characteristics.”

A good number of the masculine characteristics Clinton exhibited—more hard-line, more hawkish—are also more typically associated with the Republican party of the last 40 years and with leadership in general. But there is a downside to these characteristics, Winter says, noting the perception that President Bush has been overly aggressive, with isolating results. “If any election is going to be one where they shy away from the masculine candidate, this may be it.”

Winter conducted pilot studies in the spring and summer to expand on these ideas, asking respondents to rate how much they associate different character traits (emotional, independent, assertive) and issues (the environment, inflation, unemployment) with both political parties and genders.

The responses showed clear connections between the gender associations and party associations, with the Republicans being seen as more masculine and the Democrats more feminine. The Republicans, for instance, were associated with defense and terrorism as well as assertiveness and aggressiveness, issues and traits that were also deemed to be masculine. Democrats were aligned with compassion and sensitivity, which, along with the issues of the environment and poverty, were associated with the feminine.

The study is part of Winter’s forthcoming book, tentatively titled The Secret History of Gender in American Politics. “When we think of gender, we think of the suffrage movement, the women’s movement,” says Winter. “But what I want to show, in a variety of other areas where explicit gender issues weren’t necessarily on the agenda, nonetheless, people’s concerns, anxieties and thoughts about gender have structured how they think about politics.”

Buried Bias

This idea that we make choices based on traditional gender stereotypes suggests that some of the political choices we make aren’t really choices at all. They’re just reactions—visceral responses to stimuli on some very basic level. Even if we don’t know it, we may be making choices based on negative feelings toward blacks, women or, in John McCain’s case, the aged. “We don’t observe our mental operations,” says Brian Nosek, associate professor of psychology. “We only experience them.”

Nosek’s interest in politics emerged while he was doing research in implicit cognition, which examines thought and feeling outside of awareness and control. These kinds of implicit reactions are often useful: the natural reaction to the taste of sour milk or poison keeps you from getting ill, for instance. It’s possible, says Nosek, that some of these reactions have evolutionary roots, influenced by our experiences and exposure to things like media messages.

“They’re hard to escape, because we can’t turn them off,” says Nosek. Because we can’t run away from them, we rationalize. In the case of politics, when your negative reaction to the candidate can’t be explained away by a belief that you are unbiased, then it must be something else—foreign policy, a health-care plan or some other policy position.

Nosek—along with collaborators Anthony Greenwald from the University of Washington and Mahzarin Banaji of Harvard University—set up an online experiment during this year’s primaries to test respondents’ implicit reactions to the candidates. Named the Implicit Association Test, it looks to strip away a respondent’s layers of consciousness and reveal their implicit reactions by measuring how long it takes respondents to match positive words with black faces (including Obama’s) versus the time it took to match negative words with the images. Under rapid-fire conditions, the faster you can hit the keys to associate black faces with negative words, the more you likely have implicit negative reactions.

While the 50,000 respondents to the Web-based presidential-candidates task test were mostly those who would find their way to an online site like this—highly educated, liberal and more pro-Obama—the test found that most of the respondents didn't like the senator from Illinois as much as they said they did in prescreening questions. “Support for Obama, we found, is implicitly weaker than his support explicitly.”

Nosek has also been working with the American National Election Study to include an IAT in its data collection for 2008. And this time, while lacking the volume of his site’s nearly 2,000 visitors a week, the IAT responses gathered by the ANES will be more representative of the American public.

But the results won’t be available until after the nation chooses its next president, meaning Nosek and his collaborators will have to wait until after November to see if implicit reactions can be useful in predicting elections.

And even with Senator Clinton out of the presidential race, Obama’s run for the presidency will be giving faculty at U.Va. and around the world new data to comb through, new questions to answer, and maybe a clearer view of how the country reacts to race and gender—every poll, survey and headline a brand new piece of political science.

“I’m watching the news just like everyone else,” says professor Vesla Weaver. “Everything here is new.”

Watch a webcast of the Miller Center’s roundtable event on “Race and Gender in American Politics,” presented by Arts & Sciences magazine.

Read about the roundtable: “Panelists Ponder the Weight of Prejudice in the Presidential Election.” 

Listen to reflections and a recap on the Oscar Radio Show.

Listen to Lynn Sanders and Vesla Weaver discuss race and gender in American politics on Charlottesville Right Now.