“The study’s not working anymore,” he said.

Jonassaint told her that the subjects’ “reactivity” -- that is, the up-and-down movement of their stress levels -- had suddenly stopped. He gloomily wondered if they would have to end the experiment.

“All the work that was going into it, it just didn’t seem to be worth it,” he said. “It was one of the most frustrating experiences of my life.”

The study, designed to test whether a strong sense of racial identity protected people from the effects of stress, included 33 African-American Duke undergraduates. They came into the lab and were asked to perform a stressful task -- in this case, they made a speech and were told it would be evaluated later. Six times during the process, they rinsed their mouths with water, then placed a dental swab treated with citric acid against their gumline for 30 seconds to get a saliva sample. The swabs were then dropped into a test tube and sent off to a lab to test for the hormone cortisol, which is secreted by the adrenal glands and increases when people experience stress.

Typically, people’s cortisol levels go up and down during the experiment: It is usually high at the beginning when they first walk in, then goes down, then shoots up again when making their speech, then sinks as they relax afterward.

But suddenly that movement had stopped. The subjects were coming in with high stress levels, and the levels were staying high.

Jonassaint had a theory about why this was happening: Could the high stress levels be a response to something happening on campus? People everywhere were talking about how a local woman had accused members of the Duke lacrosse team of assaulting her at off-campus party. The accusations would turn out to be false, and the players would be exonerated. But at the time of the experiment, the first news of the case was just coming out, often with racial and gender overtones.

“When he said this, bells went off,” said Richman, an assistant research professor in Duke’s Department of Psychology and Neuroscience. Far from stopping the study, she told him she wanted to keep it going and to refocus on what was happening. “I said, ‘Actually, this is really intriguing. We have to collect enough data before the end of the semester to get a ‘before’ and ‘after’ group.’”

From a methodological standpoint, this was a unique opportunity for Richman and Jonassaint. Richman’s work focuses on people’s perceptions of discrimination and how these affect their physiological, motivational and emotional processes. Although affected by many variables, a perception of discrimination has been found to be a factor in wide-ranging health disparities between white and black Americans. Researchers are less certain, however, exactly how and why the perceptions lead to physical changes.

In particular, Richman and Jonassaint sought to unravel the complicated scientific relationship between racial identity and stress. Previous studies had suggested that a strong sense of racial identity can buffer the effects of discrimination on health, perhaps by making people feel they have more social support and thereby increasing their ability to cope with stressful experiences.

In their experiment, the two researchers divided students into two groups. One group watched a video with positive images of African Americans, such as Barack Obama and Martin Luther King, Jr. The other group watched fellow students cheering at a Duke basketball game and game highlights. The videos were intended to “activate” different parts of their identities. After watching the videos, everyone was asked to prepare and give a five-minute speech. At each step during this process, the researchers measured how the study subjects were handling the stress.

As the researchers considered changing their analysis, they searched the scientific literature but found few cases of similar studies being affected by events in the news. One involved researchers from Michigan State University and the University of California at Santa Barbara who were studying people’s beliefs in a “just world” when the attacks occurred on Sept. 11, 2001.

“I thought this was an intriguing opportunity to be able to examine how people were physically stressed during this time, to have an actual indication of how people were responding,” Richman said.

Of course, the new situation was far from ideal from a research standpoint. Because this was just part of what was intended to be a larger experiment, the sample size was small, and included only African-American students. This meant there was no control group with which to compare the results.

Still, it was an unusual opportunity, so Richman and Jonassaint continued the experiment and now began analyzing the data in new ways. They divided the participants into pre- and post-“incident” groups, using April 3, 2006 as the dividing line -- roughly when news about the March 13 party began to dominate the local headlines.

The new analyses confirmed what Jonassaint had noticed: The “post-incident” group had higher levels of cortisol, with very little change during the experiment. While the “pre-incident” group had a baseline reading of 4.6 nanograms per milliliter that jumped to 6.0, the level for the post-incident group hovered higher, at around 7.1.

The effects were most pronounced for women, and students who watched the racial identity video no longer showed results indicating the video “protected” them from the effects of stress.

Jonassaint tested the data against other possible factors -- could it be a result of exams? the time of day they were tested? -- but the results held up.

This was the first study for which Jonassaint had been responsible from start to finish. Now he was seeing the payoff -- only not in the way he had expected. The results of the study, which was funded by the National Institute of Mental Health, will be published his month in the Annals of Behavioral Science.

Richman said she wasn’t surprised by the effect. But she was struck by how strong it was. Even with the relatively small sample, the results were statistically significant.

The researchers say they would have designed the experiment differently if they could have planned for it, such as by adding a control group of white students. But they are satisfied with at least being able to confirm that this highly publicized incident had a significant impact on the physiological responses to stress in African-American students at Duke.

“The idea was just to start out with a pilot study to see if it was feasible,” Jonassaint said. “But now it has turned into something that actually makes a contribution to the scientific community. It was pretty exciting.”