t the heart of almost every argument for affirmative action in university admissions, including the one made by the University of Michigan in its current case before the Supreme Court, lies an assertion that racial tolerance and the educational experience itself improve with a diverse student body.

Yet that may not be the case in the eyes of students, faculty members and administrators. A new study has found that diversity does not improve the perception of educational quality or ease racial tensions on college campuses. In fact, it may aggravate them.

"Surveys are always questionable, including ours," said Stanley Rothman, an author of the study who is a professor emeritus at Smith College. "But it shows that those who argue diversity will improve the education of everybody haven't made their case. The data does not support them."

The study, published in The International Journal of Public Opinion Research, a peer reviewed journal, and The Public Interest, a conservative-leaning magazine, was financed by supporters of legal challenges to affirmative action and has not yet found its way into the public debate on the subject.

But it stands in stark contrast to earlier studies supporting the educational benefits of diversity, and poses a direct challenge to the academic research the University of Michigan is relying on to defend the use of racial preferences in admissions.

"A racially and ethnically diverse university student body has far-ranging and significant benefits for all students, nonminorities and minorities alike," wrote Patricia Gurin, the Michigan professor of psychology whose research figures prominently in the university's legal arguments. "Students learn better in a diverse educational environment, and they are better prepared to become active participants in our pluralistic, democratic society once they leave such a setting."

While Mr. Rothman's study did not concern itself with the latter point — that diverse colleges produce more civic-minded adults — it evaluated the question of educational quality by asking about 4,000 students, faculty members and administrators on about 140 campuses what they thought about the instruction offered at their universities. They also asked what they thought about the students around them, and whether discrimination existed in their college settings.

If, as proponents contend, diversity benefits the educational experience and promotes racial understanding, then as minority representation increases, so, too, should one's satisfaction with the college, esteem for fellow students and sense of freedom from discrimination, Mr. Rothman and his colleagues argued.

But the survey found that racial diversity produced none of those benefits. As the proportion of black students on campus increased, the regard for the educational experience declined, the opinion of student academic preparedness and work ethic fell and claims of discrimination became more common.

Given that the survey did not ask participants to explain why they felt as they did, Mr. Rothman was reluctant to speculate on the reasons behind their opinions. Indeed, the explanations could include old-fashioned racism and the perception that affirmative action had relaxed admissions standards, leading to an underprepared student body and a less challenging curriculum.

But Mr. Rothman emphasized that the data was not statistically compelling enough to prove that racial diversity actually harmed the educational experience, only that there was no evidence to show that it improved matters.

Financial support for the research came from the Earhart Foundation and the Randolph Foundation, two small institutions that have supported conservative research organizations and scholars. While the Earhart foundation acknowledged its reputation for doing so, it emphasized that it only supported "advanced academic study," not polemics.

The Randolph Foundation, by contrast, would not talk about what it supports and why. But tax records show that it has financed several organizations dedicated to eliminating affirmative action, including a $145,000 gift in 1999 to the Center for Individual Rights, the group that helped orchestrate the legal challenge to Michigan's admissions policies now before the Supreme Court.

Nonetheless, the study's authors have earned respect from academics on all sides of the affirmative action debate. Seymour Martin Lipset, another of its authors, is the former leader of the American Sociological Association and of the American Political Science Association, two posts that have earned him a sterling scholastic reputation.

That did not insulate the study from criticism, though. Just as Professor Gurin's findings have attracted critics, so have Mr. Rothman's.

Gary Orfield, a Harvard education professor who has conducted similar research and whose method was criticized by Mr. Rothman, called the study "a set of weak correlations" with a "completely inadequate response rate" that flies in the face of "generations of desegregation research."

Lani Guinier, a Harvard law professor, also took the study's method to task. In particular, she questioned whether the researchers had incorrectly emphasized race in the study, when their own findings showed that the selectivity of an institution had more of an impact on attitudes about educational quality.

Still, she agreed that the persistence of racial discrimination, even on the most integrated campuses, seemed plausible. "They may be onto something," Ms. Guinier said. "Universities cannot simply add more students of color and expect miracles to happen."