If you are conducting a faculty search, or trying to diversify the professoriate, or want to see whether various programs to do so have succeeded, the Survey of Earned Doctorates has always been a key source of information. They survey will tell you, for example, how many Latinos earned doctorates in chemistry (23 for the last year available), or how many black people earned doctorates in political science (34). If you watch the trends from year to year, and also pay attention to the total number of doctorates awarded (1,170 in chemistry to U.S. citizens, and 506 in political science), you have an instant sense of the changing or stagnant demographics of your pool.

Or at least you used to.

Citing privacy concerns, the National Science Foundation -- which sponsors the survey -- has ordered that data on subgroups beneath a certain size be blocked from release. So subgroups for which the numbers are small will no longer be available. So while we know that in 2005, six black people earned doctorates in earth, atmospheric and marine sciences, the NSF won't reveal how many earned the degrees in 2006 (covered by the most recent report). Information about the number of Latinos earning degrees in some engineering fields is gone, as are data about a number of categories for black Ph.D.'s. For Native Americans, where the base is smaller, the impact of the new policy is especially dramatic. The report was stripped of information on how many doctorates were awarded to all but 6 of the 35 subfields for which data were collected.

Because most people who focus on the study are drawn to the overall trends, where data about various minority groups is preserved because of the larger sample sizes, the issue of the missing information is only now starting to receive attention. But advocates for increased diversity in graduate education and the professoriate are frustrated by the changes. They note that educational experts of many political perspectives agree that it's hard to know how to tackle educational challenges without information about the performance of subgroups -- that's even one of the principles underpinning President Bush's favorite education law, No Child Left Behind. So removing this information, advocates say, makes no sense. They add that debates about public policy would be informed by seeing these numbers in detail -- and that the fact that the numbers are small is part of why they are important to consider.

"This hides information. It removes information," said Andreen Neukranz-Butler, human rights compliance officer for the University of Idaho and a member of the board of the American Association for Affirmative Action. If a subgroup goes from two to four doctorates a year (or falls similarly), that's important information, she said, and those working on these issues need to know it.

The Survey of Earned Doctorates is conducted for the NSF by the National Opinion Research Center at the University of Chicago. But officials of the center said that they could not talk about the changes, and referred all questions on the change to the NSF, which sponsors the research.

Jaqui C. Falkenheim, the NSF project manager for the survey, confirmed the new rules via e-mail. She said that the agency had decided to block the release of the information about subgroups with small sizes based on a review by the Division of Science Resources Statistics of its procedures to protect confidentiality. "The findings of that review revealed the need, given more restrictive rules/guidelines and heightened concerns about confidentiality/privacy, to tighten SRS procedures for releasing SED data to the public," she said.

She added that NSF "staff and contractors must comply with the laws governing the protection of information provided by respondents.... Any breach of confidentiality could substantially harm SRS's ability to conduct its surveys and obtain the cooperation of respondents in the future. Protection of respondents' confidentiality is crucial to maintaining our surveys' high response rates, our reputation as a federal statistical agency, and the reputation of NSF as a research partner with academe."

While she confirmed that the NSF "will be suppressing more small cells before releasing SED data than has been done in the past," she said that the NSF was "committed to meeting the needs of our data users to the maximum extent possible, given the constraints that apply."

Asked how large a cell needs to be before the NSF will release it, Falkenheim said that that information was covered by the new confidentiality rules as well. In some tables, it appears that the NSF is blocking cells that are smaller than six, and the number six does appear. But in other places, it appears that the cell size being blocked may be a little larger. For example, in the physical sciences, the NSF permitted the release of total data for black Ph.D. recipients (73) and the subgroups of chemistry (28), computer and information sciences (14) and mathematics (16). But there are two subgroups for which data have been blocked (earth and atmospheric sciences, and physics and astronomy) and to reach the total of 73 at least one of those categories must have more than five people in it.

Neukranz-Butler and several others familiar with the survey said that they were surprised by the concerns about confidentiality. While the survey has contained small numbers in some cells up until now, it has never named the individuals or institutions. So there has never been information on a particular person covered by any of the categories.

"The report never told you who the people were," said Neukranz-Butler. "Why are we being hampered on getting very important information?"