Odd Ban on Minority Ph.D. Stats Lifted

Jeff tries to explain how government works to readers of Science.

The National Science Foundation (NSF) on Friday reversed a controversial policy that reduced the amount of information available on new minority Ph.D.s. The move follows widespread complaints that the new rules, applied last year for the first time, would hinder efforts to attract more minority students into scientific fields.

Each year, NSF conducts a Survey of Earned Doctorates, asking newly minted Ph.D.s to provide a wealth of information on their educational history and career plans. The results can be broken down by field and by race, ethnicity, and gender (REG). In 2007, citing new federal privacy rules, NSF's statistical branch decided to suppress a considerable amount of information about underrepresented minorities (in particular, African-Americans, Hispanics, and Native Americans). Staffers feared that savvy data miners could make use of the small numbers reported in some subcategories—one Hispanic received a doctoral degree in astronomy in 2006, for example—to identify individuals.

The changes meant that any subcategory with fewer than six degree recipients went unreported. In practice, however, many more categories were also blanked out because NSF was concerned that the missing numbers could be calculated by a process of elimination. NSF also banned the use of zero, arguing that even a null set conveyed information—the absence of minorities in that category—that potentially compromised NSF's promise of anonymity to participants.

News of the new rules took several months to trickle down to researchers, institutions, and professional societies that use the data, including organizations running projects funded by NSF aimed at fostering broader participation in science and engineering. By last spring, however, they were both bewildered and outraged. There were even rumors that sinister motives were at work. "Without evidence of underrepresentation, some people might wonder whether such programs are needed," notes Shirley McBay, president of the Quality Education for Minorities Network.

Taken aback by the vociferous criticism, NSF asked McBay's group to hold a series of meetings that gave the community a chance to vent its anger and to suggest alternatives. (NSF had offered three options, which were universally panned.) Last week, McBay reported the results of those meetings to NSF's Committee on Equal Opportunities in Science and Engineering, an advisory body for issues affecting underrepresented minorities. Committee members also voiced their unhappiness with the changes and questioned why they were necessary.

That's when NSF announced it had had a change of heart. Lynda Carlson, head of NSF's Science Resources Statistics division, followed McBay to the podium and shocked her audience by declaring that NSF was rescinding almost all the new policy. From now on, only fields that award fewer than 25 total doctorates each year will be subject to any data suppression. That is likely to affect about 4% of the 280 subfields reported in the survey, says Mary Frase of the stats division. (In those instances, subfields will be combined until the minimum is reached.) In all other cases, results will be reported by race, ethnicity, and gender, even if the result is zero in some categories.

"We listened," Carlson explained after the meeting. "We didn't realize the extent to which people are using the REG tables. We can't do everything the community wanted. But we've tried to meet as many of their needs as possible." McBay says that she's pleased the statistics division "heard the concerns expressed … and has reconsidered its approach."

Posted in Education