Never has there been more federal money or pressure for education scientists to partner with practitioners—from the U.S. Department of Education’s Investing in Innovation grants to the Institute of Education Sciences’ push for more classroom-level relevance in research.
At first glance, the two cases don’t appear to have much in common. In the course of a decadelong federal lawsuit over English-language-learner programs in Arizona, lawyers for state schools chief Tom Horne last month subpoenaed the raw data from three studies commissioned by the Civil Rights Project at the University of California, Los Angeles. In gathering the data, researchers from the University of Arizona, in Tucson, and Arizona State University, in Tempe, collected comments on teachers’ and administrators’ experiences with the ELL programs—some of them critical—and promised that the names of teachers, schools, and districts would be kept confidential. Some researchers protested the request, but Judge Raner C. Collins of the U.S. District Court in Tucson ruled on Aug. 19 that the names of schools and districts could be released, but not those of individual study participants.
Next door in California, a Los Angeles Times team obtained from the Los Angeles Unified School District seven years of student test scores linked to unique teacher-identifier numbers. Richard Buddin, a senior economist with the RAND Corp., in his spare time helped the reporting team apply a value-added analysis to gauge the effectiveness of individual teachers based on their students’ test scores. He did not participate in the newspaper’s decision to match the identifiers to actual teacher names, which it released in a series of articles and a matching database beginning last month.
In both cases, the release of data thought to be confidential has sparked firestorms in the research community and with the public at large. The Los Angeles teachers’ union, which is also raising questions about the research methodology the newspaper used, has called for members to boycott the Times. Experts for and against the series continue to weigh in, including U.S. Secretary of Education Arne Duncan, who called for more public sharing of teacher-performance data but also stopped short of urging districts to do it in the same way the Times did. In Arizona, Gary Orfield and Patricia Gándara, the co-directors of the Civil Rights Project, which commissioned the ELL studies, wrote a letter protesting Mr. Horne’s data request, and Mr. Orfield said they have received phone calls from other “appalled” researchers.
“It’s certainly not good news for us in terms of the research environment,” said Martin Orland, the director of evaluation and policy research for WestEd, which conducts studies with school districts nationwide. “The biggest problem is the perception problem. We never have directly identifiable information for students and teachers, … [but] if I was on the other side, no matter what people were saying to me, I would be real careful.”
Working with districts has always been a tightrope walk for researchers, between protecting privacy and disseminating useful—but sometimes controversial—results.
Researchers agree that access—be it to student data, interviews with teachers, or classroom observations—is both the most vital and most difficult part of education studies.
Jane Hannaway, the principal research associate and director of the Washington-based Urban Institute’s Education Policy Center, said researchers “bend over backwards to provide protection” to school sources when collecting data. Yet she and Felice J. Levine, the executive director of the American Educational Research Association, agreed that research participants have become more protective of information.
“Over the last decade, increasing awareness that one’s data is available in a number of places and can be used by advertisers [and] can be accessible to employers … creates more of an inhibition about where one puts information and could be a similar inhibition to whether one participates in research,” Ms. Levine said.
Federal privacy law covers neither situation. The Code of Federal Regulations for the Protection of Human Subjects delegates confidentiality decisions to university institutional review boards, or IRBs, but in Arizona, the IRBs released the full data over the researchers’ opposition.
“It really calls into question not just the access to schools but the integrity of the IRB process,” Mr. Orfield said. Mr. Orfield, Ms. Hannaway, and other researchers suggested researchers may need a federal shield law similar to state laws that protect reporters from being compelled to name sources. “We thought the IRBs served that purpose for us, but we were wrong,” Mr. Orfield said.
Moreover, Ms. Levine said that protection procedures for researchers have not always kept pace with the rapidly expanding statewide longitudinal-data systems and federal education reporting systems, which should make data public, similar to what happened in Los Angeles.
Indeed, John Q. Easton, the director of the IES, the federal education department’s main research arm, said the institute’s national centers on education statistics and research perform intensive tests on their databases to ensure outsiders cannot backtrack through the information to find individuals. Recently, Marilyn Seastrom, the chief statistician for the IES, conducted the same tests on the data reported on state websites—and occasionally found less-than-secure information.
“She worked like a hacker, like someone would to figure out if there was a way to get things out of the data,” Mr. Easton recalled. “I think there’s acute attention paid to [confidentiality] because of the prevalence and the huge volume of data that has become available and is consequential to people.”
Separately, both the IES and the AERA are working on new guidance for researchers on ways to protect confidentiality when using state data systems and information provided through accountability reporting.
Still, some researchers argue the next generation of confidentiality guidance should make data more public, not less.
‘Era of Transparency’
“When we went down the road for human-subject protection, the object of this was to protect human subjects from bad stuff—protect their privacy, protect their informed consent—and for me, a lot of that stuff has just gone way too far,” said Mark S. Schneider, a former commissioner of the Education Department’s National Center for Education Statistics.
The Family Educational Rights and Privacy Act, or FERPA, protects data for students, but not teachers or administrators, and Mr. Schneider argued the latter two groups should not be given the same deference as students when it comes to offering confidentiality.
“I think that we want to be in an era with greater transparency,” said Mr. Schneider, now a vice president with the Washington-based American Institutes for Research. “I still believe if you are a public employee delivering a public service, maybe the expectations regarding privacy in your performance domain is a totally different question than your right to get my records from CVS [on health care]. ”
Mr. Orfield and Ms. Levine disagree. They argue that private-sector employees don’t face the same sort of public scrutiny now given to Los Angeles teachers, and while a school district superintendent may count as a public official, a teacher does not.
“I think it would really have a crippling effect on all social science, education, and health inquiry if public employees in the sector couldn’t be guaranteed the same confidentiality as any other research participant,” Ms. Levine said. She added, “In this economy, people are feeling pressed in a number of ways, and being a participant in a voluntary study is probably lower on one’s list of priorities than is providing for oneself and one’s children.”
Eric A. Hanushek, a senior fellow at the Stanford University-based Hoover Institution, said that the high-profile data releases will “almost certainly” make it harder for researchers to get information from teachers and schools. “On the other hand, I think it also demonstrates that more discussions ought to be had about how this information should be made more public. [Teachers’] performance will be subject to more scrutiny, but ultimately, I believe parents should have more access to the data.”
Secretary Duncan seems to agree with Mr. Hanushek. In a speech Aug. 25, he defended the Times’ decision to release teacher names.
“I am a strong advocate for transparency,” Mr. Duncan said, arguing that if he had the power, he would make public not only data tying teachers to student performance but student-attendance data to “hold parents accountable.”
“Let’s put out data on dropouts, college enrollment, college completion, loan-default rates, and every other kind of data that can help us highlight our remarkable success and help us better understand why too many of our children are unprepared,” Mr. Duncan said. “The truth is always hard to swallow, but it can only make us better, stronger, and smarter.”
Yet Mr. Duncan said districts and teachers should work together to find an official way to release sensitive data, rather than having it released as the Times did. He also acknowledged concerns voiced by many researchers that the biggest privacy problems may come from education decisions made by parents and community members “with a limited understanding of what this information really means.”
Mr. Schneider said as research becomes a bigger part of public education debates and more database information becomes available to the public, researchers have a greater responsibility to educate the public.
“If I’m doing something that five scholars care about, fine; I can talk to them [individually],” Mr. Schneider said. “If I’m doing something that affects however many school kids are in LAUSD, … then I believe the researcher has a responsibility not only to talk to the five other researchers who understand how beautiful this model is, but the hundreds of thousands of parents in LAUSD who might be affected by these findings.”
A version of this article appeared in the September 15, 2010 edition of Education Week as Scholars Worry Conflicts Over Data Could Hamstring Future Research