Problem to be addressed:
The use of longitudinal student data, especially those about achievement generated by NCLB, could provide a wealth of information to districts about what is and is not working in their attempts to improve student outcomes. Because of increasing concerns about the interpretation of the Family Educational Rights and Privacy Act (FERPA)1, however, many states are afraid to provide data to researchers. This includes states that previously released data for research purposes (e.g., Texas and New York) and a variety of others that have the capacity now but are afraid to proceed.
FERPA is designed to protect individual student privacy and, as a general principle, requires that individuals sign consent forms before any data about them is released by educational institutions in an identifiable way. (Data that cannot be individually identified can of course be used). There are, however, special provisions that permit identifiable data to be used by school systems, states, and the federal Department to conduct analyses that will help to improve the schools.
Perhaps because of the recent increased availability of data on student performance and on school operations, there has been new attention to ensuring the confidentiality of student records. It is essential, however, that this attention to confidentiality issues not preclude valid research and evaluation activities that are conducted within the requirements of FERPA. Because of the Department of Education's role in developing regulations for the implementation of FERPA, the Department has a crucial part in achieving these goals.
The most reliable analyses employ longitudinal information on individual student experiences, but only if these analyses employ modern panel data methods—ones that uniformly require linking individuals over time. (Note, however, that linking individuals is not the same as identifying the individuals). Attempts to make data available for public use (de-identified data in FERPA) generally destroy their research value. For example, approaches such as suppressing data with "small cell sizes" or adding some random element to some of the data destroy the possibility of conducting the most reliable research studies.
NCLB requires states to develop databases about student achievement, and IES under Congressional direction has provided funds and encouragement for states to develop longitudinal student databases. These activities make an enormous amount of sense from a public policy view of attempting to learn about the causes of student achievement. But, they are rendered useless if the data cannot be used effectively to learn about the impacts of educational policies and other factors on student performance.
The Department of Education, following IES leadership, should work to promote data access for qualified researchers. First, the Secretary should clearly articulate the need for states to provide researcher access to data so that new knowledge can be generated. Second, the Secretary should provide guidelines that ensure protection for confidential records while still allowing individual longitudinal data to be used by independent researchers. For example, providing for third party encryption of identifying information (such as SSNs, names, or addresses) would maintain privacy while permitting access to longitudinal data to bona fide researchers. Third, the Department should make it clear that protection of the data and not restrictions of the analysis and researchers is the primary goal. It is important that independent researchers be permitted to conduct scientific research and evaluations as long as they can demonstrate that they protect the confidentiality of the data and that their research satisfies the FERPA requirements.