The STL Panel Cochair, Richard A. Merrill, offered a summary and several closing comments. He acknowledged that society is moving toward greater openness and availability of federally funded research data. However, he also expressed skepticism that either the Shelby Amendment or its interpretation by the OMB defines an adequate process for making data publicly accessible.
The larger legal context from which Shelby emerges. The Shelby Amendment is best viewed from a legal perspective beginning in the late 1950s and early 1960s, he said, when only the Food and Drug Administration and a few other agencies issued administrative rules having the force of law; the practice was then uncommon. At the time, regulatory agencies generally approached issues case by case, challenging individual company practices one at a time. They did so through procedures that afforded the defenders of private-sector activity full adjudicatory rights to challenge the admission of evidence and to contest those who proffered the evidence.
Prior to the 1970s, most proposed rules were accompanied by only brief discussions of the underlying factual justification. But soon agencies began to publish bibliographies of the sources they had used as evidence. Federal agencies no longer worked on a case-by-case basis. Regulation in the United States began to change dramatically.
Nova Scotia Food Products v. United States. An important case decided by the Second Circuit Court of Appeals in 1976, Nova Scotia Food Products v. United States established two important and enduring propositions
about agency rule-making procedures.13 First, agencies were obliged to reveal to the public the information they used to propose a rule; i.e., they had to disclose the affirmative case. Second, they were required to respond to the opponent’s affirmative case by explaining why their arguments or the information should not prompt a change in agency position. In Nova Scotia, it was decided that the U.S. FDA had not followed these steps, and so its rules (for smoking fish) were set aside.
In a broad sense, then, the Shelby Amendment is a turn-of-the-century echo of a proposition laid down by the Second Circuit almost 35 years ago. But the Shelby Amendment has the potential to go significantly further. It opens the way for the kind of trial-type process that agencies abandoned some 40 years ago. Unless the scientific community appreciates the significance of this change, said Mr. Merrill, they will fail to understand the long-term implications of this legislative solution to a data access problem.
Shortcomings of Shelby. Although there are alternative instruments for approaching the disclosure problem, they are unlikely to be wholly satisfactory as long as the Shelby Amendment remains in force, noted Mr. Merrill. There are at least three ways in which the Shelby Amendment, though well intentioned, represents a problematic design.
The obligation to collect and release data is potentially serendipitous. Many federally supported researchers will have no idea that the work they are doing may someday be the basis, or part of the basis, of a regulatory action by some agency before which they have never appeared and about which they know little.
There is no “need-to-know” requirement. The mere desire of the requester is sufficient to trigger the obligations of the Shelby Amendment. As a consequence, it does not allow for the kind of nuanced, case-by-case judgment that many speakers at the workshop described as desirable. Under the Shelby Amendment there is no opportunity for balancing the interests of privacy and researcher independence and the interest of public participation in agency rule making.
The Shelby Amendment is not bilateral in its application. The amendment and its interpretation by the OMB seem to apply only to data that are generated with public dollars and that become the basis for regulatory decision making. They do not apply to data that are generated by private dollars that are submitted to support agency decisions.
The lack of direct answers to such fundamental concerns indicates that an issue of great complexity has not yet been resolved.