For Dr. James Whittenberger, the EPA’s 1979 review of its ozone standard should have been a straightforward assignment. A former dean of the Harvard School of Public Health, he was asked to chair the subcommittee that the EPA’s own Science Advisory Board had formed to review the science behind the standard. Instead, the relationship between the subcommittee and the agency soured almost immediately. The EPA’s scientific summary bounced between the two three times, drawing sharp criticism each round, until the agency simply went around its advisers. It leaned instead on a hand-picked panel, headed by a scientist who had defended the existing standard before Congress just weeks before his appointment; a study that none of its advisory committees had previously cited; and a procedure that, stripped of its jargon, amounted to asking nine experts for their best guess and averaging the results. The judge reviewing the final decision took the agency to task for its “undisciplined management of scientific advice.” Worse, by openly shopping around for experts, the EPA had given everyone else license to do the same; its own reading of the evidence was now no more credible than industry’s. In Sheila Jasanoff’s account, the episode lives on in the agency’s institutional memory “as an example of how not to use scientific advice.”
Dr. Whittenberger’s subcommittee belonged to a large and mostly unseen part of government: the more than 1,000 federal advisory committees bound by the Federal Advisory Committee Act to standards of openness and balance. They are not decision-makers, but they wield soft power, and agencies are routinely criticized for ignoring them. The government has other sources of scientific advice, too: paid experts hired by the parties to a dispute, the National Academies, and, until Congress shut it down in 1995, the Office of Technology Assessment. But the advisory committees sit closest to the regulatory agencies actually making the decisions. In other words, they are a very important structural answer to the question of how science should influence policymaking. What should that structure look like?
The first principle is collaboration. An agency that relies on science should have an empowered, independent advisory board with which it does not have an adversarial relationship. After the ozone fiasco, the Science Advisory Board used its own appointment powers in the 1980s to build a board that could talk to the EPA, seating industry representatives, public interest groups, and regulatory consultants alongside academics. Structurally, two features made this possible: the board, not the agency, chose its members, and it balanced them deliberately across viewpoints, so that no one camp could claim to have captured it. The same closeness carried a risk. When the board reviewed the EPA’s own research program in 1987-88, under a panel chaired by a former deputy administrator, its recommendations read suspiciously like the agency’s own policy goals. Hence one more structural guardrail: former senior officials should not chair reviews of their old agency.
A board with the right members and the right relationship to its agency must still decide what to do with the evidence before it. Hence the second principle, reconciliation. When stakeholders present conflicting evidence, the board should reconcile competing constructions of the science rather than act as a judge between adversaries. In the mid-1980s, the EPA’s pesticide advisers did the opposite. Reviewing the fumigant EDB and the apple-ripening chemical Alar, they set themselves up as arbiters of “good” and “bad” science, letting manufacturers pick apart the evidence one study at a time while the agency argued that the studies, taken together, pointed toward regulation. Four years of paralysis on Alar followed, broken not by a scientific panel but by a 1989 60 Minutes segment that set off a national panic. The FDA’s Cardio-Renal Advisory Committee demonstrates the alternative. In 1986, it helped end an eight-year drought in approvals of drugs for irregular heartbeats, largely through the design of its meetings. Before each one, a three-member subcommittee circulated a briefing document identifying the points on which consensus was needed, so that the full committee acted as a peer-review body for its own members rather than a tribunal. Industry testified as a source of practical expertise, not as an opposing litigant, and the FDA arrived with no fixed position. In other words, the way in which which the different actors were asked to show up in the decision-making process heavily influenced the FDA’s ability to make a productive decision.
Yet even a well-run board can be sidelined if its role is drawn too narrowly. The third principle is engagement: the board should stay involved in science-policy questions, not only narrow technical ones, and be clear about the scope of its role. In the first ozone review, questions like what counted as an “adverse health effect” were treated as the EPA’s business alone; shut out of them, the scientists were reduced to fighting over the methods of a single paper. The fix was structural on three counts. By the mid-1980s, the reviewers sat on a standing body, the Clean Air Scientific Advisory Committee, rather than a temporary subcommittee, so they built expertise and relationships from one review to the next. Their mandate let them weigh in on science-policy questions wherever they saw fit. And the EPA’s general counsel guaranteed in writing that “if the Administrator doesn’t agree with you, he has to explain the reasons for his differences.” The committee could not overrule the agency, but it could no longer be ignored. As Dr. James Crapo put it, no one was better placed to judge “what represents an adverse effect and how adverse it is” than the people who had generated the data.
The fourth principle, flexibility, is aimed less at the board than at Congress: scientific judgments should not be written into statute. In the mid-1980s, the FDA convened a panel of government scientists to consider whether animal cancer data could soundly predict human risk for six color additives. The panel largely agreed that the extrapolation was fraught. It did not matter. In Public Citizen v. Young, the D.C. Circuit held that the 1960 Delaney Clause allowed only one reading: an additive found to cause cancer in animals had to be denied. Congress had folded a scientific judgment into law and foreclosed the very question the agency’s scientists had been asked to answer. Structurally, the alternative is for Congress to set the goal and leave the science to a process that can revisit it. The Clean Air Act does just that: it directs the EPA to protect public health “with an adequate margin of safety,” and it requires the standards, along with the science behind them, to be reviewed every five years with the advice of an independent scientific committee.
None of these principles guarantees a good outcome, and each carries its own risks. But together they describe an advisory body built to be both believed and heard. The American public is nearly evenly split on the question: 51 percent want scientists to take an active role in policy debates, while 48 percent would rather they stick to establishing facts. The history of the EPA and FDA suggests that the choice is a false one. Scientists who stay out of policy debates do not become more independent; they simply lose their influence, to the detriment of the decision-maker, the scientific community, and the decision itself.