Here's your chance to have an impact on science in the twenty-first century. The government wants your thoughts on research misconduct. December 13 is the deadline to submit comments on the pending federal policy on this most touchy of topics. The final policy is due to be implemented early next year. On November 17 of this year, the National Academy of Sciences (NAS) hosted what was billed as a "town meeting" in Washington, D.C. (broadcast simultaneously on the Internet) to discuss the policy as it now stands.

The draft policy (whose text and fact sheet are available online) has been in an elongated pipeline since 1996. It is the product of a tortuous process, coordinated by the White House Office of Science and Technology Policy (OSTP), involving all of the federal agencies that dispense money to researchers. The policy has emerged from more than two decades of mostly bitter struggle with the distasteful realities of scientific chicanery. It will apply to all federally funded research including, for the first time, intramural research.

The draft says that research misconduct consists of what veterans call FFP: fabrication (making up results), falsification (manipulating, changing, or omitting data in order to misrepresent the results), and plagiarism (appropriating another's work without credit). For the first time, the definition explicitly includes an increasingly frequent and disturbing practice: plagiarism of ideas growing out of the peer review process in both funding agencies and at journals.

The policy also sets guidelines for handling misconduct allegations, including safeguards for accuser and accused. As OSTP's Sybil Francis pointed out, the highest-profile cases of the past turned into long national nightmares not because of problems in defining misconduct, but because procedures for handling allegations of misconduct were inadequate (or, in many cases, nonexistent). The policy's most significant procedural provision settles primary responsibility for preventing and detecting misconduct, and for handling misconduct cases, squarely on grantee institutions. This broad authority means that institutions are permitted - indeed, are encouraged - to establish their own standards and practices.

What the policy does not cover is disputes over authorship, unless they involve plagiarism. It also does not address most cases involving the treatment of human or animal research subjects. Nor does it cover misconduct in nonresearch activities - education, for example - even if funded by a research agency.

After many revisions, the draft policy has had an essentially friendly reception from the science establishment. At the NAS town meeting, compliments such as "fair," "workable," and "a good plan" were in the air. Nearly all the commentators were officials from federal agencies or academic institutions or professional associations (i.e., lobbyists.) The organizers wanted to encourage outside participation with the simultaneous Webcast of the meeting, and invited questions via email. But there were only a couple of e-questions, indicating a small audience for this experiment in extending NAS policy discussions beyond the Beltway.

The harshest critique came from C.K. Gunsalus, associate provost at the University of Illinois in Champaign. She acknowledged that there was much to admire in the draft, such as a long-awaited uniform federal policy across funding agencies, and a separation of the investigation and adjudication phases of a misconduct procedure. But she was pessimistic that the policy would work.

The requirement that, to be deemed misconduct, research activities must entail "a significant departure from accepted practices of the scientific community" struck her as lowest-common-denominator ethics. The excuse "everybody does it," she said, doesn't make a shabby practice ethical.

Gunsalus stressed that the policy will apply to a range of institutions, some of which lack experience in misconduct cases and have no policies of their own. They are likely to adopt the federal standards as their own. But institutional standards, she argued, should be higher than that.

Gunsalus also noted that the policy contained no mechanism for assuring that all federal agencies actually adopt it. She fears that if every agency follows its own implementation plans, each may adopt its own reporting and oversight requirements. That situation, she said, might well be more of a burden for university administrators than the present one, where different agencies observe different definitions of misconduct.

While the draft speaks of providing protection for whistle-blowers, Gunsalus said, it does not specify what the safeguards should be, nor how to deal with retaliation. On the other hand, Stephanie Bird, editor of Science and Engineering Ethics, pointed out that allegations of misconduct can devastate careers and lives even if the accused is exonerated. Bird argued for writing specific procedures into the policy to protect confidentiality.

The policy was also faulted for ignoring the treatment of research subjects. The justification for this and some other exclusions is that policies already exist to address them. David Korn, of the Association of American Medical Colleges, noted that the National Insitutes of Health (NIH)'s human subjects overseer, the Office of Protection from Research Risks, is about to get a promotion in status by moving into the office of the Secretary of Health and Human Services. Once the new office is up and running, he said, it is expected that it will address human-subjects issues more systematically.

Some commentators questioned whether the meaning of FFP was completely clear. "A lot of us don't think so," said John Gardenier, a statistician with the Centers for Disease Control and Prevention.

Falsification and fabrication can be interpreted as applying only to data, he argued, when it should also include selection of the analytic method. Scientists josh about picking a method that will yield the desired p value, he said, but manipulation of the method is no joke. The American Statistical Association has recently embraced a set of ethical guidelines for statistical practices. Gardenier argued that they should be adopted by everyone who uses statistics on the job.

Will the new policy be of significant help in dealing with research misconduct? The science establishment hopes so, because even rare instances of misconduct shatter public confidence in science. The consensus is that scientists are mostly honest and misconduct pretty uncommon. The National Science Foundation reports, for example, that it averages only three findings of misconduct per year, .01% of total projects.

Misbehavior of the types covered by the proposed policy may indeed be rare. But there are disturbing hints that other, more slippery forms of misconduct may be on the increase.

Floyd Bloom, the neuroscientist and departing editor-in-chief of Science, recounted a few disquieting anecdotes from his five-year tenure there, calling them "counterproductive scientific behaviors that seem not to be covered within the current simpler and almost universally acceptable definitions of scientific misconduct." Most of these might constitute plagiarism if only they weren't so hard to nail down. For example:

• Scientist A presents preliminary work at a colloquium attended by scientist B. B is highly skeptical but offhandedly suggests an exploration of the topic to his new postdoc, who is searching for a project. The postdoc achieves a breakthough, and he and B publish. A feels his ideas have been stolen. Have they?

• Scientist C and his postdocs make an important chemical discovery and submit a paper to a chemical journal. While waiting for the review, they refine the process so that it has broader implications. They submit a new paper to a high-profile general journal, which shall remain nameless. Both papers are accepted; neither mentions the other. When they are published, editors of both journals are furious and bar the authors from future submissions. Is this self-plagiarism and/or duplicate publication?

• A small group of scientists clone and isolate an important enzyme and submit a paper to the aforementioned unnamed high-profile journal. Among the reviewers is a senior scientist formerly active in that same research area; he tells the journal he has no conflict of interest. His review is very late, and he requires additional experiments and asks to see the revision. The authors claim that the extra experiments are inappropriate; the editor agrees and publishes. A week later a competing journal publishes the same finding from the reviewer's lab. The timing certainly invites scrutiny, but how to demonstrate that plagiarism actually occurred?

"Clever scientists are also very clever manipulators of the system," Bloom observed. "It is our observation that authorship and collaboration problems are on the rise and are a serious threat to the research enterprise."

Comments on the draft misconduct policy should be sent to: Sybil Francis, OSTP, Executive Office of the President, Washington, D.C. 20502; email sfrancis@ostp.eop.gov, fax (202) 456-6027. Comments must be dated by December 13, 1999.

Tell us what you think.

Previous Adapt or Die Articles