Reports in the popular press,(1) as well as recent studies by scientific organizations,(2) make clear that the problem of misconduct in science(3) is real and persistent. Scientific integrity is not always ensured by science's self-correcting mechanisms.(4) The effects of misconduct go beyond distortion of scientific data. Misconduct can also undermine the proper and fair allocation of federal resources when it involves peer review of grant proposals, a process that depends on honesty and integrity. The public faith in science that is necessary to ensure sufficient funding for the progress of science depends on honest applications for federal funding and honest use of public funds. Therefore, both the government and the academic community need effective procedures for dealing with misconduct in science.
Despite the need for effective procedures for establishing whether misconduct in science has occurred, considerable controversy has arisen over which procedures should be used. Most universities have only recently put in place procedures for dealing with misconduct in science and have not yet fully tested them.(5) According to an American Association for the Advancement of Science survey, scientists lack confidence in university handling of misconduct cases.(6) This article presents our study of the principles underlying NSF's procedures for handling misconduct in science, as well as the results of NSF's experience in handling allegations of misconduct. By presenting the rationale for NSF's procedures, we hope not only to persuade the scientific community of the applicability of these procedures to misconduct in science cases but also to stimulate comment and criticism.
The "Scientific Dialogue" Model
Many early cases of misconduct in science have been handled according to procedures developed at the Public Health Service (PHS) and referred to as the "scientific dialogue" model.(7) This model emulates the peer review process by assembling a panel of scientists who investigate allegations and reach a consensus on whether misconduct has occurred.(8) Although the familiarity of the procedures initially appealed to some scientists, the scientific community itself has criticized handling of cases under the scientific dialogue model for not meeting standards of confidentiality and fairness.(9) Applying methods developed for the evaluation of manuscripts, grant proposals, or tenure qualifications to the determination of ethically acceptable conduct leaves unfulfilled expectations. When proceedings may lead to sanctions, both accused scientists and observers expect certain protections. Nonetheless, some scientists resisted application of traditional American investigative and adjudicative procedures, which offer such protections, to misconduct in science cases. Perhaps this was due to concerns that scientific issues would get lost in such a setting.(10) However, theory and practice in the handling of misconduct cases has brought out the importance of procedural fairness. PHS has shown an increased awareness of procedural protections in handling misconduct in science cases by introducing the opportunity for a hearing.(11)
NSF's Model for Handling Misconduct in Science Cases
Since 1987 NSF has based its policies and procedures for handling misconduct in science cases on well-established administrative, civil, and criminal methods of investigation and adjudication.(12) NSF believes that traditional investigative and adjudicative procedures are appropriate when the issue is culpability for misconduct. These traditional procedures are preferable to alternatives adapted from the process of scientific peer review. Peer reviewers necessarily assume that authors are truthful. Because the central issue to be resolved in a misconduct case is whether an individual has been truthful, a system like peer review is inappropriate because it lacks the investigative, adjudicatory, and due process mechanisms to evaluate that issue accurately and fairly. …