Fraud: Why Auditing Laboratory Records Is A Bad Idea

Fraud in research has recently aroused the interest of Congress because it wastes tax dollars. But scientists share this concern for even stronger reasons: We build our whole enterprise on a foundation of communal trust. Since the record of some of our academic instituiton has been a sorry one, we should be pleased that public criticism is leading to improved procedures. If the matter had stopped there, the outcome would have been very salutary. Unfortunately, the reaction has gone much farther. Though it is by no means clear that fraud has increased substantially in recent years, Congress has become convinced that the problem has grown rapidly and requires forceful action. Moreover, the vindictive pursuit of David Baltimore by Rep. John Dingell (D-Mich), for a case involving normal error rather than fraud, has inflamed the discussion. In this atmosphere, Congress has suggested that we need an Office of Scientific Integrity, and that it should audit laboratory records on a random basis. Though this idea initially seemed wild to most scientists, some administrators and editors have suggested that audits are inevitable and that the scientific community should initiate them. The advantages would be forestalling an invasion by government inspectors, and replacing mere impressions of the magnitude of the problem by hard facts. However, there are good reasons for strenuously resisting this solution. In the first place, auditing would be extremely ineffective. Unlike financial accounts, where the flow of money can be traced, scientific records can be easily altered. Scientists can insert virtually any data they wish into research notebooks. A researcher who anticipates auditing can thus hide fabrication of results. The main methods for detecting fraud will therefore continue to be the same ones that have been prominent in the past--direct observation by colleagues in the same laboratory, or the discovery that published observations cannot be verified. Second, even if auditing were reliable, how useful would the information be? Some low level of fraud is inevitable, and it has not seriously impaired the advance of science in the past. Hence the most important question is not whether the frequency is 1% or 2% of all research; it is whether the harm to the scientific enterprise has now become great enough to justify major initiatives that might themselves damage that enterprise. The spectacular current advance of science suggests that we face no such emergency. Third, if the purpose of auditing is to prevent waste of tax money, legislators should compare the potential saving with the expense. Since fraud is undoubtedly uncommon relative to the whole scientific enterprise, and since auditing is an inefficient means of detecting it, this approach does not promise to be cost-effective. Fourth, a bureaucracy built to monitor fraud will be tempted to extend its range--especially since legal counsel at the NIH has insisted on the broad term misconduct instead of the technical legal term fraud. Misconduct includes misappropriation of ideas (for example, from grant applications or refereed manuscripts) and inadequate credit to predecessors. Such problems are more frequent than fraud and of wider concern to scientists. But even if the present informal ways of handling them within the scientific community do not always yield a satisfactory outcome, blunt dissection by an Office of Scientific Integrity is hardly likely to be better. Fifth, the self-regulation of scientists has fostered creativity and enthusiasm and has therefore enhanced contributions to productivity and to culture. The threat of random search, and even more the threat of regarding any honest error with suspicion, will inevitably encourage less bold, pedestrian research. Sixth, by taking away some of the fun from research, the proposed measures would impair the recruitment of bright young people into science. This effect may be hard to recognize until a great deal of harm has been done. Finally, a climate of suspicion encourages institutions to overreact. One career has already been destroyed by Harvard's failure to distinguish between two kinds of plagiarism--the misdemeanor of incorporating material in a review without rewording it, and the felony of falsely claiming credit for a discovery. Fraud in science is an old story. Given the range of human nature, a low level is inevitable. But it has probably been less frequent here than in most human enterprises--not because scientists are particularly selected for integrity--but because nature is our ultimate auditor. Even in clinical research, where the problem seems to be the worst, fraud has not been serious enough to prevent a spectacular rate of advance. Scientists surely all wish to see fraud attacked vigorously. Nevertheless, we must help legislators and the public to recognize that auditing, and the prospect of heavy-handed public excoriation for normal error, are likely to do more harm than the disease they aim to cure. Bernard D. Davis is a microbiologist in the Bacterial Physiology Unit at Harvard Medical School.

Fraud: Why Auditing Laboratory Records Is A Bad Idea

Interested in reading more?

Become a Member of