TCS Daily

Fraudulent Witch-Hunts

By Glenn Harlan Reynolds - October 2, 2002 12:00 AM

Research fraud is in the news again. At Bell Labs, researcher Hendrik Schon has been fired for allegedly faking data in numerous research papers, many of which were published in highly prestigious scientific journals. At Emory University, historian Michael Bellesiles is appealing the findings of an independent commission that was assigned to investigate charges that his Bancroft-Prize-winning book Arming America contained fraudulent or fabricated data; many observers expect him to be
fired, too.

Other academics have been in trouble over the past year, too: Stephen Ambrose and Doris Kearns Goodwin were the subjects of plagiarism scandals, and Joseph Ellis was suspended from teaching at Mount Holyoke for a year for peddling fictionalized accounts of his military service in Vietnam. (Unlike the others, Ellis's scholarly work was not implicated.)

The Schon and Bellesiles cases - because of the subject matter, their seriousness, and because, in Bellesiles' case, federal grant money was involved - are likely to create the biggest impression. In fact, if we're unlucky, we could face a rehash of the disastrous science-fraud witch-hunts of the 1980s and 1990s.

Those witch-hunts - and that is not too strong a word - took place in response to earlier fraud scandals, like the Summerlin mouse case - in which dramatic skin-transplant results from black to white mice were actually accomplished by inking in black patches on the fur of white mice with a felt-tip pen - that led members of Congress to conclude that academia wasn't up to the task of policing itself, or, less charitably, that there was political hay to be made by attacking science fraud.

There is no dispute, of course, that academia, like other human endeavors, has always had its share of fraud. The Piltdown Man - a cobbled-together phony fossil that may have started out as a practical joke - is one example; the work of Boston University cancer researcher Marc Straus, whose data looked too good to be true because they were fake, is another. And since, as scientist Thomas Ray has noted, every successful system accumulates parasites, it's no surprise that a system as successful as American academia has picked up some intellectual tapeworms along the way.

But - as with tapeworms - it is important to avoid cures that are worse than the disease. Congress failed to heed that advice the last time around. Hearings led by then-Congressman Albert Gore, Jr. led to an increased focus on fraud in the sciences. Rep. John Dingell's investigations subcommittee took the matter further, and NIH was pressured to set up an office dedicated to "research integrity."

Unfortunately, both anti-fraud operations turned out to operate along witch-hunt lines, processing anonymous allegations into bogus accusations. Nobel Prize winner David Baltimore was dragged through the mud for years in response to ungrounded allegations made by an unhappy graduate student against one of his proteges (during the investigations NIH officials, while investigating fraud, leaked false information to the press, and even faked the results of forensic investigations in order to make their case look stronger than it was); AIDS researcher Robert Gallo was investigated for, (as journalist Malcolm Gladwell put it), the unapproved use of compound sentence structures; and researcher Rameshwar Sharma lost his job to an investigation revolving around an obviously innocent typo in a grant application that was, in any case, denied. All of these people - along with many others - were ultimately exonerated as the cases against them collapsed upon examination.

These experiences led to an increased interest in due process, something that scientists sometimes see as a barrier to the truth. As Bernadine Healy put it, "I came full circle to thinking that an adversarial system was necessary. It had become obvious that this was a totally polluted system where these scientists got behind closed doors and worked out their venom, taking down their colleagues. It was a Star Chamber, a hideous travesty of justice."

It was indeed. The travesty resulted from two things: the belief by scientists (and scientific bureaucrats) that there had to be very visible and immediate action against perceived fraud or their funding would be jeopardized, and the belief that due-process protections were merely technical tools for lawyers trying to avoid a good-faith search for truth. By the time it was all over, though, the investigators were themselves guilty of what Gladwell called "the fraud fraud," concluding that, "up close the alleged sins of scientists appear minor," and scientists recognized that such things as impartial hearings and the right of accused persons to examine evidence against them can play a vital role in promoting truth - and justice. (You can read a law review article in which I discuss this issue at more length here.)

This experience suggests that any response to these new scandals needs to be more measured - and less prone to hysteria, manipulation, and political grandstanding - than last time around. The conventional wisdom, after all, is that the academic process of replication and scholarly criticism will discover fraud. Though the discoveries may have come a bit slowly in the Schon and Bellesiles cases, they did come along - and they did so without the injustices that accompanied the more bureaucratic approach. When you create a new bureaucracy to deal with justice, the result tends to be rather more bureaucratic than just. With luck, the opportunities for political grandstanding provided by the war and the stock market's decline will keep the issue of academic fraud from being overly politicized this time around.

If there is a lesson - from the Schon and Bellesiles affairs in particular - it is that peer review is overrated as a check on academic fraud. Peer review is hard work, but it is under-rewarded in academia, meaning that people tend not to put a lot of time into it. In addition, peer reviewers are often influenced by the stature of the reviewee and by the excitingness, or political correctness, of the findings involved. So while peer review is reasonably good at discovering obvious methodological problems and outright crack-pottedness, it is of very limited use in spotting actual fraud. In fraud cases the data will always look good.

When fraud is discovered, it is usually by another researcher whose skepticism is aroused. Yet uncovering fraud usually isn't considered as valuable to an academic career as original research is; worse yet, some scholars who expose their colleagues as frauds face resentment from those who dislike seeing their field's dirty laundry aired. But the absence of consequences for fraud can only make the problem worse. If we want to discourage fraud, we need to ensure that the people who discover it are recognized for their contributions - which, after all, spare other members of the field years or even decades of wasted effort based on fraudulent work - and properly rewarded. And, of course, we need to ensure that those who commit fraud are properly punished. For although anyone can make mistakes, conscious fraud is a betrayal of academia's most central principles and deserves to be treated as such. If the administrators of academia want to ensure that their fields are held in high regard by the public, they should bear that in mind.



TCS Daily Archives