As a story in today’s ScienceNow [subscription required] by the indefatigable Jennifer Couzin details, the last week has brought more “expressions of concern” from leading journals over prominent papers written by leading scientists. The latest concern regards papers in the New England Journal of Medicine in 2001 and 2004 and The Lancet in 2004 about oral cancer research done by Jon Sudbø of Oslo’s Norwegian Radium Hospital. The papers had claimed to make progress in distinguishing which mouth lesions are most at risk for developing into cancer — a breakthrough that might make earlier treatment possible and save many lives. But after NEJM editors found duplicated figures in the 2001 paper, writes Couzin, they examined the papers more closely and apparently pressed Sudbo, who admitted to fabricating some of the evidence. (Intriguing that once again, a duplicated figure led to the undoing of subterfuge. Several of the more notable frauds of the last few years — Hwang, Schon, and now Subdo — came to light when editors or readers noted that illustrations or graphs had been duplicated to buttress supposedly different data sets or manufactured altogether.)
If Sudbo indeed made major fabrications, he will have committed a fraud that is, if less elaborate than Hwang’s, perhaps more appalling, since he’s dealing not with potential breakthroughs but with patients presently in treatment.
Will we see an increase in fudged papers revealed? It’s possible, at least for a while, as editors, tired of getting fooled, may look more closely at papers making or following up large claims, or about which they or others have doubts they ignored before but now bring forth.
My piece in yesterday’s New York Times on errors in scientific journals lacked room to consider a key factor generating the sort of fraud that has haunted science lately: The way publishing concentrates and broadcasts not just the sort of error that John Ioannidis writes about, but power and money (its imprimatur), which can corrupt and lead to the sort of fraud Hwang indulged in.
While top researchers a couple generations ago lived modestly, today they command fat salaries and wield immense power. These perks provide extra motivation not just for cheating but for the overreaching — the half-conscious stretching of data and interpretation — that starts many episodes of fraud. As science historian Horace Freeland Judson shows in The Great Betrayal: Fraud in Science (Harcourt, 2004), many fraudmeisters start fabricating to cover prior, half-innocent data tweaking. Jan Schön and Victor Ninov, for instance, brilliant young physicists working separately at top labs, apparently believed sincerely, even as they moved from culling “aberrant” data to contriving fabrications, that they had made legitimate discoveries (Schön a one-molecule transistor, Ninov a new element) that needed bolstering. Exposed in 2002, Schön and Ninov insisted they never invented data. Neither denial seems credible. Yet in both an early belief in a possibility — an intense bias — that grew to overwhelm first judgment and then ethics.
Hwang’s stem-cell fraud dwarfs these episodes, of course. But it reveals a couple elements common to some of the biggest scandals of the last few years, hubris and a willingness to exploit power to create or maintain scientific triumphs that bring even more power. Journal publication gives researchers power because it seems to come from on high, bearing an authority that — as Ioannidis’s argument that most journal findings are false suggests — is illusory. The closed peer review system, with its granting of irreproachable, anonymous (rather Godlike) approval, is a key element in granting that authority and through it the power that researchers can gain through publishing. That’s one more reason some want to overhaul it.