Book Review: On Fact or Fraud

A Review of On Fact or Fraud: Cautionary tales from the front lines of science by David Goodstein (Princeton University Press 2010)

Review By Eugenie Samuel Reich

The sociologist of science, Steven Shapin, once told me that he could not see the interest of studying fraud in science. Fraud, according to Shapin, revealed more about pathology of a few solitary individuals than anything about science as a whole. I've always felt instinctively that Shapin was wrong about this, and wished that he and other sociologists of science had spent more time understanding fraud. As a result, I was very happy to find a book that starts out from the same assumption that I have: that cases of fraud in science - including alleged, suspected and actual cases - can reveal something about the way science works. On Fact and Fraud: cautionary tales from the front lines of science is an accessible, well-written contribution to a relatively understudied area. It's by David Goodstein, whose background as a working physicist and administrator at the California Institute of Technology places him far from the sociology of science. Goodstein's book also contains some of the ingredients needed to prove Shapin wrong in his view of fraud, although I found that it stopped disappointingly short of reaching the conclusions that the ideas in it point towards.

Goodstein starts out by presenting a kind of back-of-the-envelope theory of fraud in science. According to Goodstein, fraudsters:

  1. are under career pressure
  2. think they know the answer their experiments will give and so do not bother to do the experiments properly
  3. are working in a field where results are hard to replicate

The theory is a good start; the problem is that it doesn't go far enough. As Goodstein himself says, almost all scientists are under career pressure. Many scientists also think they know what their experiments will show. And science is also often hard to replicate. Why, then, do some scientists act fraudulently while others do not? Come to that, why do some frauds escalate into sensations that prove particularly embarrassing to the scientific community (the case of stem cell researcher Woo Suk Hwang of South Korea is a good example) while others are nipped in the bud early, for example, by a vigilant supervisor, trainee or colleague of the fraudster?

Goodstein's book falls short of answering these crucial questions. His theory on fraud is supplemented by some ideas about the social structure of science, which is described as containing a 'reward system' and an 'authority structure.' The 'reward system' covers all the ways that scientists express esteem for one another. The 'authority structure' guides and controls the reward system and includes the institutions in which scientists are employed (Goodstein mostly considers universities, but presumably this could include government and corporate laboratories as well). As scientists strive to work their way upwards in the 'reward system', scientific fraud is described by Goodstein as a temptation 'lurking quietly in the shadows', although never as something central to science. If this is granted, it becomes unclear how the ideas that Goodstein is presenting have helped to explain fraud. The problem is that Goodstein seems too reluctant to admit that fraud could be directly encouraged or magnified by features of the 'reward system' and 'authority structure.'

Further details, on the role played by the 'reward system' in fraud cases of the past, might have helped this account. For example, many scientists I have spoken to, especially junior ones still moving up the career ladder, complain about excessive pressure and rewards for getting into prestigious journals -notably Nature and Science -- which have tended to publish short accessible papers reporting on the latest results instead of long technical papers on experimental methods. Papers in such journals impress colleagues, and count heavily in a candidate's favor in tenure review procedures at universities, so they are undoubtedly an important part of today's 'reward system.' Yet cases of fraud in recent years have also made clear that the incentives to publish in these journals can act to the detriment of the truth. The physicist Jan Hendrik Schon was a post-doc and later staff scientist at Bell Labs in New Jersey at a time when submissions to Nature and Science were prized by managers. Schon became obsessed with the two journals and published eight papers in Nature and Science in 2001 -a young scientist at a top university would usually be lucky to publish two in a year. As I report in my book on Schon's case, when Schon was asked to write a long technical paper emphasizing methods rather than results, he even needed advice from a colleague about what other journal he might publish in!

Incentives operated similarly for the cloning scientist Woo Suk Hwang, who was prominent in South Korea at a time when the country was keen to improve the profile of its scientists internationally. Hwang took extreme steps to get two papers into Science, including fabricating the details of ethical protocols on the donation of human eggs, adding a figure of questionable data to a manuscript in response to questions from editors, and recruiting a respected American stem cell scientist, Gerald Schatten of the University of Pittsburgh, to co-author the work. Looking at such examples, as well at some of the breakthroughs that have lead to Nobel prizes (the painstaking work on blood antigens that lead to the discovery of hepatitis B by Baruch Blumberg is a good counterexample) one is driven to the conclusion that the scientific system works best when scientists place their emphasis on methodological and technical details that are the nuts and bolts of scientific breakthroughs. A change in science that rewarded solid methods instead of sexy results would make life harder for data-fabricators who tend to lack information on plausible methods because they have not really done the work.

Part of Goodstein's approach is to compare examples of fraud with examples of 'irreproachable' science. His main example of the latter is the discovery of high-temperature superconductivity in 1986. The period that followed the discovery of high-temperature superconductivity is famous for having involved lots of hyped and excessively enthusiastic claims. When I was reporting on Schon's case, I was regaled by tales of corner-cutting and ethical breaches in this area of physics. Goodstein does not consider any possible link between major frauds and an environment of hype. Yet Schon, whose case he does discuss, produced some of his more remarkable breakthroughs in the field of nanotechnology. Woo Suk Hwang was working in stem cells, also a field contaminated by hype at the time. Following Goodstein's theory we would gather that high-temperature superconductivity avoided a fraud scandal in part because the effect was easy to reproduce, and in part because it was a surprising phenomenon so that no one could have expected the results to come out the way they did. But this does not make the science of high-Tc 'irreproachable,' to use Goodstein's term. In fact ethical breaches did occur - one scientist told me that during high-Tc he learned about an important experiment by being asked to review a paper about it and immediately directed his own laboratory to study the same effect, hoping to get his own paper out on a competitive timescale. This scientist didn't see the problem with using the review process as a source of tip-offs when a field was very competitive.

Why then, does Goodstein steer so far clear of finding any fault with the scientific system. The reason, I strongly suspect, is that he is part of science himself, and wishes to avoid harming it by being too candid. Goodstein made his name as an authority on scientific misconduct while serving as vice-provost at the California Institute of Technology. He thanks the provosts at Caltech that he served under in the acknowledgment section of his book. His career at Caltech traced the emergence of a unified federal policy on research misconduct, including controversies such as the notorious Baltimore case, when Nobel Laureate David Baltimore defended a collaborator against accusations of misconduct at MIT. Baltimore went on to become president of Caltech, including at the time that Goodstein was vice-provost. Goodstein is up front about all of these connections, but in several places allows the bias that arises from it to influence his accounts without apparently realizing he is doing so. For example, he discusses two cases of fraud in biology at Caltech that he says were handled in an exemplary fashion, but omits to discuss the case of Luk Van Parijs, an associate professor of immunology at MIT who in 2007 was found guilty of misconduct committed at Caltech while working in Baltimore's lab, including on two papers on which Baltimore was the senior author (the research was retracted in 2009). Despite widespread rumors and credible allegations against Van Parijs' work at MIT, Caltech failed to investigate Van Parijs' work at their institute until irregularities were brought explicitly to Baltimore's attention by a journalist (myself, while working on an article about the case for New Scientist magazine). Another place Goodstein's bias interferes is in his account of Robert Millikan, a Caltech professor who won the Nobel prize for measuring the charge on the electron using oil droplets. Historian of science Gerald Holton published some pioneering historical work in 1978 showing that Millikan, despite claiming in one paper to be reporting on every drop he studied, in fact omitted to report on some of the droplets he had studied. Goodstein lapses into a troubling apologism for seriously unethical misbehavior when he writes that Millikan probably forgot about the omitted droplets.

Taken together Goodstein has generated a useful primary source that documents, in that it suffers from, some of the bias that senior people in today's scientific community have when it comes to dealing with fraud. The major problem that institutions are usually reluctant to investigate or to publicize their own failings - as with Caltech in the case of Van Parijs - is barely addressed in the book and ideas bearing on the motive for fraud are not filled out in detail. Parts of the book nevertheless make for a thoughtful contribution to the landscape of reading on fraud and select sections - such as the introduction - may be worth adding to the syllabi for ethics and misconduct courses.