Unless you are a chemist or a habitual follower of scientific misconduct stories, it’s possible that you missed the saga of Bengü Sezen.
From 2000 to 2005, Sezen was a graduate student in chemistry at Columbia University, working in the laboratory of then-Assistant Professor Dalibor Sames. She appeared to be a talented scientist in training, and during her graduate studies was lead author on three papers published in the Journal of the American Chemical Society. Columbia University conferred upon her a Ph.D. in chemistry (with distinction).
But, as it turns out, her published results were not reproducible, an issue raised by chemists at Columbia and elsewhere as early as 2002. Further, the results were irreproducible for very good reason: as reported by Chemical & Engineering News, investigations by Columbia University and by the U.S. Department of Health & Human Services (which is home to the Office of Research Integrity) revealed
a massive and sustained effort by Sezen over the course of more than a decade to dope experiments, manipulate and falsify NMR and elemental analysis research data, and create fictitious people and organizations to vouch for the reproducibility of her results.
In the wake of the investigations, Sames has retracted the papers coauthored with Sezen (Sezen refused to retract them on the grounds that she stood by the work), and Columbia has revoked the Ph.D. it granted Sezen.
The evidence from the investigations supports the hypothesis that Bengü Sezen was a liar masquerading as a chemist, that she claimed to have done experiments that she hadn’t, to have obtained NMR spectra that she created (in part) with correction fluid, to have built molecules that she didn’t build. She committed fraud that introduced not just mistakes but lies into the scientific literature.
But she didn’t — she couldn’t — do this alone. She didn’t commit her fraud as a principal investigator (PI). Rather she did it as a scientific trainee, a graduate student working under the supervision of Dalibor Sames (who is currently an Associate Professor at Columbia). It’s worth examining what responsibility Sames bears for what happened here.
The PI is instrumental in creating new scientific knowledge, and certainly this is a source of some obligations to the rest of the scientific community. All scientists participating in the knowledge-building project are supposed to provide accurate reports of the experiments they performed and the results they obtained. As well, they are supposed to do what they can to eliminate possible sources of error in their results. Making up results, obviously, is right out.
But here, there is no evidence to suggest that Sames made up any results; Sezen did that. Does this mean that Sames bears no responsibility for the fraudulent papers that came out of his lab and bore his name?
Indeed, I think it’s fair to say that Sames bears a significant amount of responsibility, and not just as Sezen’s scientific collaborator. Along with producing new scientific knowledge, a PI at a Ph.D. granting institution is also working to train new scientists, and this role surely brings with it additional responsibilities.
One of these is a responsibility to ensure that the trainee knows what she’s doing — that she understands the scientific question she’s been asked to tackle and the experimental strategy being used to tackle it, that she knows how to actually perform the experiments and collect the data, keeping accurate records of how she set things up, what she measured, and the results. The PI cannot assume that the trainee knows everything when she arrives in his lab; she has come to the lab for training to help her become a scientist. This means that the PI is responsible to provide this training, maybe himself, maybe at the hands of trusted associates.
Given the practice of coauthoring research papers with trainees, there’s a clear incentive for PI’s to make sure their trainees learn what they need to learn to conduct clean experiments and master data collection. The PI’s name appears in the author-line of the paper alongside the trainee’s. His scientific reputation is tied to the quality of the work the paper communicates. Thus, as a matter of routine quality control, it should matter to the PI if the trainee can control experimental parameters or keep a coherent lab notebook.
But beyond quality control, the PI has a responsibility to help the trainee embrace the norm of organized skepticism. As well as technical skills, the trainee needs to gain good scientific judgment, which includes a constant effort to notice ways that one’s scientific results or conclusions could be mistaken. A goal of science, after all, is to work out how various phenomena in our world really are — which means that scientists are committed to ferreting out accounts that get those phenomena wrong.
Often, scientists are propelled to cast a skeptical eye on their own findings by the prospect of other scientists scrutinizing them for errors, especially since these other scientists may be in competition for the same grant money, the same jobs, the same hope of recognition for an important discovery. However, a trainee may not snap into skeptical reflection so automatically. As the trainee’s advisor, the PI can help with this by regularly greeting the trainee’s exuberance about a run of good results with careful questions about the seventeen or so potential sources of error that ought methodically to be checked. (Trainees can also suffer from an excess of skepticism, fearing that none of their results will hold up when someone else tries to repeat the experiment. A good advisor can help such trainees take reasonable steps to confirm the goodness of their work without going overboard.)
Verily, the “organized” in organized skepticism should remind us that the careful checking of scientific work is generally conducted by a community of scientists (the community upon which we depend if we’re after something like objective knowledge). As it turns out, most new scientists are trained in the context of a laboratory group. They share their environment with other trainees at various stages, and they learn to function together to do things like troubleshoot experimental difficulties, track down relevant literature, perform sensible analysis of their data, and anticipate objections to their conclusions. The lab group, in other words, can function as a microcosm of larger scientific communities, providing an opportunity for the trainee to learn ways to interact productively with collaborators (and perhaps even competitors) under the supervision of the PI.
To the extent that the PI recognizes the importance of the interactions between scientists to build knowledge claims and make sure they hold up, the PI might be motivated to take the pedagogical value of such interactions within his own lab seriously.
However, the evidence from the investigations of Bengü Sezen make it clear that Dalibor Sames had a problem on this front. In particular, at least three other trainees in the Sames lab brought their concerns about Sezen’s work (and their inability to reproduce it) to Sames. Rather than take these concerns seriously, however, Sames dismissed two of these trainees from his research group (which, in the absence of another professor stepping forward to fund and train these students, amounts to dismissing them from their graduate program altogether); the third apparently decided to leave the program him- or herself, although it seems likely that this decision was shaped by an awareness that Sames was unlikely to be supportive of the concerns this student voiced.
Sames, it appears, had decided that Sezen was a star, a trainee with great potential who was producing exciting results. On the basis of this judgment, he seems to have decided that the other trainees in his lab who could not reproduce Sezen’s work were technically incompetent, or worse, troublemakers who were jealous of Sezen’s scientific prowess. It is striking that he seems to have made these judgments without enough hands-on involvement in the scientific work of any of these trainees to determine whose work was actually reliable and whose was not. Worse, Sames seems to have forgotten entirely that he had just as much responsibility to mentor the students pointing out problems as he had to mentor the student producing what seemed to be impressive results.
How should a PI working to mentor all the trainees in his research group respond to such a situation? When there is a disagreement about whether the experiment works, a scientific mentor should ask where the data lead, and should be ready to help the team work out a strategy to collect reliable data and follow them. A scientific mentor should also emphasize that the scientific truth (as far as the data can establish it) is more important than which set of hands or eyeballs first identifies it or whose hunches are vindicated by the evidence. This means reminding all his trainees that every scientist’s findings are susceptible to error — and that good scientific work includes helping others to locate and avoid potential sources of error.
The students in the Sames lab who spoke up about Sezen’s too-good-to-be-true results did exactly what they were supposed to do. They reported their own data and where it led. They communicated their findings within their research group, trusting, I imagine, that their PI was interested in the truth about the bits of the world they were studying. And, for their efforts, they were squashed like bugs.
Bengü Sezen got a Ph.D. from Columbia University (although it has since been revoked). Dalibor Sames got tenure and a promotion. These three scientific trainees got shown the door.
Maybe they found their way into another Ph.D. program, or another scientific career that didn’t require Ph.D. training. Or maybe they gave up on the world of chemistry as a snake pit where politics matters more than sound scientific methodology. In any event, it seems fair to conclude that they did not get the scientific training and mentoring they deserved from Dalibor Sames.
Academic laboratories make new knowledge and new scientists. The second of these functions is crucially important if we do not want the ability to build new knowledge to die out with the present generation of PIs. (As well, ignoring the training function of these laboratories makes it look an awful lot like university science research laboratories exist primarily to exploit cheap student labor.)
Dalibor Sames dismissed two honest students, discouraging a third to the point of quitting, and anointed Bengü Sezen as a skilled and trustworthy scientist (which the evidence now shows she wasn’t). This suggests that Sames did not succeed in the task of training new scientists.
Moreover, there is no indication that Columbia University has sanctioned Sames at all for his role in Sezen’s fraud, nor for dismissing the students who raised reasonable questions about her work. To all appearances, then, Columbia University doesn’t take the job of training new scientists very seriously. If it did, not only would there be some non-trivial consequences for Sames (which might include serious mentoring in how to mentor scientific trainees), but there would also be some serious effort to arrange the training of graduate students so that they would not be so subject to the whims and judgment of a single faculty member.
Chemists-in-training in a graduate chemistry program have the potential to impact the larger community of chemists, for good or for ill. The chemistry faculty at Columbia University, as a group, has a duty to all the students in the Columbia graduate chemistry program — to make sure these students learn how to be good chemists, but also to make sure they are not being punished for asking reasonable questions or sharing unwelcome results. Further, members of the chemistry faculty arguably have a duty to get involved when a colleague drops the ball, whether with respect to mentoring his trainees or with respect to exercising appropriate skepticism about his own results.
There might be a natural impulse to side with one’s fellow faculty member over a graduate student (who is, after all, just passing through your department). However, if you’re doing the job of training new scientists right, graduate students are not transients in one’s scientific community. They are your scientific sons and daughters. The future of your discipline depends on the lessons you teach them.
What lessons have Dalibor Sames and Columbia University taught the rising generations of the community of chemistry?
* * * * *
For more on the Sezen fraud (and Sames’s involvement), I recommend William G. Schulz’s coverage in Chemical & Engineering News (here, here, here, here, and here) and the indispensable coverage at ChemBark. I also wrote about it on my other blog as the case unfolded (here, here, here, here, and here.)
