Sunday, October 30, 2005

Aw Mom, scientific misconduct again?!

Can you believe there's another story in the news about a scientist caught fabricating and falsifying data? Also, the sky is blue.

This time, the miscreant scientist is Luk Van Parijs, who was an associate professor of biology at MIT until they fired his ass. His offenses include "fabricating data in a published scientific paper, in unpublished manuscripts, and in grant applications"; apparently, he also admitted to falsifying data. Gareth Cook and Marcella Bombardieri write about the incident for the Boston Globe.

Why does this kind of thing keep happening? Maybe if we knew, we could put a stop to it. In the meantime, here are my thoughts about the various players in this affair. (All quotes are from the above-linked Boston Globe article unless noted otherwise.)

The other researchers in Van Parijs's lab at MIT
The investigation began in August 2004, when a group of researchers in Van Parijs's laboratory brought their concerns to university administrators.

Well done, other researchers! The Globe article doesn't give details about who these researchers were. They may have included postdocs, graduate students, technicians, and visiting scholars. In any case, it is a good bet that they were lower on the foodchain than Van Parijs.
[Alice] Gast [associate provost and vice president for research at MIT] praised the scientists who made the initial allegations, saying that the university depends on all of its members to defend the integrity of research, even if it means the awkwardness of challenging friends and colleagues.

Keep in mind, if the other researchers really were lower on the foodchain than Van Parijs, that we're talking about something significantly more awkward than challenging a friend and colleague — we're talking about challenging the boss. You know, the guy whose letter of recommendation will make a big difference to your future as a scientist. The guy who needs to sign off on your dissertation. The guy who pays your salary.

Were there conversations in which these other researchers raised their concerns with Van Parijs before they brough their concerns to university administrators? The article doesn't say, but it's hard to imagine that there weren't. Maybe the questions were oblique. But you want to be reassured that the research projects of which you are a part are on the level. Really, there are so many messy consequences that flow from your boss being a cheat that any other theory with evidence to support it is preferable. Only when you're sure that he's really crossed to the other side do you get the admistrators involved.

Standing up for the integrity of one's own research, and of scientific research as a whole? That's a good thing. Mad props to the whistle-blowers.

MIT
MIT said Van Parijs quickly admitted fabricating data, as well as falsifying data, which means changing it in a misleading way. The confidential investigation was conducted by MIT scientists whom Gast declined to name. Van Parijs was placed on paid administrative leave in September 2004, and did not have access to his lab, she said...

Gast said that MIT is working with his coauthors on retracting published errors and that all of Van Parijs's colleagues have been very cooperative. The university is also preparing a report to the US Office of Research Integrity, an agency of the Department of Health and Human Services that investigates scientific misconduct involving federal funds, so that it can perform its own investigation. She said the university immediately stopped the spending on his grants when the problem was discovered, and would work with the government to determine what money needed to be returned.

MIT did well in taking the concerns about Van Parijs's work seriously. They started investigating the allegations quickly. They stopped spending the grant money, recognizing that the funders of the research thought they were funding the collection of actual data rather than fabrications. They are involved in correcting the problems in the published scientific literature stemming from Van Parijs's misconduct. MIT gets that science demands a high standard of integrity, and they don't want the MIT brand associated with behavior unbecoming a scientist.

At the beginning of the investigation, they put Van Parijs on paid administrative leave. After more than a year of getting the facts, they were comfortable enough in their assessment of his deeds (and his character) to fire him. Even if Van Parijs had been tenured (he was not), MIT would have been able to remove him "with cause". Scientific misconduct is a pretty clear cause for removal.

Kudos to MIT for finding the facts and taking appropriate action to remove a bad actor from its faculty. Beyond cleaning its own house, MIT acquits itself well by its "ongoing work to correct the scientific record". Not mentioned in the story is how MIT handled the dissolution of the Van Parijs lab; I hope efforts were made to find reasonable positions for the postdocs, grad students, and technicians displaced by this dissolution.

Luk Van Parijs
It goes without saying that a scientist ought not to fabricate or falsify data. Fabrication and falsification suck. These deeds are varieties of deception. Deceiving the people reading your papers in the journals, or the people reading your grant applications and deciding whether to fund your research, is crappy. And, fabrication and falsification suck even more when done in papers on which you have coauthors. You're dragging good scientists down with you. Even when you've been taken out of the game, they still have to worry about corrections, retractions, and the lasting impact on their reputations.

But wait, what's this?
In an e-mail sent to the Globe last night from his MIT account, Van Parijs said, ''I was shocked at the timing and manner in which MIT made the announcement since I had cooperated with the investigation to the fullest of my capabilities."

Excuse me? Dude, you were caught. Indeed, when confronted, you confessed your wrongdoing. MIT had to fire you. It's not like anyone was going to be able to trust any of your data again. You knew what you were doing was wrong ... and you still did it.

You cooperated? Great. But that doesn't mean MIT can, or should, keep your bad acts a secret. They can't protect your reputation without simultaneously hiding the fact that you put crap into the pool of scientific knowledge. They are not going to keep things hush-hush so you can, maybe, get a job in science someplace else and not have to work under a cloud of suspicion because of your prior bad acts.

Get a clue. You're not a bewildered first year grad student who doesn't know the rules of science. (Indeed, it's quite possible that your grad students were the ones with the moral compass and the courage to rat you out.) You're an associate professor in your mid-30s. If you don't know how to be an ethical scientist by now, is there any good reason to think you'll pick it up at some point in the future?

Van Parijs's mentors
He worked in the lab of Dr. Abul Abbas at Brigham and Women's Hospital for three or four years, until about 1997, according to Abbas, now chairman of the pathology department at the University of California, San Francisco. Abbas said he did not see any indication at the time that Van Parijs might falsify data. He said he is talking with people at Brigham and Women's Hospital to decide whether to investigate the work Van Parijs did while in Abbas's lab.

One imagines if Abbas did see an indication that Van Parijs's, or any other graduate student under his supervision, might falsify data, he would have grabbed him by the shoulders and given him a good shaking. A good research advisor wants you to learn how to do good research. Learning how to obtain clean, reliable data is a lot harder than learning how to make up data that suits your hypothesis.

[H]is career had been highly promising, including being hired as a postdoctoral fellow in the lab of famed Nobel laureate David Baltimore.

''He was a very personally attractive, excited, and thoughtful guy who cared about a wide range of science," Baltimore, now president of the California Institute of Technology, said in an interview yesterday. ''When I first heard there was a question about his work, it came as a very great surprise to me."

Dr. Baltimore, we meet again. Why is it that you are always surprised when someone raises a question about a collaborator's work? Is it that you have an optimistic view about your collaborators? Or do you maintain a certain ... distance from the day to day details of what they're up to? I'm not going to call it willful ignorance, but maybe you should consider collaborating more closely when you undertake collaborations.

The broader culture of science
By no means do I want to say that circumstances made Van Parijs cheat. He owns this mess. But, I have to wonder just a little if, given some changes in the culture in which he was working, we'd have fewer cases like this.
'He got a job and finished his postdoc training much faster than average people," said Xiao-Feng Qin, who worked on the same laboratory bench as Van Parijs at Caltech and is now an assistant professor in immunology at MD Anderson Cancer Center in Houston. ''I think people thought he was a golden boy, because he finished so fast."

You have to get results, and publish them, to stay in the game of science. Doing more, and doing it faster, gives you an edge over the competition for the scarce resources of jobs and grants. Getting really amazing results helps too.

But getting good results, in the real world, takes time. Doing things right sometimes means doing things over. Doing things right often means your hypotheses don't pan out. Writing up clear articles that present your findings accurately and explain what they do, and do not mean, takes time too.

Valorizing high output at great speed over careful research may be a problem. At the very least, we ought to not simply marvel out the golden boy's productivity; we ought to look at his output very carefully to make sure it holds up. Science isn't supposed to be about an impressive biography or being "personally attractive" and fun to talk with. We're supposed to be trading in real results!

Will the community of science learn the lessons from the good and bad actors in this case, take them to heart, and nip future misconduct in the bud? It would be nice to think so. But it wouldn't surprise me if we've got another case like this six months from now.

"Prove me wrong, kids. Prove me wrong!"


Technorati tags: , ,

0 Comments:

Post a Comment

<< Home