Map of life expectancy at birth from Global Education Project.

Monday, August 23, 2010

Crime Against Truth

Yeah, it's really ugly when politicians and professional yackers make stuff up that isn't true. I've spent a lot of binary digits here wailing and moaning about our non-reality based political culture. But let's face it, we expect politicians and Glenn Beck to lie, that's their job, at least as they see it.

But if we humans are to continue our slow, painful climb out of the darkness, the scientific enterprise absolutely must be trustworthy. Scientists are human and they are often tempted by rewards other than truth -- prestige, money, fame, power. If they succumb to any of these at the expense of truth, it's a betrayal not only of the institutions where they work, and the scholarly community of which they are a part, but of all humanity. False findings and wrong conclusions can get embedded in the literature and take decades to expunge. Meanwhile investigators are led down false paths, funding is squandered, opportunities for important discoveries are missed, patients are mistreated - the consequences cascade endlessly.

That's why the case of Marc Hauser is so deeply disturbing. He was not only a prominent scientist and influential public intellectual, but part of a community of important thinkers and well-known champions of reason to which I personally feel a deep attachment, including Stephen Pinker and Noam Chomsky.

Arthur M. Michalek and colleagues (and no, Donald L. Trump is not The Donald) review the directly measurable costs to their own institution of a single case of scientific fraud, not nearly as important as Hauser's. They come up with a bottom line of $525,000, but that does not include the so-called intangible costs as falsehood stains the very fabric of the field. In fact, unlike the Hauser case, that didn't really happen here. This particular fraud was nipped in the bud, but Hauser has had three fairly important articles retracted or corrected, and his entire body of work must now be called into question.

Tom Bartlett's account (see my first link) suggests that Hauser may have been guilty more of wishful thinking than of conscious fraud. He coded the behavior of monkeys, but it turns out that only he could see the behaviors that supported his hypothesis. Graduate students who observed the same data saw the opposite.

However, that the design of his studies even made that possible is a grievous fault. The only legitimate way to do this is to have coders observe videos of the monkeys' behaviors while blinded to the stimuli being tested. It's just elementary that having the investigator, who has a compelling personal stake in the outcome, do unblinded coding in this way is invalid to begin with. It's a mystery how this work even got published if the methodology was accurately described.

So no, Hauser doesn't get off the hook as being sloppy or careless rather than deceitful. There's just no excuse for this. However, it is not a crime, and it won't even cause him to lose his job. Harvard has already determined that he committed scientific misconduct, and if NIH makes the same determination he will be disqualified from getting NIH funding for three years. That's it, that's the worst that can happen. To be sure, as a practical matter he's going to have a hard time convincing a study section to recommend him for funding even after three years, his reputation is in tatters, and he'll likely end his career as a tenured non-entity. But I'm still going to take this opportunity to scold him.

We have to pile on to make the penalty for this sort of thing not worth the risk. In my group, we agonize over getting every little detail right. We spend days and weeks going through our data and documenting everything we do, every time we touch it, who does what, and making sure that every case is labeled correctly, described correctly, and assessed correctly. When something gets screwed up -- and that will happen -- we make absolutely sure we have it straightened out before we move ahead. It's a major pain in the ass, it slows us way down when it comes to publication and grant writing, it's boring and it's annoying. But it's what you absolutely must do.

1 comment:

Cervantes said...

I generally avoid long-distance diagnosis but I'm going to say Schizophrenia.