Can You Criticize Science (or Do Science) Without Looking Like an Obsessive? Maybe Not.

We need to normalize the pursuit of accuracy as a good-intentioned piece of the scientific puzzle.

Russell Crowe as Javert in 2012’s Les Misérables.
Russell Crowe as Javert in 2012’s Les Misérables.
Universal Pictures

The scandal that exposed fundamental flaws in the published work of eating-behavior researcher Brian Wansink—called Pizzagate, not to be confused with the Pizzagate conspiracy theoryhas been instructive on many levels, but perhaps one of the most valuable takeaways is what it’s revealed about the process of correcting the record in academic journals. As Jordan Anaya details in his Medium post “How Journals Responded to PizzaGate,” Anaya, Nick Brown, and their colleagues have faced extreme obstacles in their efforts to expose problems in Wansink’s published works. What’s scary about this story is how much work was required by Anaya and Brown, and how resistant the academic journals have been to outside criticism.

Wansink’s is about as easy a case of scientific misconduct as you can find, excepting people such as Diederik Stapel who flat-out admit that their work was fraudulent. Wansink’s case is easy because he described some of his questionable research practices on his own blog and also because there were lots and lots of false numbers in his published work (more than 150 errors in the first four papers that people looked at, and then lots more errors kept coming in). Also, while Wansink has had an effective organization at Cornell University to publicize his work, he doesn’t seem to be making any concerted effort to defend his work, and to his credit he has not been attacking the people who have uncovered errors in his work.

So, it should be a simple case: There are lots of errors right on the surface, and there’s not really anyone defending the work. But, still, it took months of pushing and pushing and pushing for Anaya and Brown to get journals to admit there were problems. As Anaya explains, “If something is going to take several months, you might expect it to be a sufficient correction of the record, or at a minimum accurate. Unfortunately, we didn’t find either of those to be the case.”

My point here is not to go over the pizzagate story one more time. Anyone paying attention should now know not to trust the claims, published or otherwise, coming from Wansink’s lab. The point I want to make is how much effort was required by Anaya and Brown to get any changes at all, even in this easy case where there were actual and demonstrable untruths in the published papers. Not just questionable research practices, misinterpretation of statistics, overblown claims, etc. Those alone would be enough of a reason to disbelieve a published claim and enough of a reason for journals to post a correction. But in these cases, you have actual clear errors, and very minimal changes.

So then the question arises: Why would Anaya and Brown put in all this effort? Why are they so focused on this work, if they’re not getting any results? Is something wrong with them?

The answer is of course not. Instead of asking why are Anaya and Brown obsessively pursuing Wansink, why not ask, why was Wansink obsessively doing this flawed research? The answer to that latter question is actually obvious: It benefited Wansink’s career, also Wansink presumably thought, and thinks, that this work increased scientific understanding and benefited people, including food producers, restaurants, and everyday consumers.

The exact same thing is true for Anaya and Brown: Exposing bad research can increase scientific understanding and benefit people. If people can be helped by the promulgation of true facts about eating behavior, then presumably they can also be helped by the exposure of false claims (whether the errors were intentional or not).

But what about their careers? Here Anaya, Brown, and their colleagues are getting slammed both ways: If exposing bad work helps their careers, then their efforts here are self-interested and therefore suspect, bringing Wansink and others down only to boost themselves. But if exposing bad work does not help Anaya’s and Brown’s careers, if it’s not part of their jobs, and they’re just doing it out of a crusading sense of altruism, then this can be perceived as even stranger: After all, what sort of person gets off by obsessing over others’ mistakes?

Neither of these assessments are useful ways to categorize Anaya and Brown’s behavior—and one way to see this is to think about scientists’ behavior in general. Yes, we’re obsessive about our work and this obsession can be good for our careers. Anaya and Brown have been obsessive because they care—they care about getting things right, and they care about the larger benefits of cognitive and behavioral science for people’s lives, and the potential negative effects of the spread of false and misleading claims. Wansink’s got to feel that way too, right? Otherwise why do the science in the first place, if you don’t think that right information is better than wrong information?

To me, this is the Javert paradox. Inspector Javert, recall, is the character in Les Misérables who pursued Jean Valjean, a poor schlub whose misfortunes all began when he stole a loaf of bread. The problem with Javert (played by Russell Crowe in the 2012 film adaptation, pictured above) was not that he was indefatigable in his pursuit. No, the problem was that he was indefatigable in his pursuit of a guy who stole a goddamn loaf of bread to feed his family.

And here’s the Javert paradox, as it applies to this case: Their pursuit of Brian Wansink—or, I would say, their pursuit of the scientific truth in the Wansink saga—has taken many, many hours of work from Anaya, Brown, and their colleagues. It’s taken so much time and effort that the critics look like Javerts.

But … if they hadn’t been so thorough and careful—as Anaya puts it, “I know Wansink’s work better than he does, it’s depressing really”—then I suspect none of this exposure would’ve happened. Recall that the Wansink lab was called out five years ago on data problems, and the researchers just bobbed and weaved and never acknowledged any problems. And recall that, after the first batch of pizzagate errors came out (more than 150 wrong numbers in only four papers!), Cornell tried to dismiss the whole story. And then, of course, there are the journals that have required months of nagging to make even the smallest corrections.

So, here’s the Javert paradox: Suppose you find a problem with published work.

1. If you just point it out once or twice, the authors of the work are likely to do nothing, to either ignore the criticism entirely, or to brush it aside, or to make only the most minor corrections and then claim, without good evidence, that their main conclusions are unaffected. And the publicity machine goes on; the erroneous claims get featured in the press, students continue to work in the lab producing more questionable findings, etc.

2. If you really pursue the problem, reanalyzing data, considering alternative theories, contacting journals, writing articles on the problem for general audiences, then you look like a Javert.

Sometimes, though, this is the only way to effect change. It will look a bit obsessive, because it takes a lot of effort to pursue this sort of case. If Anaya, Brown, etc. want to do this, they can’t really pursue lots of others. Once you start on a project, you have to keep going on it. This is characteristic of science in general: You work on a project that you feel has some importance and some interest to you, and then you pursue it in depth.

In the end, I think either option is OK. It’s fine to point out an error and then let others follow up if they want to, and it’s fine to be obsessive. That’s part of what it’s like to be a scientist, or a journalist, or even a citizen.

Andrew Gelman is a professor of statistics and political science at Columbia University and author of several books, most recently Bayesian Data Analysis.