Monday, March 28, 2016

A review of "Derailment", Diederik Stapel's autobiography

In 2011, social psychologist Diederik Stapel was accused of faking his data.  As allegations mounted, Stapel admitted to fraud and was fired his university post.  The incident was widely covered in the news (I blogged about it here) and is one of the precipitating events for the current conversation in psychology about reproducibility.

Two years after the scandal broke, Stapel wrote an autobiography, "Ontsporing", or "Derailment".  An English translation of this autobiography is now freely available for anyone to read.

If Stapel is an admitted liar, why should we take his autobiography seriously?  The thing is, although less severe forms of scientific misconduct are more common, outright fraud is
rare -- about 1.97% of scientists admit to committing fraud in surveys on the subject (Fanelli, 2009).  Rarer still is for someone admit to fraud publicly and then go on to write about their experiences.  Stapel's autobiography therefore has value in that it provides a window into the psychology of someone who did not just tiptoe the shallows of scientific misconduct, but who dove in headfirst.

In other words, I thought I might learn something about why Stapel decided to commit fraud by reading about his experiences in his own words.  So I did.



---

Stapel describes reading a book in college by Benjamin Kouwer entitled "Het Spel van de Persoonlijkheid", or "The Game of Personality".  The central argument of this book is that personality is a fiction, an epiphenomenon, a game played through endless transactions between people and their environments.  As Stapel puts it,

 ". . . a person is like an onion. As we peel off the layers of our public self, our private self, our ideal self, our moral self, our sense of self-confidence and our feelings of self-esteem, all of the layers and depths and identities and roles that we recognize in ourselves… then at the end, there’s nothing left. We don’t have a core."

This premise forms the narrative backdrop against which Stapel tells his story of fraud.  It is a somewhat unsettling frame.  If personality is an empty construct, then there is little use appealing to personality as an explanation for why people do the things they do.  If personality has little explanatory power, then, given the right circumstances, anyone has it within them to do anything --  including commit scientific fraud.

This line of argument has particular resonance in social psychology.  One of the great intellectual traditions in social psychology is situationism, which is premised on the idea that the forces outside oneself are a great deal more important for explaining human behavior than the traits people carry around with them.

Situationism is often portrayed as an optimistic philosophical position -- after all, if situations have more influence on behavior than traits, then, given the right combination of situational influences, any child can learn higher math, any criminal can be rehabilitated, and even the most ingrained stereotypes and prejudices can be changed.  However, situationism has a darker side.  If situations are structured to promote antisocial behavior, we should not be surprised when people behave antisocially.

Indeed, Stapel describes a myriad of institutional forces that drove him toward fraud.  Foremost among these is the relentless publish-or-perish culture that pervades academic science.  Competition for resources in academic science is intense.  The pool of PhDs continues to expand, whereas the number of positions, grants, and awards does not.  This resource scarcity encourages young scientists to view their colleagues as potential competitors, whom they must outcompete in Darwinian fashion by publishing a greater quantity of papers in better journals than their peers.  This competition creates a perverse set of incentives: they encourage people to use any means necessary to create papers that are "publishable", while discouraging people from working in the teams that might provide oversight from questionable practices.  Stapel himself says,

"I’d started working alone because I preferred it that way, being able to keep control over everything I was doing, but the more I turned inward on myself, the more unfeeling and unprincipled I became. Maybe I did want to cry, or be angry, or throw up, but I’d spent years arranging my professional life so that there was nobody around to feel those emotions. In my solitude the emotions just ebbed away." 

In addition competing for publications (and the grant money that can result from publications), the expansion of media of all sorts, including scientific media, has provided a fresh resource over which to compete: eyeballs.  In the context of scarce public funds, media attention can provide an indirect means of obtaining resources, as researchers who are adept at courting media attention can obtain lucrative speaking engagements.

Attention can also be a motivating force on its own.  Stapel describes how his intense craving for attention and approval came to dominate his work:

"I’d gone into science out of a fascination for the content of the subject, but I found myself more and more often in situations where the content wasn’t the most important thing, or in some cases wasn’t important at all. I was an enthusiastic scientist and I wanted to do my research as well as possible. But 'as well as possible' is difficult to be objective about, and like a lot of people, I need objective, 'honest' feedback about my performance. 'Good? Is that A-minus good, A good, or A-plus good?' So over time, my need to score and my quest for applause came to dominate. Scoring gets you applause, and an objective high score gets you loud applause. And loud applause is great, because it drowns out some of the doubts you’re having about whether what you’re doing and who you are is really worth anything."

Stapel became adept at creating studies that would provoke figurative applause.  The foremost example of this is a paper where he purported to find that messy environments lead to discrimination against outgroups (Stapel & Lindenberg, 2011).  This paper has a message that is easy to communicate to non-specialists.  The message has immediate implications for how to improve peoples' lives -- just clean up those things that make the environment messy.  The paper was placed highly in the journal status hierarchy -- it was published in Science, which is about as high as you can go.  Finally, the findings are plausible without seeming too obvious, a combination that makes for good news, if not good science.

This, of course, is the problem with science designed to be newsworthy rather than true.  Science that draws attention needs to strike just the right balance between confirming and defying peoples' prior expectations.  The truth, on the other hand, does not care at all about this boundary.  There is a big difference between what is attention-grabbing and what is true.

This leads to the final category of incentive -- the relentless drive for simplicity in a field where our phenomena are chaotic and complex.  Psychology wants desperately to be viewed as a "hard" science.  The problem is that, much as we wish otherwise, psychology is not physics.  Our data are noisy.  Most datasets contain warts that make interpretation difficult.  And yet, the relentless drive for simplicity, for a clear theoretical "story", drives psychologists to sand out the rough patches that make data confusing.  We wish to be taken seriously so badly that we carry on the pretense that our data are just as neat (and our theories are just as rigorous) as those in the "harder" sciences.

Stapel himself felt this drive for simplicity keenly:

"I did what I could. I philosophized and analyzed until my head hurt, and I tried to produce even better, more incisive and smarter descriptions of what I’d developed or discovered, but I kept getting the feeling that it wasn’t enough. Everything was just too complex. My stories weren’t elegant enough, and my results weren’t exciting enough, to get published in the top journals. Even after I’d deployed all my favorite statistical and methodological tricks to pimp my studies, they weren’t good enough. Even my best efforts were seen as mediocre. Not impressive enough, not interesting enough, not innovative, much too complicated. I wasn’t good enough."

The drive for simplicity led Stapel to create data that were, quite literally, too good to be true.

The pressures that Stapel describes are recognizable by anyone who has spent time around academic science.  I have felt these same pressures, as have many other scientists.  And yet, if we accept the situationist premise that Stapel's actions were due in large part to these pressures, the logical conclusion is that all scientists that are caught in this system have the potential to become another Stapel.

Herein lies the temptation to dismiss Stapel as a liar, a bad actor.  If Stapel is a bad actor, then the fault for his fraud lies in him rather than the system in which he was situated.  Blaming Stapel for his actions becomes a form of apoligism for scientific business-as-usual.

Stapel himself seems to feel some temptation to absolve the system by taking sole responsibility for his actions:

"I saw the cookie jar, and everything became possible. Others didn’t see it, or if they did, they thought about the problems and the risks, acted differently, were smarter, were wiser, felt the moral pressure to conform.
I, me, always me.
It’s not the environment. It’s me."

It is almost self-evidently true that certain aspects of Stapel's personality predisposed him to committing fraud.  As I mentioned in the beginning of this post, and as Stapel himself articulates, not every scientist commits fraud, yet every scientist is exposed to the same systemic pressures to cheat.  There must be personal factors that distinguish bad scientific actors from good.

Yet, we also can't fully absolve the system in which scientific practices take place.  To do this robs the Stapel incident of its power to teach, and willfully ignores the systematic problems that many of us already privately acknowledge.  One need not subscribe fully to situationism to admit that incentivizing people to pursue publishability at the expense of truth will create more Stapel-like behavior than if these incentives were not in place.

Fortunately, these incentives may be changing.  From the recent editorial by the incoming Journal of Personality and Social Psychology editor Laura King that her section will allow greater data imperfections to the large-scale preregistered ego depletion replication project, there are abundant signs that psychologists want the system that produces their science to promote truth.  In this context, we should remember Stapel not just as an embarrassing footnote, but as a reminder of the kind of scientist that rotten incentives can produce.

No comments:

Post a Comment