Wednesday, May 1, 2013

The Psychology of Academic Deceit

About 18 months ago, I did a meta-post on Diederik Stapel, of Tilburg, and before that of Groningen. In that post, I linked to Retraction Watch, which recounted Stapel's misdeeds, a social psychologist who didn't falsify, but outright manufactured, data-sets in numerous scientific papers, 54 of which have now been retracted. Stapel, now known as the Lying Dutchman (see this post by Catarina) was an international star and rose to be Dean of Social and Behavioural Sciences at Tilburg before he was fired for academic fraud.
The New York Times Magazine has now published (28th April) a fascinating story by Yudhijit Bhattacharjee on the man and his career. Here is some of what we learn:
  • Stapel was impatient with the messiness of empirical data and started at first to falsify, but after a while to manufacture data-sets in support of hypotheses he found plausible. For example, he published a study, supposedly conducted at Utrecht railway station, claiming that when the station was garbage-strewn (during a strike) people sat further away from a black man planted in one of the seats, than when the station was clean.
  • He became famous because of his elegant and compelling presentations, which were aided by his good looks, charm, and bourgeois good manners.
  • He was exposed by his (more working class, less good looking) friend, Marcel Zeeland, who overcame his initial scepticism when cornered by suspicious graduate students, and decided to do the right thing. Zeeland now avoids meeting Stapel as much as he can. He is filled with conflicting emotions of sorrow and anger.
  • When Stapel's younger colleagues and graduate students approached senior figures in the US and elsewhere for advice about what to do about their growing suspicions, they were told that it would be prudent to keep silent. (The students became suspicious when Stapel refused to provide raw data.)
  • Finally, two of the students hit upon the ruse of cornering Zeeland at a conference in London, where he couldn't easily get away from them before hearing them out.
Stapel is presenting a front of unqualified contrition. But there is something rather arrogant about even this, in the way of a prominent divine who flogs himself to pray for forgiveness. He has a tendency to present his story as a symptom of bad things in the profession—which, undoubtedly it is, but it is hardly for him to cast aspersions. (The NYTM story coincides, by the way, with the publication of Stapel's own account of his crimes. In her post, Catarina asks about this memoir: "Is it “a way to try to make money off of his terrible decisions”, as suggested by Bryce Huebner (to whom I owe the pointer to the article on Twitter)? Or is it a case of someone who is so used to being in the spotlight that any form of public attention is welcome?" Indeed! )
One of the more interesting insights in Bhattacharjee's article comes when he interviews Stapel's parents. Elsewhere in the article, we find Stapel crediting them with instilling in him a culture of achievement: "That's what my parents' generation was like," he said, "You are what you achieve." (Interesting way to motivate fraud!) But Bhattacharjee writes
On our visit to Stapel's parents, I watched his discomfort as Rob and Derkje tried to defend him. "I blame the system," his father said, steadfast. His argument was that Stapel's university managers and journal editors should have been watching him more closely. [Stapel shook his head, "Accept that this happened."]
This demonstrates little care for the graduate students who are seriously harmed by the retraction of their joint publications.
What are we to learn from this? I suppose the content must take some of the blame. What are we to say of hypotheses like that supposedly proven by the Utrecht study? Are journal editors to blame? Only if you think that they should be looking at raw data. The Utrecht study is not replicable in the strict sense. (Of course, you could repeat the experiment, but what exactly would it show if you came up with negative results?)
As for his collaborators, was it simple naïveté, or something worse? Judge for yourself, from this story:
A colleague, Ad Vingerhoets, asked Stapel to help him design a study to understand whether exposure to someone crying affects empathy. Stapel came up with what Vingerhoets told me was an "excellent idea." They would give elementary-school children a coloring task in which half the kids would be asked to color an inexpressive cartoon character, while the other half would have to color the same character shown shedding a tear. Upon completing the task, the children would receive candy and then be asked if were willing to share [it] with other children.
Stapel cooked up some results that delighted Vingerhoets. While writing up the paper, however, he (V.) wondered about gender differences. But Stapel told him that the data had not yet been entered into a computer. "Vingerhoets was stumped. Stapel had shown him means and stadard deviations." He consulted a retired professor who asked, "Do you really believe that someone with [Stapel's]status faked data?" V. dropped the matter immediately.