We apologize for what must seem like a constant drip, drip, drip, but we have three more retractions from Diederik Stapel to report, all from the European Journal of Social Psychology.
The first, in order of publication date, involves a 2006 paper by Stapel and co-author Camille Johnson — whose name has appeared on several other of Stapel’s retracted papers — titled “When nothing compares to me: How defensive motivations and similarity shape social comparison effects.”
The notice reads:
The following article from European Journal of Social Psychology, “When nothing compares to me: How defensive motivations and similarity shape social comparison effects” by Diederik A. Stapel and Camille S. Johnson, published online on 26 September 2006 in Wiley Online Library (wileyonlinelibrary.com), has been retracted by agreement between the co-author, the journal Editor in Chief, Tom Postmes, and John Wiley and Sons Ltd. The retraction has been agreed following the results of an investigation into the work of Diederik A. Stapel (https://www.commissielevelt.nl/). The Levelt Committee has determined that this article contained data that was fabricated by Diederik A. Stapel. His co-author was unaware of his actions, and not in any way involved.
Then comes the retraction of a 2009 paper, “Judging the unexpected: Disconfirmation of situation-specific expectancies,” by Stapel and Marret K. Noordewier:
The following article from European Journal of Social Psychology, “Judging the unexpected: Disconfirmation of situation-specific expectancies” by Marret K. Noordewier and Diederik A. Stapel published online on 1 January 2009 in Wiley Online Library (wileyonlinelibrary.com), has been retracted by agreement between the co-author, the journal Editor in Chief, Tom Postmes, and John Wiley and Sons Ltd. The retraction has been agreed following the results of an investigation into the work of Diederik A. Stapel (https://www.commissielevelt.nl/). The Levelt Committee has determined that this article contained data that was fabricated as supplied by Diederik A. Stapel. His co-author was unaware of his actions, and not in any way involved.
Finally, the journal is retracting a 2010 article by Stapel and Joris Lammers, “Racist biases in legal decisions are reduced by a justice focus”:
The following article from European Journal of Social Psychology, “Racist biases in legal decisions are reduced by a justice focus” by Joris Lammers and Diederik A. Stapel published online on 1 December 2010 in Wiley Online Library (wileyonlinelibrary.com), has been retracted by agreement between the co-author, the journal Editor in Chief, Tom Postmes, and John Wiley and Sons Ltd. The retraction has been agreed following the results of an investigation into the work of Diederik A. Stapel (https://www.commissielevelt.nl/). The Levelt Committee has determined that this article contained data that was fabricated as supplied by Diederik A. Stapel. His co-author was unaware of his actions, and not in any way involved.
The lone-wolf narrative here underscores the disheartening fact that individual fraudsters will always find a way to escape detection. It also throws into high relief the risks of the oft-cited idea that each author is responsible for particular parts of the paper, but only the corresponding author is responsible for the entire thing. That may be the only reasonable way to go in many fields with half a dozen or even dozens of co-authors, but in some fields — including psychology — maybe co-authors should ask to see the original data as a matter of course. We’d welcome thoughts on that.
Hat tip: Clare Francis
For too long, psychological researchers have used a modus operandi in which a sole researcher conducts all of the statistical analyses out of view of others (even the co-authors). This practice would be bizarre in commercial aviation and accountancy and so one would hope that psychologists have learned their lesson after the recent series of misconduct cases. Just like in aviation co-authors should at the very least see the data and preferably conduct an independent analysis of the data. In such a co-pilot model both fraud and honest error are greatly reduced.
The more I think about it, the more I like that idea. It has at least one big advantage: it avoids imposing new compliance costs on people like reviewers, editors, and academic institutions who can’t really do the job well without being unreasonably intrusive. The job that reviewers & editors might be able to do, if it were culturally acceptable in academia, is to tell the author, “Good paper, but this is the kind of study where we’d really like to see a co-author.” That approach might involve some culture change, but not too much. Is this the sort of thing you had in mind?
In the same way, teaching professors and graduate advisors might get in the habit of telling students, “Yes, that’s a provocative paper, but of course its only an unreplicated, single-author study — so who knows?” That may sound awful, but consider some of the other scholarly folk wisdom passed on as rules of literature triage. Its more relevant than many such.
Every paper he’s authored or coauthored should be retracted at this point.
No kidding… He should be in jail.
A junior researcher could make a nice career out of actually conducting the experiments Stapel outlined in his papers. It would be a tremendous service to the field to know what actually happens.
Psychology has no special status. The other author(s) might have asked to see the original data, but if the original data was fabricated then what?
It is not easy to fabricate data that look real. Uri Simonsohn has just posted his paper on his fraud detection methodology. He started out with the descriptive statistics and the raw data really established that the data were suspicious.
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2114571
And now that this paper is published (at least as some sort of tech report), all fakers have to do is add this constraint to the way they fake data! Looks like this is they way Uri will try to get tenure (not sure they even have tenure The Wharton School). Hope it works!
He had to publish it. The idea of him having some secret method that is above scrutiny is ridiculous.
I agree with the comment by Jon Beckmann. The Simonsohn paper can be used as a handbook for fabricating data.
While it’s true that a fraudster could just fake the raw data, I think a culture in which all authors check the raw data would still prevent a lot of fraud. For example, the raw data is likely to contain a set of dates on which the data was collected. If 5 coauthors, who all know the data collector personally, were to browse over those dates, they might well spot things that don’t make sense – a data point collected on a day when everyone was away at a conference, or when one of the coauthors was in a meeting with the guy.
That kind of thing. It’s a lot easier to fool a stranger than to fool your friends.
“I think a culture in which all authors check the raw data would still prevent a lot of fraud. ”
Well, with that logic, a culture in which each author collects their own data and actually replicates the study before publication would be even better! No offense, but, NS, obviously you don’t have a real academic job as you don’t seem to have a realistic view on the time academics have to publish at the rate that is expected of them by institutions.
No offence taken… because you’re not talking about my suggestion – see the original post:
“Maybe co-authors should ask to see the original data as a matter of course.”
I was just saying that yes, that would help prevent fraud. Which it would. I didn’t say it would be feasible (although it would be maybe an hour’s work, per author, per paper, in most cases, so yes it would.)
Why not just make all authors responsible for the paper instead of just one as a matter of course? Could that provide a check in the system? I can’t help feeling that some co-authors who are at least negligent wrt to their involvement in fraudulent studies and at worst complicit, get off scot-free.
All authors ARE responsible for the paper they collectively publish. I do not buy into the scenario that each of Stapel’s coauthors was clinically gullible. But this scenario limits the damage, so let’s pretend it is actually true.
Apparently all the authors aren’t responsible otherwise there would be some sanction, note the byline in the retractions: “His co-author was unaware of his actions, and not in any way involved”. Maybe this is a problem that the journals could pretty simply do something about. If your name is on the paper as one of the authors and you were NOT involved in the fraud then it is up to you to prove it – change the burden of proof. At another level, how is that someone can claim authorship of a study when they don’t know what is going on?
“All authors ARE responsible for the paper they collectively publish. I do not buy into the scenario that each of Stapel’s coauthors was clinically gullible”
Perhaps an exception to this would be Stapel’s students ? They may have not received proper training or a proper education to fully grasp what was going on, and what their responsibilities (should they have any) were.
I agree with chirality. In addition to the obvious fact that one who authors a paper has to answer for what’s in it, most journals ask that all authors sign a form attesting that they have read and agree with what has been submitted. Most also require that each author’s contribution to the work be specified.