In a recent Retraction Watch guest post on the “Eysenck affair,” James Heathers notes the extraordinary possibility that as many as 61 Hans Eysenck publications might be retracted. I believe this figure is a significant underestimate.
This reckoning has been a long time coming. The issues surrounding Eysenck’s 1980s/1990s collaboration with Ronald Grossarth-Maticek and their unbelievable results linking personality to health outcomes have been known for decades. Many eminent researchers, including Tony Pelosi and Louis Appleby, had lined up to criticise this research even while it was still ongoing.
In 2010, I published a lengthy biography of Eysenck, Playing with fire, that detailed the context for this collaboration. In the final full chapter, I explained what prompted Eysenck to team up with this outsider figure and laid bare the extent of Eysenck’s deep and longstanding relationship with the tobacco industry. It was backed by extensive archival research and interviews with key players (including two days with Grossarth-Maticek). I had hoped it would provoke a reappraisal and remedial action. But the impact was minimal.
More recently, Tony Pelosi published a sharply critical article with strong editorial support from David Marks. Marks called for an inquiry and compiled a list of 61 suspect works. Some action has resulted, but what the final outcome will be is still uncertain.
Kings College did conduct an internal inquiry earlier this year. The Kings inquiry flagged 26 joint-authored papers as “unsafe,” stating that they would seek retractions where possible. But for reasons that remain unclear, they did not include publications where Eysenck was the sole author.
The Kings inquiry decision to exclude Eysenck’s sole-authored publications was criticised by Marks, who suggested that it might be an attempt to lay the blame on Grossarth-Maticek, who had improperly listed his affiliation with the Institute of Psychiatry (now affiliated with Kings) during part of this collaboration. Whatever the case, it is a decision that is hard to defend. This should not be about institutional and personal reputations; it should be about the reliability and integrity of science.
To my mind, the Kings decision appears to be neither consistent nor comprehensive. But then again, Eysenck was always a hard man to pin down. Some background is necessary to explain why this is the case.
In the early 1980s leading up to his collaboration with Grossarth-Maticek, Eysenck published a series of articles and chapters on stress, personality and disease. Much of this work cited Grossarth-Maticek‘s published research, and not uncritically. Eysenck did not yet present Grossarth-Maticek’s data as the basis for his work, but that would rapidly change.
Soon after he retired in 1983, Eysenck visited Grossarth-Maticek in Germany. Their first joint effort was a 1985 conference paper. However, 1986 appears to be the pivotal year as far Eysenck’s published work went, when he first began to present data from Grossarth-Maticek’s research. From 1987 on, a steady stream of papers and book chapters appeared, some jointly authored by a cast that included Eysenck and Grossarth-Maticek. However, other works were authored by Eysenck alone.
See, for example, this 1987 paper in the Journal of Anxiety Disorders. In this paper, Eysenck presented hitherto unpublished data from Grossarth-Maticek drawn from a 1986 joint conference presentation. This data is the key component of the paper’s novelty, for its knowledge claims. And from my knowledge of this literature, such is the case for virtually all the other sole-authored Eysenck papers and book chapters that Marks identified as potentially retractable.
Remember that these are papers and chapters that the Kings inquiry ignored. Yet they depend on the same dubious data. They are just as “unsafe” as the jointly authored papers. There is no sense of diminished responsibility that is sometimes (rather problematically) given to the “unwitting” co-authors of dubious papers. Official guidelines do not generally allow such charity: if your name is on the paper, then you are responsible for its contents. However, the Kings inquiry appears to imply diminished responsibility where Eysenck was the sole author. This seems inexplicably inconsistent. Even if Eysenck didn’t gather the data he was presenting, there is no ambiguity in his responsibility.
Eysenck himself claimed he did the due diligence with Grossarth-Maticek’s data. But from my research, I know that Eysenck knew there were problems. It was impossible to precisely determine what kinds of malpractice had taken place. Much of the data might have been genuine at some point, but various forms of manipulation had to have transpired. At one stage, Grossarth-Maticek’s long-time statistician Hermann Vetter complained in print that the figures he was being fed were being doctored in some way. However, in the context of potential retractions, it matters little how it was done or by whom. Both Grossarth-Maticek and Eysenck must be seen to be responsible for the work that bears their names.
Then there is the issue of comprehensiveness. Marks’ list is significantly incomplete. I have identified at least 27 more potentially retractable papers and chapters. That would take Eysenck’s potential retractions to 88, vaulting him into the Retraction Watch Leaderboard stratosphere across all sciences, not just psychology. Some of these additional pieces are co-authored articles and chapters (meeting the Kings criteria), though many are sole authored pieces by Eysenck (including the 1987 journal article mentioned above). This expanded list gives a much better estimate of the extent of the problem. I will be forwarding it to the relevant bodies. For now, the focus on Eysenck does not take into account Grossarth-Maticek’s other potentially problematic publications.
Finally, there is the issue of other work that depends on this questionable data; in particular, the meta-analyses that include this data. These include this highly cited 2008 meta-analysis assessing whether stress-related psychosocial factors contribute to cancer incidence and survival. It is dangerously misleading to use the Grossarth-Maticek data in such summary analyses, as has been pointed out. The inclusion of this data distorts what we think we know about the relationship between personality and health, falsely inflating what is at best the very modest contribution of the former to the latter.
In sum: if science is to be truly self-cleaning, the job needs to be done properly – consistently and comprehensively.
Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].
I just wanted to add a note of thanks to Buchanan for his work on Eysenck.