The Journal of Health Psychology has just published an extraordinary pair of papers that call for a new inquiry into a 30-year old case of probable scientific fraud.
According to Anthony J. Pelosi, author of the main paper, the case was “one of the worst scientific scandals of all time” and yet has never been formally investigated. The journal’s editor, David F. Marks, agrees and, in an editorial, also calls for the retraction or correction of up to 61 papers.
The scandal in question is one I had never heard of before, but the facts are jaw-dropping. Beginning in 1980, a Dr Ronald Grossarth-Maticek reported that he had discovered a cancer-prone ’emotionally repressed’ personality. Someone with this personality type was, he claimed, at very high risk of later developing cancer. A second personality type predicted ‘internal diseases’, such as stroke and hypertension. Even more remarkably, Grossarth-Maticek said, a brief course of psychotherapy was enough to virtually eliminate the excess risks.
Despite the fact that Grossarth-Maticek was claiming to have found a way to prevent most cancers, his work was largely ignored. Then, at the end of the 1980s, he started a collaboration with Prof. Hans Eysenck, of the Institute of Psychiatry in London (now part of King’s College London).
Eysenck was an eminent and extremely influential psychologist in Britain, perhaps the most prominent of his era, so the papers that Eysenck and Grossarth-Maticek published together around 1990 were widely read. Eysenck had no role in the data collection of any of these studies, but his name was an endorsement of their credibility.
The attention Grossarth-Maticek attracted was not positive, however. A barrage of harshly critical responses appeared, starting in 1991 with a broadside from statistician Bernard H. Fox and followed in 1992 with a piece by a young Anthony J. Pelosi and Louis Appleby. Many more critics soon joined in.
The gist of the criticisms was that Grossarth-Maticek’s results were simply too good to be true. Here’s one of the many remarkably strong effects found in Grossarth-Maticek’s data, as discussed by Fox (1991):
This table represents the most important discovery in medical science since penicillin… if it were true. According to these data, Dr. Grossarth-Maticek was able to predict, with virtually perfect certainty, who would get cancer in the next 10 years.
Out of 763 people who were classed as not cancer-prone, not a single one had developed cancer 10 years later. Of those with a cancer-prone personality type, 159/172 or 92.4% had cancer (or had already died of cancer) 10 years later. All this based on a pen-and-paper interview measure of personality.
As Pelosi puts it in the new paper, this result, along with the rest of Grossarth-Maticek and Eysenck’s work, represents “what must be the most astonishing series of findings ever to be published in the scientific literature.” It’s simply very difficult to believe that these kind of massive effect sizes are real.
The critics also dug up many inconsistencies and problems in Grossarth-Maticek’s data, such as evidence of duplicated questionnaire responses and admissions of ‘corrective’ data manipulation:
It’s fair to say, though, that these data anomalies were just the icing on the cake. The main reason that people were skeptical of Grossarth-Maticek and Eysenck’s results was that the results were unbelievable.
The controversy rumbled on for a few years but by the late 1990s, it had petered out, no formal investigation ever having taking place. Eysenck died in 1997. Pelosi calls for an investigation to happen now – better late than never. Marks, in his editorial, says that he has addressed open letters to the British Psychological Society (BPS) and King’s College London requesting an inquiry. He also calls for the establishment of a ‘National Research Integrity Ombudsperson’.
In a subsequent post I’ll look into more detail at Eysenck and his motivations in this affair, but here are my initial thoughts on the new papers and the call to investigate.
I think an investigation is needed and should have happened long ago. King’s College London bears the obvious responsibility for this, as Eysenck’s institution. I’m less sure if the BPS are obligated to investigate, but in the spirit of good science, they should do, or at least assist King’s. (Both organizations have previously declined to get involved.)
It might be said that this is all ancient history now, and there is no need for an investigation after so long, but I think this is entirely the wrong attitude. If anything, the fact that these frankly bizarre results are still in the literature (and, as Pelosi points out, still being cited) 30 years later makes the scandal even worse.
What exactly is Eysenck accused of? Eysenck didn’t collect the data in question, so he probably can’t be accused of data manipulation as such. The question is whether he willingly published data which he knew to be unreliable. I would say that he must have known, and certainly should have known, that the data were not credible. He was one of the world’s leading academic psychologists. A young grad student handed a dodgy dataset by a senior professor, he was not.
King’s College may argue that they have no responsibility to invesigate Grossarth-Maticek because he never worked for the Institute of Psychiatry (although he claimed to, in some publications). Nonetheless, even if they only set out to investigate Eysenck’s actions vis a vis his work with Grossarth-Maticek, they will have to consider events that happened before the two met, because this was when the data were (allegedly) collected. I think any inquiry into Eysenck (if it happens) must encompass the question of the origin and validity of the data.