A researcher in Germany has lost a paper on determining the sex of panthers after a now-former colleague objected to his use of data.
Here’s the notice for “A method to assign sex to leopard Panthera pardus specimens using quantitative cranial data:”
This article has been retracted at the request of the Editor-in-Chief. It was established that the author of this article did not have permission to use part of the data and therefore had no right to publish.
Here’s the original abstract:
Sexual size dimorphism of skulls is well described in leopards Panthera pardus (Linnaeus, 1758) and other felids, but which cranial measurements should be used for reliable sex predictions under significant geographical variation is unknown. In this paper, I describe an approach on using the geometric mean (GM) of cranial measurements as a reliable predictor of sex of leopard skulls. The dataset included 49 male and 25 female skulls belonging to leopards from throughout the Middle East from which 22 cranial measurements were taken. The logistic equation has demonstrated its statistical power to predict sex from GM. A nonparametric 95 % confidence interval of GM is 57.54–67.24 mm in females and 68.99–77.53 mm in males. Nine scenarios of separating sexes in the transition zone of GM = 60–68 mm are applied and compared using the integer cutoff values and receiver operating characteristic curves as the model validation tool. The GM = 63 mm is defined as the reliable cutoff value so that the skulls <63 mm are females and those ≥63 mm are males in almost 80 % of cases. If the lower cutoff GM = 60–62 mm is taken, then female skulls are correctly classified less often (in 44–64 % of cases) than males (83.7–89.8 %) because the smaller skulls assumed to be females often belong to smaller males. Taking few measurements may inflate GM and misclassify females as males. Therefore, smaller skulls should be measured for as many as possible variables in order to ensure their accurate sexing.
The journal’s editor, Andreas Schmidt-Rhaesa, tells us:
We learned about the background in this case directly from a person, whose data were used.
The paper’s author, Georg-August-Universität Göttingen postdoc Igor Khorozyan, described what happened:
I and an Iranian scientist were the co-authors of the paper. Just before submission, we had a serious conflict between us which made further collaboration in paper writing impossible. I left his data in the paper and acknowledged his name in Methods and Acknowledgements sections, but removed from co-authors.
But perhaps a leopard can change his spots. Khorozyan continued:
Later after retraction, I excluded his data from the analyses and completely re-written the paper with new data, having obtained exactly the same results. Now it has been reviewed in one of the journals.
Hat tip: John Hutchinson
This brings up a sometimes very difficult topic. If you do collaborative work, and then the relationship goes bad, do you have the right to block publication of the results?
The other question is, can you eject an author due to non-cooperation and still use the data they produced?
Neither question has cut and dried answers.
Another question is whether it is acceptable to re-do the work of an ex-collaborator and then exclude them? You might say ‘yes’, but if they qualified to be an author I think you cannot exclude them simply by repeating the physical work they did.
That’s a difficult question but one that any researcher will face in their careers sooner or later. Especially when the field encourages collaborative work.
I personally think that a collaborator has the right to block publication. Of course, it would be good to discuss it first but what happened in this story wasn’t right either. That is, to kick someone out from the author list and move them as an acknowledgement. Certainly, the opportunity to split the work into two lesser papers should have been discussed. Of course, if the two authors hate each other, the idea of sitting down to talk is perhaps impossible… The term “block” is ambigious; each one is blocking the other by not choosing to work it out peacefully.
As for “re-doing” the work and then exclude them, I would say “no”. It isn’t the same since the first time it was done with the collaborator, you had no idea it would work. So, there was an amount of risk involved. Doing it again knowing that it will work isn’t the same. Add to the fact that the author that was kicked out probably contributed intellectually and not just the data. Can you redo the work by “forgetting” all the conversations you’ve had? Perhaps impossible…
Instead of all those good questions you’ve asked, I think the lesson to be learnt is to pick your collaborators as carefully as you would pick your friends!
The co-author only contributed data and never expressed willingness to discuss the paper or be involved in data analysis. So, the entire paper in its original and re-done version was written solely by myself. Re-doing of the work was also risky, as I had to deal with a smaller sample size and was not sure at all if results would be the same.
I think the rules are quite clear in most cases:
There are two issues at stake:
Ownership of the actual data… Usually the person/group who employs the data collector owns the data. Often times a university will give that to a PI in their contract. A pharm company will retain ownership. In most countries this is owned material, and the owner has absolute right to block publication. This is independent of authorship status. A data owner may not qualify for authorship, but she can block publication. Similar rules apply for writing.
Second, authorship. Typically, authorship only goes to people who meet all criteria. If you have a falling out with someone, simply writing the paper from scratch and not showing it to them should elimate them from qualifying as an author. The only issue here is whether their idea has been plagiarized… But I suspect that in a case like this, it may not have been.
Without knowing this case at all, I think there are many ways a project and authorship can go wrong. I once developed a project in biophysics to measure VLA4/VCAM1 and SDF1 with an atomic force microscope (AFM) on mouse lymphocyte-like cells – i.e. I had the idea, developed and pursued the project over years and even invested a lot of effort to get a well known immunologist (who once was a postdoc at Harvard) to cooperate on this project with me an my other collaborator, especially to be part of the same project with just human lymphoma cells. Things went wrong, but to my biggest surprise the others published this work and didn’t even ask me to be last author as discussed before. The editor – in this case – did not retract the paper, but the university lab does not mention one of the authors as former PhD student, nor includes the paper in the publication list.
But the labs get millions of funding to proceed – as I find – not very successfully.