It seemed like an egregious violation of academic standards.
A researcher forwarded us a data access agreement from the University of Oxford, in which Schedule 4 read as follows:
Data Users who make use of Licensed Data must cite the relevant primary original publications (Fairfax BP et al. 2012 Genetics of gene expression in primary immune cells identifies cell-specific master regulators and roles of HLA alleles. Nature Genetics 44, 502-510; Fairfax BP et al. 2014 Innate Immune Activity Conditions Effects of Regulatory Variants upon Monocyte Gene Expression. Science 343: 1246949). Authors utilising genotyping data are required to list the following individuals as co-Authors on any publications: “BP Fairfax, S Makino, JC Knight”.
Funding for the project was provided by the Wellcome Trust under awards Grants 088891 [B.P.F.], 074318 [J.C.K.] and 075491/Z/04 to the core facilities at the Wellcome Trust Centre for Human Genetics), the European Research Council under the European Union’s Seventh Framework Programme (FP7/2007-2013) (281824 to J.C.K.), the Medical Research Council (98082, J.C.K.) and the National Institute for Health Research (NIHR) Oxford Biomedical Research Centre.
In other words: You want to use our data? Sure thing, just cite our papers and make us co-authors. We showed this to a few scientists, all of whom found it unusual, to say the least. One used much more colorful language before the word “ridiculous.”
But to the Oxford team’s credit, they’ve decided to change the language now that we’ve brought scientists’ concerns to their attention. Julian Knight, the “JC Knight” in the agreement, wrote us this message in response to some questions about their rationale:
You raise an important point and in practice when the question of authorship has arisen, we have discussed this with the scientists involved using the data with a view to authorship only where there has been a scientifically appropriate contribution within a formal collaboration.
Looking again at this phrase within Schedule 4 of the data access agreement in light of your email, I agree that as written this was overly prescriptive and moreover is something we have not followed. We are therefore revising this to reflect our actual practice. The revised agreement will not require users to include as co-authors any of the currently listed individuals on the schedule.
We are fully committed to data access and have sought to do so, with all requests for genotyping data received to date resulting in data being released. We are a small group of scientists, in contrast to large Consortia, with recruitment, cell purification and data generation being essentially carried out by 3 individuals.
Ben Fairfax, the “Fairfax BP” in the agreement, followed up with a phone call. He told us that the agreement had been drafted a few years ago as the data were being developed, and that the intention was always to share the genotyping results but to make sure the small team of three were receiving some sort of acknowledgement. No one had ever raised questions about wording — and they have never actually insisted on co-authorship:
We have most definitely provided that data to people, and I’m sure it’s been published, and is in papers that are now in press, and we’re not authors but are in the acknowledgements.
While many groups were sharing data at the time, Fairfax is concerned that fewer are today. That makes it more difficult for other groups to try to reproduce the findings, and to collaborate. But he acknowledged that the agreement didn’t do what his team hoped it would:
There was an element of naivete. We should have rephrased it, most definitely.
And so they will. Fairfax tells Retraction Watch:
We don’t want to go on other people’s papers if we haven’t contributed.
Kudos to the Oxford team for acting swiftly.
Progress – the original Schedule 4 was flagrantly out of step with data management policies of the funders and rules for authorship.
My university has a datacentre (http://data.3tu.nl/repository/) through which you can make available datasets. These datasets are assigned a DOI, making them easily citable, e.g. following the standards suggested on http://www.datacite.org/whycitedata.
This makes it easy to share data (upload once, then just send people a link if they want your data and haven’t already found it) and to receive credit for your work (tell people to cite the dataset).
You would expect that the original paper was cited, as it describes how the data was collected and it is also good manners, as they did the work. It is not the first time I’ve seen a request/demand for authorship and that is definitely not OK., and if I see more I will let you know. Almost as serious and much more common is the demand for authorship when the data is privately held. It explains the large number of papers with huge numbers of authors in the medical world especially using genetic data.
I am a collaborator and author on the 2014 Fairfax et al paper. I wasn’t part of the design of the experiment, and wasn’t aware of the clause you mention, and would have argued against it had I been aware.
But I want to point out that Julian Knight and his group have been exemplary in releasing their raw data for other researchers – the data was released online immediately upon publication of the paper, with a simple application process that has never refused an applicant. Many other very similar projects have not shared data, or only shared it under collaboration agreements, or at least have had major delays in putting the data up. I am not saying that they should get massive praise for abiding by the principles that all scientists should follow, but I personally would be delighted if all the consortia I work with would release data as swiftly and completely as it was released for this paper. The section in the agreement was a mistake, but these guys are definitely on the side of good in the open data wars.