The editor received correspondence from a third party suggesting that a paper published in 2005 in the journal by four co-authors contained suspect data. The suspicion was based on the observation that in three separate figures all error bars were identical. Indeed, the third party subjected 32 publications from the same group of authors to independent statistical appraisal and concluded that 21 (66%) of the papers were suspect, in comparison with none of a large control sample of papers from other institutions publishing in the same journals (detailed confidential report supplied). Inspection of the figures did indeed indicate a consistency of error bars that were highly unlikely to have been generated by chance variation, suggesting that there is a prima facie evidence for concern about data fabrication.
1.
Given that multiple journals appear to be implicated, what sort of coordination between editors, or within/between publishers, would offer the most effective way forward?
2.
What should be our next action with regard: to (i) the authors of the challenged article and (ii) the third party?
Although multiple journals appear to be involved in this complicated case of suspected data fabrication, the general view was that the journal should pursue its own case in the first instance. Due process should be seen to be done and there is always the danger that waiting for a concerted effort from a number of journals could take several months or even years. Hence the advice was that the editor should write to the authors requesting the raw data, stating that it had been brought to his attention that the error bars were identical in three separate figures. This should be done without making an allegation or accusation of data fabrication but giving the authors a deadline of one month to respond. If no response is received, a second letter should be sent to the authors informing them that if no satisfactory response is received then the authors’ institutions will be contacted and informed of the situation.
Other advice offered was to have an independent statistical adviser review the paper, independently of the third party.
A number of the journals involved are published by one publisher who was represented at the meeting. The publisher agreed to encourage its editors to write to the authors requesting the raw data. However, it was agreed that it would be best to wait and see if this editor receives a response before the other editors took any action.
The Council of Science Editors (CSE) has been contacted as many of the journals are also members of the CSE. The CSE are very interested in pursuing this case through their governing body and are also eager to form some type of collaboration with COPE
The editor received a response from one of the authors in which the author stated that he had re-evaluated the results of all of the parameters and found an error in one of the figures. The raw data and covering note have been forwarded to the third party who raised the suspicion in the first instance and also to an independent statistician so that they can analyse the data and evaluate the author’s comments in the covering letter. The editor and publisher are awaiting the response from the third party and independent statistician, and will not act until such time as they have received this information.
The publisher in the meantime is also pursuing the matter.
Further update (August 2007)
The editor has received a response from the third party and the independent statistician and is in the process of reviewing the information.