Imagine an important public institution, say, for the sake of example, the police.
Imagine that serious and specific accusations of misconduct have been made against a high ranking officer on a whistle-blower website. These have been picked up in the media. Although there is no suggestion that anybody has been physically harmed, those acts, if proved true, may have costed significant amount of public money and may have had severe consequences on the well being of many people and businesses. The media reports are also a concern because of the damage made to the public trust, essential to the police mission.
Imagine then, that the press release announcing the investigation says nothing of the potential consequences of those putative acts, stresses that the serious and specific accusations are in fact only anonymous comments on a website, indicates that the investigation procedure will be completely opaque to public scrutiny with an undefined timeline, and, finally, concludes with an entire paragraph devoted to the glorification of the work of the accused (and indeed highly qualified and otherwise commendable) officer.
This is, of course, science-fiction. The police would not adopt such a course of action because they know full well that this would only disqualify the investigation and do nothing for the prestige of the (maybe wrongly) accused officer.
This is however very close to what two major scientific institutions have just done.
Last week, the CNRS and ETH Zurich published press releases announcing investigations into allegations of scientific misconduct. Retraction Watch, covering these press releases, “found some of the language in the announcements puzzling. Call us old-fashioned, but generally it’s a good idea to actually do an investigation before saying that “the studies’ findings are not in doubt.”
True, especially in the current context. The scientific enterprise is suffering from a reproducibility crisis. One of the drivers of this crisis is the lack of publication of negative results which is itself a combined consequence of the publication system and of the methods of evaluation of researchers based on where they publish rather than what they publish [I got more (serious) congratulations for my April fool spoof paper in Nature Materials than for my PloS One paper published the day after].
Scientific institutions such as the CNRS and ETH Zurich should be leading the way to change those practices. They should not, at the onset of an investigation, rule out that “studies findings” (maybe) based on data manipulation are *not* in doubt. Instead, they should set firm plans to test how much of this body of work is solid and how much is not. Surely damages to human knowledge and the integrity of the scientific record should be major sources of concern, yet they barely feature in the press release. It would seem that the main (and almost exclusive) concern related to accusations of scientific misconduct is the damage done to the accused until proven guilty/innocent. That concern for individuals is warranted. It should not stop to the accused. If the charges are proved correct, then there are probably a number of other individuals, less prominent and well-known, who have directly suffered to different extent and for whom redress is unlikely to ever happen: the reviewers of papers and grants who have wasted their time on “diagram/chart” which had been “manipulated”; the competitors which may not have had access to such impressive data and therefore would have failed with their papers and grant applications; the PhD students who might have spent three years trying to reproduce some of these experiments without success [you would not have heard about this since negative results are not published] and may have left science in disgust at the end of the process, etc.
If you’re interested, see also this conversation about the CNRS press release via Twitter (with critical contributions from @b_abk6 and others).
#CNRSvoinnetPR in 140 ch “world-leading researcher with unquestioned track-record irritated by anonymous allegations: CNRS investigates”
— Raphael Levy (@raphavisses) April 15, 2015
and of course, PubPeer