On scientific misconduct

Dear Sir, If we are to understand the relationship of science and society, it is vital that we also understand the extent to which the profession of science is characterized by integrity (Bornmann, 2013). Shady practices go on at the fringes of the profession: For example, some scientists have been bought by the tobacco companies and oil companies to produce research “findings” to order (Michaels, 2008; Oreskes & Conway, 2010). Scientists produce fraudulent results for various reasons: personal advancement; pressure to produce results; poor oversight of students, trainees, and subordinates; or ignorance of research rules and norms. Publishers can be out to make money rather than deliver scientific truth and they can always find authors whose motivations fit their ambitions (Bohannon, 2013). But these cases may be news-worthily scandalous precisely because they are atypical and do not characterize the profession of science as a whole (Merton, 1942). After all, it would not make news were we to discover that a used car salesperson had sold a vehicle knowing it to be less perfect than it appeared because we know that that profession is the icon of dishonesty (or, at least, once was). It is a good thing that social scientists continually try to estimate the honesty of the scientific profession. Here, however, we point to a peculiarity of the method used in many of these studies and ask if it can deliver the right answer to the question. The predominant method is to survey scientists and ask if they have encountered scientific misconduct in their careers (according to their own behavior or the behavior of colleagues). The results are then presented as a percentage of the proportion of scientists asked who answer “yes.” A high percentage is taken to indicate a relatively dishonest profession. For example, the first metaanalysis of surveys on research misconduct including 18 studies shows: “A pooled weighted average of 1.97% (N = 7, 95% CI: 0.86–4.45) of scientists admitted to have fabricated, falsified or modified data or results at least once—a serious form of misconduct by any standard—and up to 33.7% admitted other questionable research practices. In surveys asking about the behavior of colleagues, admission rates were 14.12% (N = 12, 95% CI: 9.91–19.72) for falsification, and up to 72% for other questionable research practices” (Fanelli, 2009). But the conclusion that this is a dishonest profession does not follow. Thus, if a survey asks people whether they know of anyone who has ever told a lie and every respondent says “yes,” this does not mean that the society surveyed is characterized by dishonesty because nearly every person has lied about something at one time or another in their lives, even if it was only to protect another from a truth they are better not knowing. What would be needed determine whether this was a dishonest society would be the proportion of lying acts among all the acts ever undertaken. One lying act per person would mean that everyone had encountered a lie (at worst their own), whereas one lie per person is a very small number and characterizes an essentially honest society. Of course, one murder per person in a society is not a small number and should a survey discover this then we would certainly be justified in saying this was a murderous society. Thus, to make sense of the existing surveys, we first need to know what it means for a society if some of its members cheat rarely or sometimes. We need to know how surveys based on scientists would relate to a survey that would be based on scientific acts. It may be impossible to invent a method based on acts but at least the problem should be acknowledged: The number of scientists who have encountered dishonesty does not map onto the honesty of the profession in any straightforward way. The needed discussion turns on: (a) how science compares to other professions in terms of the reporting of experiences of misconduct; (b) how many reports of encounters with scientific dishonesty constitute a dishonest science; (c) the variation in the way reported negative experiences signify a serious problem where different kinds of acts are concerned (e.g., science, lying, politics, banking, car sales, murder). “a” has an empirical solution; “b” and “c” are much more matters for sociological and philosophical analysis.