As a post-doc, I used to assume that published work, particularly in high-ranking journals, was something you could rely on. Now the feeling is changed. A commentary recently appeared in Science (1) voiced a question that I have been asking myself since sometimes: how much of the published scientific work can nowadays really be trusted? The progress and success of science are intrinsically based on building new findings on the bases of older ones (“standing on the shoulders of giants”, as Isaac Newton put in his letters). Can science survive without this trust? Is the problem really that bad? What are the causes?

One indication that indeed the rate of scientific misconduct is increasing comes from the observation that the percentage of articles retractions is constantly increasing, due for the most part to scientific misconduct and in particular to fraud (2). Although retracted articles are still rare (about 1/1000), the problem rate is certainly much higher as only a small percentage of flawed articles are indeed likely to be retracted (3). The thought is disheartening, as it means that there is a lot out there that really cannot be trusted.

Interestingly, and Editorial on Infection and Immunity recently demonstrated a strong correlation (retraction index, p<0,0001) between retraction frequency and the journal impact factor (3), and this is particularly true for retractions due to fraud (2). Thus, one of the main causes for the rising rates of scientific misconduct may well be the increasingly high pressure to publish more and in higher impact journals, as publications number and impact factor are getting one of the main evaluation factors for career progression and grant assignment. In a world where independent positions and the success rate in grant applications are constantly shrinking, the competition, which is of course an essential driving force, is perhaps becoming overwhelming and counterproductive.

We cannot do much unfortunately to increase funding levels. Better and less shortsighted planning for funding should be pressed on. Also, we should push for an evaluation of scientific production more concerned about quality than about numbers. Indeed, famous scientists, editors of important scientific journals and prestigious scientific societies such as EMBO have recently launched an initiative (the San Francisco Declaration on Research Assessment, http://am.ascb.org/dora/) aiming at containing the indiscriminate usage of the journal’s impact factor in assessing grant applications and scientific carreers, among others.

We need to reinforce at every level the concept that cheating is not an acceptable behavior. At the same time, the amount of pressure post-docs and students are exposed to in the labs is important: as in biology, neither too little nor too much! Also, let’s try to preserve and communicate to everybody our delight for discovery. A science all business/no fun can be a gate to fraud.

Scientific misconduct is also a matter of ethics, and has treacherous boundaries. Perhaps not so many published data have been truly faked, but things like setting aside a control that does not fit, or selecting a particular set of data against others that do not make sense (to one’s model, anyway), can really be considered mere simplifications of an experimental system that is too complex, or could they dangerously near misconduct?

References
1. Couzin-Frankel, J. Shaking up Science. Science 2013, 339:386-389
2. Fang et al., Proc. Natl. Acad. Sci. USA 2012, 109:17028-17033
3. Editorial, Retracted Science and the Retraction Index. Infection and Immunity 2011, 79:3855-3859