|Misconduct in science
||[Jul. 11th, 2011|10:16 pm]
Who'da thunk it that scientists are just as prone to cheating as people in other professions? A Nature article(non-paywalled version) found that around a third of around 3000 respondents admitted to having engaging in some kind of scientific misconduct, ranging from falsified data to the comparatively benign sin of not keeping proper records related to research projects.
Looking at the percentages of people confessing to various behaviours puts me strongly in mind of one of Dan Ariely's studies, documented in Predictably Irrational, where he gave people tests with monetary rewards for correct answers, and then progressively made it easier for each subsequent group to cheat. His overwhelming conclusions were a) that people will cheat if you let them, but only up to a certain point, and b) that removing money from the equation by, for example, giving the participants tokens that were exchanged for money just a minute or two later, greatly increased the incidence of cheating because money is Serious Business.
We can see the same kind of cheating going on here, with relatively large numbers of respondents admitting to some kind of misconduct, but most of it being of the minor easy to rationalise variety and further away from the parts of their work that really pay their salary (funding and publishing). For example, 7.6% of all respondents admitted to circumventing minor aspects of human-subject requirements but only 0.3% circumvented major aspects. Similarly, almost no one failed to disclose involvement in firms whose products were based on their research (0.3%) but around 15% allowed funding sources to pressure them into changing their design, methodology, or results, despite the same type of objectivity being called for.