Comment The perverse incentives in biomedical research (Score 2) 167
My experience is that most of the work is done by the first one or two authors under the direct supervision of the last author (there are exceptions when the lab is very big and the PI has delegated most of the supervision to postdocs or staff). Generally, the corresponding author on the paper bears much of the responsibility for the data being published under the assumption that he is supervising the research and is intimately involved in analyzing the results and writing the paper. Many journals now require a statement, which briefly outlines the authors contributions. Having said that it is not unheard of that a student or a postdoc will manipulate data and the PI in his willingness to prove a hypothesis will not be overly critical.
The deeper problem is that there is a huge pressure in the biomedical field to publish often. The PI will not be able to receive grants unless he/she has demonstrated a track record of productivity. If he/she doesn't get grant funding he/she will not be receiving full salary and will not get tenure. At the same time the competition for funding is furious. The percent of applications being funded by NIH are in the low teens and for some NIH institutes they are under 10%. So not only you have to publish, but you have to publish more and in better journals than 90% of the people in the field to be competitive. All this puts huge pressure on the PIs to publish. Few of them publish rubbish and some resort to fraud. Students and postdocs are under similar pressure to be "productive" not only from their PIs, but also because their future prospects depend on the research they publish.
The irony is that the current situation is to a large degree caused by the expansion of the NIH budget in the past. Public and private research institutions rushed to build lab space, recruit scientist and train students to take advantage of the NIH grants (this expansion still continues!). The incentive for the institutions is that they get 40% or more (up to 100%) on top of the grant award as an overhead. So a typical $250K per year grant from NIH will pay directly to the university at least $100K per year in overhead in addition to sponsoring the PI and staff salaries from the direct costs. The NIH budget, however, did not continue to grow rapidly after the initial jump during president Clinton's time in office and has actually shrunk in the past year. The result is that now you have a large number of scientist desperate for grant money and not enough grants to fund even a small fraction of them. The current incentives do not reward the quality of the research, but the speed by which it is done and its quantity. This is a very perverse situation. It also creates a negative feedback loop where the poor quality of the published data prevents people from defining valid hypotheses or identifying viable lines of research. As a result they waste time and are subject to even more pressure to publish junk.