"We've had a couple of papers in Science in the last four or five years that plainly involved scientificmisconduct, ultimately discovered on investigation and publicised," he told reporters last month."We were asked questions about those papers and I said, editorially, each time that there is no way that the peer review system can be made proof against misrepresentation of data."It has been suggested that on particularly contentious or high profile research, reviewers should do morethan just read through manuscripts to check they add up; could materials also be submitted for independentanalysis and verification?
"You can only assess the science in terms of what is in the paper, and the data looked accurate," Dr StephenMinger, director of King's Stem Cell Biology Laboratory, UK, told the BBC News website."I know at least one person who reviewed the original paper, and they were convinced the data was real;and short of going to the lab, and physically inspecting the data yourself, and saying 'I want to see the cells,I want to do the DNA analyses myself' - you just cannot do that physically."On that, Dr Kennedy is in agreement. Insisting replication of data by an independent group be part of thereferring process would be a nightmare, he believes."I think that to install a procedure by which replication by a third party was required for acceptance of a manuscript would imposesimply an enormous load on our readership and the scientificcommunity," he said.Colin McGuckin, professor of regenerative medicine at NewcastleUniversity, UK, commented: "We don't want to become bureaucratic, because that will hold up everything. But sometimesas a scientist, your biggest critics will be in the organisation youwork at. They're the ones you should look to for advice."The Newcastle stem cell expert said there should be internal review procedures within universities andscientific institutions to evaluate research before it is submitted to academic journals for peer review.Freelance publications consultant Liz Wager agrees there needs to be improved governance of scientificresearch by universities and institutions."This kind of thing has to be policed at the departmental and institutional level; they actually know what'sgoing on. They need to create an environment in which a whistleblower can feel safe."She added: "Once the paper reaches a journal, it can be one that is on another continent. The journal can'tinvestigate, speak to lots of other people involved or look at the lab notes."Peer review is good at calming down over-optimistic claims and improving the presentation, but theevidence shows it is really bad at picking out very major fraud."
Of course, when fraudulent scientific papers enter the larger literature, there is always the possibility thatthey will be exposed when other researchers try to replicate, or repeat, the findings themselves.
We think it would beharder for people toplagiarise work once you cando extensive word searchesand access more materialfree on the internet
Robert Terry, Wellcome Trust