One-Third Of US Scientists Admit Wrongdoing
By Maura Lerner
The types of misbehavior range from claiming credit for someone else's work, to changing results because of pressure from the sponsor.
"Our findings suggest that U.S. scientists engage in a range of behaviors extending far beyond falsification, fabrication and plagiarism that can damage the integrity of science," the authors write in a commentary piece in tomorrow's journal.
The survey, which was led by Brian Martinson of the HealthPartners Research Foundation in Bloomington, questioned more than 3,200 scientists around the country about a long list of questionable actions. They range from outright fraud to improper relationships with research subjects.
Among the findings: 15 percent said they had changed the design, methods or results of a study in response to pressure from a financial sponsor.
Fewer than 1 percent admitted to "falsifying or cooking research data." Slightly more, 1.4 percent, said they had potentially improper relationships with students or subjects.
But significantly more -- 12.5 percent -- said they had overlooked others scientists' use of flawed data or questionable interpretations. And 7 percent admitted ignoring "minor" rules for protecting human subjects. Six percent said that they failed to report data that contradicted their previous work.
Martinson, a sociologist, said the fact that a third of those surveyed admitted to one of the top ten violations suggests the problem doesn't lie with a few "bad apples."
Scientists, he said, are "one of the hardest-working groups of people that I know." But he said there may be something about their working environment -- the mountains of rules, the pressure to compete for grants and produce results -- that ends up compromising their ethics.
"There's been this kind of idea that scientists... are super-humans or something, that they're immune from these kinds of pressure," he said. "But scientists are human."
He said this is the first survey of its kind, so it is not known whether the misbehavior is more common now than in the past.
--------------------------------------------------------------------------------