Map of life expectancy at birth from Global Education Project.

Thursday, October 04, 2012

More on scientific fraud

It turns out I'm not just imagining things, it seems there is more of it lately -- or at least it's being caught out more. According to this study, the rate at which scientific papers are retracted has increased ten times since 1975. And contrary to the conventional wisdom, 2/3 of retractions are due to fraud, plagiarism, or duplicate publication, not unintentional error.

The rate went up noticeably after 2005, when Congress flatlined NIH. This is unfortunate but not unsurprising. The dynamic is that a period of increased funding for research increases the supply of Ph.D. students, post-docs and junior faculty who are looking to make it as scientists. When funding gets tight, some of them get desperate. Right now NIH is funding less than 10% of proposals, and if you can't get yours into that very lucky category, you can't have a career.

There's more to it than that, or course. For instance, there is also institutional culture. The report says that 43% of retractions came from just 38 labs. That means thousands of labs and research centers had none. (Mine included, I'm happy to say. Indeed, as far as I know, my entire university remains pure as the Antarctic snow.)

A commitment to stable federal support for research, growing with inflation and GDP, would help a lot. Then we'd have a world in which the number of people who make it onto the first rung of the scientific career ladder is reasonably matched to the number who can look forward to having a decent career, allowing for the appropriate level of attrition and people who just don't cut the mustard. But that's my selfish, parochial point of view. More important, we would continue to make steady progress in basic and clinical sciences with less wasted effort and investment, and maybe some day we'd have effective treatments or prevention for Alzheimer's disease or diabetes or cancer or MS, and people would get the right care that met their particular needs, at lower cost.

Wouldn't that be nice.


Anonymous said...

Science and technological innovation thrive when they are cooperative, and can get decent funding, or earn decent money, which may be quite minimal to keep it going.

When young ppl are invited to join, paid decently, not exploited, mentally tortured.

Science dies when it is made hyper competitive, individualistic, when it is grossly subsumed to financial interests, when it enters the raffle game of casino finance (by holding out the hope of future gains, future growth), when it becomes subject to ‘ratings‘ of various kinds, when money counts more than ideas. I’d add when it is subject to various ‘fashions’, memes, tropes, or temp scientific dogma, that have to be embraced to get ahead.

Sure young ppl fraud all the time. Much of the fraud is legitimate - using bad stats, going for pre-formed conclusions, etc. The line between inventing data and massaging is so thin as to be invisible.


Cervantes said...

Well, I agree that it's always possible to be tendentious and how you present data. But there are clear standards -- not universally enforced, alas, but there's a growing movement to improve peer review.

Post hoc analyses, data mining, failure to model interactions, cherry picking, spinning conclusions -- all this goes on. But I don't think the line is invisible -- or at least, the invisible part of it is very thin. These bad practices can be clearly defined and reviewers with proper access to the data can find them.

The problem is we usually don't have that access, we have to take authors' word for it.