Misconduct? Spend More on Science, not Less

What’s behind paper retractions? (29)
by Adam Marcus and Ivan Oransky, Labtimes 04/2015



Paperbasket
Photo: zettberlin / Source: PHOTOCASE

A recent survey suggests that in the USA, 28 billion dollars are wasted on irreproducible studies. Why not give that money to replication analyses?

The retraction last month by Science of a paper about attitudes toward gays and gay marriage was justifiably disheartening. It was yet another in a growing roster of high-profile studies marred by misconduct.

But focussing too much on such misconduct ignores the much larger problems science is facing in reproducibility. And that failure to reproduce results carries a hefty price tag. A new study in PLOS Biology by Leonard Freedman, of the Global Biological Standards Institute, and his colleagues at Boston University, estimates that waste associated with irreproducible, preclinical studies costs the government and private funders $28 billion a year, or about a quarter of the overall spend on such research. That’s nearly the entire budget of the US National Institutes of Health.

Chronic underfunding creates perverse incentives

An understandable, if wrong-headed response, to all that misspending, however, is to suggest that there’s too much research. The reality is that chronic underfunding by the federal government of basic science has created a system of perverse incentives, in which taking shortcuts and outright deceit are all but inevitable.

Rather than starve the beast to kill fraud and increase reproducibility, the better approach is the opposite: Spend more, not less, money on research – just spend it smarter.

Labs short on dollars have to choose. They can push on with new studies or attempt to replicate earlier work. As Freedman’s group found, many are opting for the former route. That’s not surprising, given the ways career advancement in science happens.

In that way, the cycle feeds waste: funding squandered on research that can’t be reproduced crowds out investment in more productive work, and exacerbates the problem of underfunding of the sciences by the federal government. Here’s how: Many experts in research misconduct believe that funding pressures are at least one driver of science cheats. The shrinking pool of grant money – the $30 billion budget for the National Institutes of Health is some 22% lower, in real dollars, than it was in 2003 – heightens the fierce competition to publish in top-tier journals, bylines that can jumpstart a young career, cement a tenure application and, most important, attract even more grant money.

To be sure, simply throwing more money at scientists is not going to produce more robust, fraud-free findings. A more sensible solution is to substantially increase federal funding for research, with strings attached. Portions of grants should be earmarked for replication analyses – studies that ought to be conducted before researchers submit their work for publication. Here’s a proposal: Add enough to the NIH budget to cover replications, too – about $8 billion, using the figures in Freedman’s team’s analysis – but make sure the extra amount is spent on replication, not new studies.

Lessons learned reduce waste

If at least half of the NIH budget is already being used on studies that turn out not to hold up, why not spend a fraction of that on making sure they do? Over time, the lessons scientists learn from flawed studies should reduce the waste – so there’s a chance this could all pay for itself, eventually.

And universities should devote a much greater percentage of the overhead built into grants – which ranges from 20% to 85%, according to Nature – to monitoring the work of their faculty and research staff. Such internal policing helped to catch Scott Reuben, a US pain specialist, who went to federal prison for faking data in drug trials.

The spectacular flame-outs of scientists caught red-handed surely don’t help the public image of the field. But researchers rightly point out that retractions in themselves are not a symptom of a broken science machine. Indeed, the opposite is true: science is self-correcting. It’s designed to catch errors and, over time, truth wins out.

What’s more, spending a bit more on a regular basis can only make science more efficient. Isn’t that worth it?



The authors run the blog Retraction Watch: http://retractionwatch.com



Last Changed: 08.07.2015




Information 4


Information 5


Information 6