The Financially Driven Erosion of Scientific Integrity

All else being equal, if you pay for something bad, you will get more of it. If you punish something good, you will get less of it. These basic rules of economics apply as much to junk science and scientific integrity as they do to junk food and political candor.

Science and the scientific method are the jewels in the crown of Western civilization. The ascertainment of facts, construction of reproducible experiments, development of falsifiable theories, impartial training and meritocratic advancement of practitioners, and - most importantly - integrity of the publication process by which a well established body of truth can be confidently assembled all underpin the respect accorded to science by the citizenry. In modern times, this respect translates into tax dollars.

Unfortunately, today those tax dollars are corrupting the process. Unprecedented billions are doled out by unaccountable federal and state bureaucracies run by and for the benefit of a closed guild of practitioners. This has created a moral hazard to scientific integrity no less threatening than the moral hazard to financial integrity that recently destroyed our banking system.

According to a report in Nature Reviews Drug Discovery, nearly two-thirds of the experimental results published in peer-reviewed journals could not be reproduced in Bayer's labs. The latest special issue of Science is devoted to the growing problem of irreproducibility. The Wall Street Journal reports that Amgen, Pfizer, and others have abandoned research programs after spending hundreds of millions pursuing academic research that could never be replicated.

And, of course, there is the heated controversy over claims of impending doom made by climate scientists, each trying to out-sensationalize the other as they bid for taxpayer money to fund their research. Declaring an active area of investigation "settled," demonizing critics, and promoting unfalsifiable theories may qualify as topics of study within political science, but they are alien to the scientific method. This kind of behavior imperils not just the scientific community. When translated into international policy, it imperils the entire global economy.

Exactly what is going on here? To find out, we must analyze the motivations and behaviors that lead scientific investigators first to job security and academic freedom and, sometimes, onward to fame and fortune.

Tenured faculty at our research universities sit at the pinnacle of the scientific community, acting as Principal Investigators (PIs) on government-supported research projects. To become a PI, one must serve an undetermined number of years as an indentured apprentice, first as a graduate student and then as a post-doc. Pleasing one's PI is the key to graduation, publication, and advancement.

The lion's share of actual laboratory work is done by these apprentices. PIs preside over the process. They formulate theories, as they try to become the first to plant a flag in virgin research territory. Well-funded PIs assign apprentices to conduct experiments that will confirm their theories, sometimes assigning multiple teams to work on the same project in parallel. They write grant applications seeking to expand their domain along with the number of apprentices they can support. And they put their names on all the papers that come out of their fiefdoms, acting as gatekeepers in the process.

Fame goes to those that publish first, not necessarily those that do the best, most thorough, or most reproducible science. Grants go to those with the most fame. The work of PIs who publish novel work in prestigious journals may be peer-reviewed, but it is rarely replicated by their fellow PIs.

For some PIs, life gets even better. Thanks to the Bayh-Dole Act of 1980, PIs can patent the results of their work even if it was supported with tax dollars. The benefit to society is that both venture capital and industry investments can be attracted to translate breakthroughs into the market that might otherwise languish in academia if they were unprotectable.

The downside is that some peripatetic PIs make a specialty of dashing about planting broad and vague patent land mines. These PIs may never reduce a single idea to practice, but the mines they plant sit and wait to extract tribute from future entrepreneurs who do. Either way, personal fortunes can be made.

This environment is rife with moral hazard.

While outright data fabrication does occur, it is rare. The bigger threat to scientific integrity is the temptation to cherry pick results as they are produced by a Darwinian horde of apprentices clamoring for admission into the guild. Failed experiments never get reported, the definition of failure sometimes including results that call a PI's pet theories into question. Confirmation bias pervades the process much more so than in industry since the consequences of spending billions drilling a dry hole are severe.

But what are the consequences for publishing a paper with irreproducible results? What becomes of tenured PIs whose junk science leads us down blind alleys, polluting the literature while precipitating hundreds of millions of dollars in someone else's losses?

They write another grant application.