Science is Dead, Long Live Science

Jim Martin, Why Research Fraud Is Getting Worse, "Until we disincentivize number-fudging, academic scientists will continue to cheat."

Once a rarity, research fraud is on the rise at some of the nation’s most prestigious universities. What is most disturbing is that the fraud in question too often involves tenured professors with sterling reputations who betray the public’s trust.

Most recently, the venue was Florida State University, where Professor Eric Stewart was terminated for “research misconduct” and for the unprecedented number of his articles that were retracted. Next in line was the City University of New York, which found “egregious misconduct” in data management and recordkeeping on the part of Hoau-Yan Wang, a professor in its School of Medicine who was working on an Alzheimer’s drug.

Earlier this decade, Harvard Business School accused Francesca Gino, a prominent professor, of data fraud in four behavioral-science papers. (Ironically, Gino’s research concerned why people lie.) Acting on a tip from the social-science research blog Data Colada, Harvard placed Gino on unpaid leave and is seeking to revoke her tenure. After Harvard came Stanford, where the president, Marc Tessier-Lavigne, resigned after a series of investigations revealed that he had failed to live up to standards “of scientific rigor and process” and had not corrected the record on numerous occasions. The founders of Retraction Watch estimate that “at least 100,000 retractions should occur every year.” When retractions are underperformed or underreported, public confidence in research is severely undermined.

 

In the past, universities were reluctant to hold their faculty accountable for misconduct because research brought in millions of taxpayer dollars. Schools still want the money, of course, but the fear of exposure on the Internet or consequences in U.S. News & World Report’s annual rankings has left them with little choice but to intervene. That explains what happened at Harvard, Stanford, and CUNY most recently. It also applies to the other elite universities, including Duke and Cornell, that have hosted misconduct over the past few years.

The best illustration of how the system works was seen in early October, when University of Pennsylvania researcher Katalin Karikó was awarded the Nobel Prize for medicine despite a strained relationship with her employer. According to the Wall Street Journal, Karikó’s prize offered a glimpse into “the clubby, hothouse world of academia and science, where winning financial funding is a constant burden, securing publication is a frustrating challenge, and those with unconventional or ambitious approaches can struggle to gain support and acceptance.” It’s a flawed system—and one that occasionally presents researchers with incentives to fudge the numbers—but it is highly resistant to change.

There is always a fraction of people willing to cheat if the incentives exist, the likelihood of getting caught is low, and/or the penalties for getting caught are not draconian. The incentives to cheat are quite high (literally career making), the chances of getting caught are low, (most peer review doesn't actually examine the data in it's rawest form, where cheating is easy), and the penalties might not be as big as you think.  I know of only one really big case of academic fraud close at hand. A friend of mine, as a graduate student, once determined that a professor at her school fabricated a huge mass of data for a report to a government agency, because the instrument necessary for the work was out of order during the period the numbers were allegedly produced. She reported his to the administration. The professor left the school, but was picked up by another university, because he was a prolific grant getter. Moving is a bitch, but fraud like that should have been an academic death sentence, and it wasn't. In an interesting coincidence, my aunt and uncle knew the guy from church, and said he was a very nice man.

I assume it starts small. 'Hey, if I move this one data point, the statistics become significant, and I can publish this six months worth of work, and anyway, it's true!' soon becomes 'Oh well, I know Baltimore Harbor is contaminated as hell, I can make up numbers that look right."