Wednesday was interesting for political scientists. Our social media feeds were full of angst in response to the news that a very influential member of our discipline had requested a retraction of a very widely reported finding published by a very prestigious journal on which he had been a co-author. The data upon which the finding rested appear to have been fraudulently produced. Thus, a process of shaming has begun. It is a necessary process. Yet it misses a very important part of the story: science actually worked.
Not much political science research gets major coverage in outlets like Bloomberg, The Washington Post and “This American Life.” The now retracted finding did (here, here, and here), and that is partly because it was published in a journal that all scientists — not just social scientists — read. A retraction of an article published in such an outlet is major scientific news, and to the best of my knowledge, no political science article has ever been retracted from such a publication. And because some U.S. lawmakers oppose funding for political science research, people are particularly concerned that this “black eye” will contribute to such critiques.
“Do not fudge the data” is, of course, an important scientific norm. Public confidence in science rests in no small part upon our upholding it. So the news that the authors of one of the most widely disseminated findings our discipline has produced of late had violated that norm was met with consternation and concern. A political science study had joined the pantheon of famous academic frauds, including the 1989 cold fusion fraud, the 2011 retraction of the vaccine-autism study, and the 2013 case of serial fraud in social psychology.
The reaction to all of these cases is publish shaming. Shaming is the standard process by which human societies reproduce norms. Norms are most readily apparent when they are being violated, and if we want the norm to persist, large groups of us must raise the alarm and call out the violators for their poor behavior.
Meanwhile, a second and more important process also just unfolded: the scientific process worked. In fact, those of us who teach undergraduates how to do science had just been handed a valuable story that we can tell in the classroom for years.
One of the challenges teaching science is undermining the popular misconception that science is practiced in isolation. This mistaken view depends on a caricature: the ethically pure, unimpeachable “scientist” in a lab coat conducting experiments to divine truth. Eureka!
Unfortunately, scientists turn out to be human beings, which is to say some of us are just as likely to succumb to temptation (cheat, commit fraud, etc.) as any other large collection of human beings. Indeed, we have norms against such behavior precisely because such behavior is tempting. Were it not, or were we a collection of ethically pure humans, the norms would be unnecessary.
Hearing that scientists are no better or worse than other collections of humanity, one might wonder whether the results reported in scientific journals are trustworthy. This is why transparency is so important: making publicly available all of the information required to replicate the research. And the retraction that has gotten everyone’s attention is the outcome of the transparency required by scientific journals.
Ira Glass, of “This American Life,” has two blog posts describing what we presently know about how this apparent fraud was uncovered. The short version is that some PhD students set out to do additional research along the lines of that in the article, but started getting different results, and thus sought out additional information to figure out what they were doing wrong. It turns out they were doing nothing wrong.
So they contacted the better known of the study’s authors, and when he contacted the co-author who had collected the data, the co-author could not provide the data, and so the well-known author wrote the journal editor and publicly requested a retraction. All of that work was then made available for public review online. The fact that it happened quickly — from publication in December 2014 to retraction in May 2015 — is all the better.
To be sure, political scientists will still have to take their lumps. This is how norms are reinforced, and as social scientists we can all understand that. But the glass is half full. To expect zero fraud is unrealistic. Fraud detection is thus very important, and episodes such as this one demonstrate why science works, despite the shortcomings of human beings.