A few weeks ago, Harvard University’s president, Claudine Gay, resigned. The controversy surrounding Gay began when she, along with MIT’s Sally Kornbluth and the University of Pennsylvania’s Liz Magill, testified in Congress about alleged antisemitism on the campuses of those revered universities following the October 7, 2023, Hamas attacks against Israel. All three college presidents immediately came under fire for their alleged hypocrisy on matters of free speech. Shortly afterward, Magill resigned. Gay seemed to weather the political storm, securing the apparent endorsement of Harvard’s Board of Trustees, but was forced to resign weeks later after allegations of academic dishonesty surfaced against her.
Without delving too deeply into the political controversy here, I want to take a step back and look at the proximal reason for Gay’s resignation: academic dishonesty. I recently listened to an episode of the Freakonomics Radio podcast that was all about this subject, albeit without reference to Claudine Gay or the other two college presidents mentioned above. It was a fascinating episode, and it raised some interesting questions about what academic fraud is, why it matters, and how some people unwittingly become involved in it.
Academic fraud consists of several interrelated practices. Claudine Gay was accused of plagiarism, which is pretty straightforward: the use, without proper citation, of another scholar’s work in your own papers. Another, subtler practice of academic fraud is the manipulation or exclusion of data that skews the results of the study—usually to secure an outcome favorable to the scholar. The most egregious case of academic fraud is the outright fabrication of data in a study, which, again, is usually done to make the scholar’s work look far more interesting and applicable than it actually is.
In the Freakonomics Radio episode, the host, Stephen Dubner, interviewed a group of academic researchers who run a blog called “Data Colada.” Years ago, before they had a blog, the group decided to run a weird experiment to demonstrate how to commit academic fraud and obtain a result that makes absolutely no sense. The professors recruited people to be part of a study to see whether listening to the song “When I’m Sixty-Four” by The Beatles actually makes you younger. I don’t know all the particulars, but basically, the study’s participants listened to three songs, including the aforementioned Beatles song, and reported their age. Sure enough, by omitting the data from one of the songs, the researchers were able to show that participants who listened to “When I’m Sixty-Four” actually became younger.
It’s a silly study that obviously shows incorrect results. But it’s only obvious to us because we know that people don’t actually get younger. That’s physically impossible. But in most academic settings, the meaning of the results aren’t as obvious to observers. If, for instance, you’re investigating some social psychological process—i.e., signing off on emails with “Thanks in advance” to increase the chances of a response—it’s not as obvious to us when the study returns incoherent results.
That is also a problem when researchers fabricate data. In the Freakonomics Radio podcast episode, the main study under investigation was one that purported to show that signing one’s name at the top of a contract would make one more likely to tell the truth than signing at the bottom. In the portion of the study that relied on real-world information, the researchers investigated auto insurance data. They ran a simple experiment: Give some participants insurance forms with signatures at the top of the contract and others forms with signatures at the bottom, and have participants record how many miles they drove. Participants would have to pay more money if they reported more miles driven, so they have an incentive to lie and say they drove fewer miles.
Lo and behold, the researchers found that drivers who signed at the top of the insurance forms were more honest than those who signed at the bottom.
This finding makes intuitive sense. If we are forced to sign our name to a form at the top, it reminds us that we’ve made a commitment to be honest. And the results of the study had immense promise across domains. Imagine, for instance, if the IRS began requiring people to sign their names at the top of their tax forms. How many more billions of dollars could be collected in taxes if signing your name at the top was all it took to guarantee an increased rate of honesty? Sure enough, the paper—published in the prestigious Proceedings of the National Academy of Sciences (PNAS) and edited by the well-known behavioral psychologist Daniel Kahneman—began to attract attention from businesses all over the world. Those businesses tweaked their forms to put the signature line at the top in the hope that they would cut down on fraud from customers. At least one of the paper’s co-authors was hired as a business consultant as a result of the article’s publication, and several of the authors were hailed as rising stars in academia.
But eventually, people started to get suspicious. For one, the study didn’t replicate. This means that other researchers who took the same data and ran the same experiment could not obtain the same results. This is an obvious red flag in the world of academic research, but it isn’t necessarily due to fraud. Another red flag is the outright fabrication of data, which this study appears to have relied upon. One of the paper’s co-authors was a Harvard Business School professor who was known to be a mentor of many up-and-coming researchers at that august university. In his appearance on the Freakonomics Radio episode, he explained that he became suspicious when he noticed that many of the drivers in the dataset had reported driving upwards of 20,000 miles. Most Americans drive somewhere around 13,000 miles per year, so 20,000 seemed too high. Evidently, there was no good answer as to why these data were so high, other than the hand-waving quasi-explanation that the data had been collected for longer than one year (even though there was no indication in the study that this was indeed the case).
Eventually, PNAS retracted the study, and Harvard and Duke, where two of the co-authors worked, investigated the researchers for academic fraud. One author was suspended from Harvard without pay, and she sued Harvard and the aforementioned team at Data Colada, which originally raised objections to the paper, for defamation.
The episode is quite interesting, and I’d recommend listening to it for its careful consideration of both sides of the case, as well as the thoroughness of Dubner and his team. More importantly, it raises important questions about academic dishonesty, the peer review process, status-seeking practices among academics, the integrity of research findings, and the questionable nature of some fields of study.
It’s hard to understate the importance of honesty in academia. Science is one of the drivers of human progress, and fabricated findings or omissions of integral data blunts the advance of this fact-finding field. Moreover, science skepticism has grown in recent years, often justifiably so. But this skepticism often goes beyond healthy levels and morphs into a generalized distrust of the scientific method—the most rigorous and time-tested way of discovering truth about the world that is currently available to us. This extends beyond the hard sciences and into the social sciences, which, despite their fickleness and history of skewed results and bias, remain important ways to theorize about human existence.
Part of the reason the scientific method is so useful is that, in order for a scientific paper to get published, it must go through a peer-reviewed process. This means that editors at prestigious journals send paper submissions to editors, whose job it is to interrogate the findings of the paper to ensure their validity. In the insurance case mentioned above, prominent psychologist Daniel Kahneman—who won the Nobel Prize in Economics for upending decades of assumptions with psychological data—was apparently in charge of reviewing the paper. Despite the prestigious name, Kahneman evidently didn’t think to question the provenance of the data, and instead green-lighted the study to be published in a renowned journal. Of course, this raises the question: How reliable is the peer-review process? It’s hard to imagine that this is the only case in which the process failed to produce the desired results. In fact, there’s a whole controversy in the social sciences called the Replicability Crisis that implicates this entire process. To simplify, the Replicability Crisis was precipitated in part by the Data Colada team and others, who went through peer-reviewed and published papers to see if they could recreate the findings. In some fields, like psychology, something like a third of papers don’t replicate. Often this is because the original dataset was too small (for you non-data nerds out there, the bigger the dataset, the more reliable the results), but it’s often the result of academic dishonesty. Why aren’t these errors caught before they’re published? In theory, this is the job of peer-reviewers; in practice, it’s the job of third-party researchers like the Data Colada team. This is a questionable process, at best.
But to me, the most interesting part of the Freakonomics Radio episode was that Dubner interviewed one of the co-authors of the fraudulent insurance paper extensively. That co-author, Max Bazerman, was apparently innocent of any academic fraud, despite the apparent guilt of his co-authors. (I’m choosing to take his testimony at face value, since, as I’ll reveal shortly, that alone raises some interesting questions. Of course, he could be lying to protect his own reputation.) As mentioned above, Bazerman is well-known at Harvard Business School for being a mentor to up-and-coming researchers. He serves on many dissertation committees and assists promising students with their research. He revealed that his name is on several papers for which he performed almost no research. In other words, after the experiments are run—and often, I suspect, after the papers are written—he appends his name to the paper for publication. Dubner asked him why he does this, since it sometimes puts him into hot water (like in the insurance paper). His response, as I recall, was that it helps lend legitimacy to a paper written by a less well-known author, which could increase the paper’s chances of publication. He evidently does this out of the goodness of his own heart, as he likes to help start academics’ careers.
Of course, Bazerman is likely not telling the whole truth. Academic reputations are built in large part on how many papers an author has written and how often those papers are cited. So Bazerman actually benefits by attaching his name to papers that he had very little to do with—so long as the papers aren’t fraudulent. But I submit that appending your name to a paper for which you had little to no input in its design, experimentation, or writing is its own form of academic dishonesty.
In the end, as with many human endeavors, much of it comes down to status-seeking behavior. Researchers want to be world-renowned, to be known as rising stars in academia, to land lucrative consulting gigs, and to have their bylines appear in prestigious academic journals, newspapers, and magazines. They want tenure not only for job security but also for clout. They want to land appointments at Ivy League universities not just for the healthy salary (which is actually lower than they could make in the business world) but also for the access to politicians and other elites that comes along with the resume. Unfortunately for the rest of us, these goals can conflict with the ostensible purpose of academic research: to advance the sum total of human knowledge.
Academia isn’t alone in this corruption. Journalism is increasingly warped to fit the viewpoint of a certain set of elites, and even courts have become politicized over time. In fact, Jonathan Rauch, a center-left intellectual at the Brookings Institution, recently wrote a book (The Constitution of Knowledge) that addresses these concerns and notes that these fields must crack down on the corruption in their ranks in order to reclaim the trust of all Americans.
But we should be able to trust academia. That ever-decreasing numbers of Americans do not is cause for deep concern. Academic fraud may seem like a niche problem for Ivy Leaguers to deal with, but it has far-reaching implications.