How A Major Scandal Is Shaking Up The Booming Food Science Field

Senior Contributor
03.01.18 4 Comments

UPROXX/Shutterstock

You likely don’t know Brian Wansink’s name, but you know his work. As the head of Cornell University’s Food And Brand Lab, Wansink has developed, for years, the theory that psychology and environment play an outsized role in how you eat, what you eat, and how it affects you. He’s the academic behind the study claiming men eat more pizza to impress women and the “bottomless bowls” theory about people eating endless soup when their bowls are refilled.

If for nothing else than how much mainstream traction his papers have gotten, Wansink has probably had a strong influence on how you approach what you eat. The premise is always the same: Wansink claims diet and exercise matter less than environment. Change your plate size, the colors in your dining room, the company you eat with, Wansink’s research argues, and you’ll also change your waistline. But as intriguing as these ideas are, they’re also not holding up to scrutiny. The researcher is at the center of a scientific scandal showing he’s massaged data, made technical errors, and otherwise engaged in misleading work in order to “go viral.”

The backlash started on Monday with a detailed Buzzfeed article accusing Wansink of a host of scientific crimes, including self-plagiarism, recycling papers, and, most gravely, the scientific version of putting the cart before the horse, called “p-hacking” or data dredging. To understand why this latter practice is reviled, you need to remember that the scientific method starts with a hypothesis: You figure out what you want to learn, basically, and then you do the research. Wansink is accused of doing it in reverse; he would collect data, or obtain data from other studies, and work the data around until it could fit a hypothesis that would get attention.

The problem with p-hacking is that any scientific study’s data is going to have a degree of error, some form of statistical weirdness that allows the data to line up when looked at through a specific lens. The classic example is the fact that when ice cream sales rise, so do drownings. The two have nothing to do with each other, of course; the correlation is because when it gets warm, people go in the water more often and also eat more ice cream. It’s easy to spot an implausibility like that, of course, but what if a bunch of people who eat ice cream also happen to commit murder? Without the right degree of rigor, it could be a coincidence, or it could be a causal link.

Around The Web