Last year, when I was reviewing food researcher Brian Wansink’s book about ways to trick yourself into eating better, I noticed that the way he did some of his research seemed fishy. A statistician agreed with me. But I posted the book review anyway.
Why? Because Wansink’s findings are fascinating. He says we enjoy expensive buffet lunches more than cheap ones, even if the food is the same. Kids behave better at dinner if you cut up their food for them. You’re more likely to overeat in a messy kitchen than a clean one.
Nobody in the field was questioning his conclusions, so we didn’t either.
But it turns out that the research coming out of the Cornell Food and Brand Lab, which he runs, may have serious problems. A few months ago, Wansink was accused of pressuring a student to produce publishable results from a failed experiment. Cornell investigated four papers that came out of that experiment, and concluded that Wansink was guilty of “inappropriate data handling” but that his actions don’t meet the definition of scientific misconduct. Wansink has agreed that, going forward, he’ll have independent experts check his work.
We’ve posted tips from Wansink’s lab many times over the years—and now we’re not sure which of them are reliable. Besides those four papers that Cornell looked at, critics have questioned dozens more, saying that sometimes the numbers don’t add up. This is related to the problem that the statistician and I noticed in Wansink’s book: he waded through giant data sets looking for anything that stood out. That approach can yield lots of false positives—in other words, things that seem like they are significant but are just coincidences. (If that sounds confusing, this xkcd cartoon may help you understand.)
Since we posted so many life hacks that came out of this lab, we feel it’s only fair to let you know that they’re suspect. Below is a list of our posts that were based entirely or partly on Wansink’s research. Let us know if we missed any.