The News About Mindless Eating Will Surprise You
Popular ideas are challenged by retractions of the original published studies.
Posted October 6, 2018
You’ve undoubtedly heard the phrase “mindless eating,” referring to the tendency for people to eat more than they otherwise would due to the influence of factors outside their control. In settings from school cafeterias to the grocery store, the mindless eating movement suggested a “paternalistic” approach in which healthy food is more prominently placed than junk food to lead people to select the fruit and veggies rather than the tater tots and brownies. By engineering the location of food choices, those who want to promote good eating habits can, according to this view, ensure that their patrons veer toward good rather than bad choices. You’ve probably also heard that if you want to lose weight, you should cut down on plate size. The same amount of food appears larger on a small plate than it does on a large plate so, the mindless eating research suggested, keep your plate sizes to a minimum, and you will feel that you’ve had enough to eat.
Much of the research on mindless eating originated from the lab of Dr. Brian Wansink, a Cornell University nutrition professor who made the tricky transition from academia to prominence in the public eye through savvy marketing of his published findings. However, something seemed amiss to the scientific reviewers and others who took closer looks at the data. The publication process is supposed to ensure that science is properly conducted, not only for the rest of the research community, but also to be able to give a seal of approval before the findings are applied in the real world.
As first reported in the New York Times and Atlantic magazine , Wansink was forced to resign from his prestigious position at Cornell when the revelations of the multiple flaws in his research could no longer be dismissed as flukes. Multiple scientific journal editors published retractions of the studies that, only a few years earlier, drew acclaim for their potential to overcome the dangers of consumers being mindless when they made food choices. These retractions ranged from the Pakistan Journal of Nutrition for a 2002 article on soy consumption in developing countries (Appetite, 2018) to Preventive Medicine for a 2012 study on how the name of a food influences its attractiveness to schoolchildren (2018). Dr. Howard Bauchner, Editor-in-Chief of the renowned Journal of the American Medical Association (JAMA), wrote in an editorial (2018) that the AMA retractions came after Cornell University was “unable to provide assurances regarding the scientific validity” of 6 studies that had come under scrutiny.
Indeed, as reported by Anahad O’Connor in the New York Times, the problems arose due to a tactic engaged in by Wansink referred to as “p-hacking,” which “is tantamount to casting a wide net and then creating a hypothesis to support whatever cherry-picked findings seem interesting — the opposite of the scientific method.” Making matters worse, Dr. Wansink prodded researchers in his lab to mine their data sets for results that would “go virally big time.” That Cornell couldn’t validate Wansink’s findings was due not only to this manipulation of results, but also to some of the errors in his own reporting of findings in the retracted papers.
The question is whether, in hindsight, it would be possible to know of these scientific sleights-of-hand. One of these studies, which was reported in a brief Research Letter in JAMA (Wansink & Cheney, 2005), was ironically published next to a set of 4 corrections of previous JAMA-published studies. The Wansink and Cheney letter described the study as an attempt to determine whether serving bowl size would influence the quantity of food a person would freely select from a buffet-like offering. A sample of 40 graduate students, purportedly invited to a Super Bowl party, were asked to sign a consent form indicating that their responses would be evaluated to “food and commercials in party environments” (p. 1727), thus disguising the purpose of the experiment. They were offered 10-inch plates and a one-cup serving scoop which they could use to help themselves to snack food. A research assistant then weighed their plates, so that the difference between pre- and post-snacking could be quantified (the plate and its waste were weighed again after participants were through eating). Because 5 participants didn’t take any food (4 women and 1 man), their data were eliminated from the analyses.
Comparisons of the pre-post snack plate weights revealed that those given large bowls took and ate about 145 calories (about 54%) more than those who filled smaller bowls. So far, so good. In fact, the researchers compared the means of the two groups only after taking into account body mass index, body weight, hours since last meal, age, and education. However, drilling into the results further shows that men not only ate more than women, but that men weren’t affected by bowl size in the extent to which they loaded up on snacks. Nevertheless, Wansink and Cheney went on to conclude that “small environmental factors an have a large influence on food consumption… although this study was not conducted in a medical setting, it is possible that if a physician giving diet-related advice recommends using smaller serving bowls, patients may serve themselves smaller portions” (p. 1728). We’ve now lost the male-female distinction, as well as translating the 145 calorie difference into practical terms (perhaps amounting to half a cup of snack food, depending on the food’s exact composition). Additionally, since the experiment was conducted in a party atmosphere, where presumably Super Bowl fans were also drinking alcohol (this was not mentioned or measured), the results may not translate into a more sober solitary setting in which dieters have no choice but to see what they’re eating while they’re eating it.
Not stopping there, however, the authors conclude that “an appropriate area for further research is whether these same cues… can be used to encourage people to eat greater amounts of healthier foods such as fruits and vegetables” (p. 1728). These overblown conclusions should have provided yellow flags to the journal’s editors. However, such flowery language and hyperbole have unfortunately become part and parcel of the media-driven research enterprise in which journals and academic institutions pursue headlines by journalists who aren’t as tuned into the kind of “p-hacking” that O’Connor described.
In a second paper, Wansick and collaborators (2012) were allowed to publish a “Corrigendum” (otherwise known as a “mistake”) to a 2012 study on the effect of vegetable names (e.g. “X-Ray Vision Carrots” vs. “Food of the Day”) on consumption of mini carrots among children ages 8-11. Wansick et al. (2018) now acknowledge that the children were in preschool and so were only 3-5 years of age. The children could not read, so the signs labeling the vegetables were read to them by adults. The mistake in the ages of the children was discovered, Wansick et al. note, “upon locating overlooked documents from 2007 and 2008.” (p. 114). Children ate more (10 vs. 5 of the little carrots), but did not take more, of the X-ray carrots. Expanding on this field study, a second experiment extended the data collection period to 40 days and did use elementary school students.
The “corrugated” study included no less than the following errors: 1. Number of children, 2. Number of children in the “treatment” school,” and 3. Misreporting by a factor of 2 of the percent change in consumption (“We inadvertently left out a factor of 2 when calculating the average in an Excel spreadsheet” (p. 115). Undaunted, the authors conclude that they have shown that “children of a wide range of ages” can be affected by the names given to their food.
In his quoted statement in The Atlantic , Wansick defended himself as follows: “Having people say, ‘I do something differently because of your research, and it works’ takes away the sting of someone pointing out the degrees of freedom in an F-test were wrong.” Making a mistake in one statistical test isn’t enough to cause a researcher to be let go from their host institution, so this defense seems to lack credibility. Furthermore, just because a researcher believes his theory to be correct doesn’t mean that the rest of the world has to change its lifestyle to go along with this theory. The field of nutrition research suffers enough as it is from findings that one day show you should eat more cheese and another day that dairy should be eaten only at your peril. The flip-flopping in nutritional advice may be what leads people to throw up their hands in dismay and confusion and just fill their bowls with as many unhealthy snacks as they want.
Hamblin points that “the shadow cast by Wansink’s story has worsened a real problem: declining credibility in nutrition and behavioral sciences, and declining public trust in science generally.” At a time when science has more powerful tools than ever to analyze the real threats to health and well-being that confront the world, this is an unfortunate outcome. On the positive side, the risk of being found to be mistaken (whether on purpose or not), the availability of more and more open-access data sets and data sharing, and the heightened vigilance due to cases such as Wansink’s and others may provide the course correction that will allow scientists and consumers to feel more confident in the science that matters most to their lives.
Wansink, B., & Westgren, R. (2018). “Profiling taste-motivated segments: The case of soy”: Retraction. Appetite, 125, 566. https://doi-org.silk.library.umass.edu/10.1016/j.appet.2018.03.013
Wansink, B., Just, D. R., Payne, C. R., & Klinger, M. Z. (2018). “Attractive names sustain increased vegetable intake in schools”: Retraction. Preventive Medicine: An International Journal Devoted to Practice and Theory, 110, 116. https://doi-org.silk.library.umass.edu/10.1016/j.ypmed.2018.02.028
Wansink, B., Just, D. R., Payne, C. R., & Klinger, M. Z. (2018). “Attractive names sustain increased vegetable intake in schools”: Corrigendum. Preventive Medicine: An International Journal Devoted to Practice and Theory, 107, 114–115. https://doi-org.silk.library.umass.edu/10.1016/j.ypmed.2017.11.032
Wansink, B. & Cheong, J. (2002). Taste profiles that correlate with soy consumption in developing countries, Pakistan Journal of Nutrition 1: 276; DOI: 10.3923/pjn.2002.276.278).