3 ms·
How a Controversial Study About Kids and Cookies Turned Out to Be Wrong, Again
- danso 9y agoPrevious discussion: https://news.ycombinator.com/item?id=15351006 https://news.ycombinator.com/item?id=15351006 Cornell professor Brian Wansink professor has been under fire the past year for publishing studies with egregious statistical errors. So far he and Cornell have weathered the storm by saying that the errors were a result of human error and differences of opinion in statistical judgment. But the revelation here -- that Wansink knowingly promoted a study as involving elementary school children, when it involved preschoolers -- seems to indicate that Wansink and his colleagues conspired to publish an outright lie. Wansink isn't admitting to such malice, but he has agreed that the study should be retracted, not just updated with a "corrected" analysis. The age group difference is enough to completely invalidate the study, which purports to show how grade school children would be more likely to chose healthy food if the healthy food "had a sticker of Elmo", a tactic that seems would have different effect on an 8-year-old versus a 3-year-old. But there's an additional element of nefariousness in that this study became the evidence for a $22 million federal program about school evidence. It would not be federally-funded evidence if it had been known that the study did not involve K-6 schools. Certainly, prestigious studies have been published with egregious accidental errors. And some studies have been published using fake/copied data, or dishonest analysis, with the blame being assigned to rogue researchers. But this kind of "mistake" would seemingly have to involve not just its TED-famous leader, but his students and his collaborators.