2023 Author: Bryan Walter | [email protected]. Last modified: 2023-05-21 22:24
Brian Wonsink, a renowned researcher in the psychology of food and eating behavior, who once was in charge of official dietary recommendations in the United States, resigned from his post as professor at Cornell University. Wonsink's 13 articles have been recalled by scientific journals, and the university accuses him of manipulating data and statistics and other behavior unworthy of a scientist.
In Cornell, where the scientist worked for 14 years, Wonsink ran his own laboratory, the Food and Brand Lab, which studied the psychological characteristics of human eating behavior. The main idea in their research was the ability to influence this behavior relatively easily through external factors like plate size (people eat more from larger plates) or even movie watching (people eat more popcorn when they watch sad movies). Wonsink received the 2007 Shnobel Prize in Nutritional Science for one study that used a bottomless, automatically filled bowl of soup.
As BuzzFeed News notes, Wonsink was quite famous: he was invited to the morning TV shows, the New York Times and Oprah Winfrey magazine wrote about him. In addition, a scientist who previously led the USDA's healthy eating committee is still running a $ 22 million program to "improve" canteens in 30,000 schools.
Cornell said in a statement that Wonsink "misconducted scientific research, including inappropriate publication of data, problematic statistical methods, failure to comply with scientific documentation and storage requirements, and inappropriate authorship." Initially, in the spring of 2017, the university did not find any problems in Wonsink's actions, but the investigation was resumed in November 2017 after another withdrawal of the scientist's articles from journals.
Wonsink announced his retirement from the university the day after three JAMA journals recalled six articles by the scientist and his coauthors, published between 2005 and 2014 - now the total number of recalled articles has reached 13. In the spring, these journals issued a special reporting that the editors asked Cornell to double-check the articles' findings, which the university could not do without access to the original data. Wonsink himself, in a comment to BuzzFeed News, was surprised at the withdrawal of his articles, explained that his group simply threw out the paper questionnaires after the data from them were digitized, and expressed confidence that the results of the withdrawn articles could be reproduced.
In fact, Wonsink suffered not even because of the most "inappropriate scientific behavior", but because he did not particularly hide his willingness to go to any lengths for the sake of a bright scientific result - and inadvertently attracted the attention of people interested in his conclusions. In November 2016, Wonsink wrote on his blog about a graduate student who took a "failed" study with an unconfirmed hypothesis, on which the laboratory spent money and resources, and analyzed the data from different angles until she received an interesting result (as a result, based on this data Wonsink and his colleagues published four articles, which had to be officially revised.)
In English, this behavior is sometimes called "cooking" data, when you are trying by all means to get at least some meaningful result from the data that does not support your hypothesis. Statistical methods (or rather, their arbitrary use) allow you to achieve this, but such results are not considered correct, and doing so is unethical: in this way you fit the hypothesis to the data. Nevertheless, this occurs often enough that there are special terms for all these actions: the fact is that, despite many years of discussion of this problem, to publish a work in a scientific journal with a negative result - when your assumption has not been confirmed - is still difficult, and such results attract immeasurably less public attention. It is extremely unprofitable not to publish at all (in the academic environment they usually say “publish or disappear”, everything is so serious), so there is a serious motivation for “cooking”.
Wonsink himself, as can be seen from the explanations in the blog, did not think that he was writing about something inappropriate - in his opinion, the graduate student in the post simply deeply and in detail analyzed the initial data. But commentators took it differently, the post caused wide discussion and interested Tim van der Zee, Jordan Anagnu, Nicholas Brown and James Heathers - also graduate students and scientists interested in statistics and the quality of scientific publications. Van der Zee, Ananja, and Braun published a detailed breakdown of four articles on pizza, in which they pointed out, in particular, inconsistencies with sample size and some statistics. Wonsink eventually announced that the work would be rechecked, and released edits to them, but there was no turning back: critics began to study his other high-profile studies.
The public history of Wonsinka was made a big investigation by the same BuzzFeed News, who read his correspondence with the laboratory staff and interviewed one of them - at that time, Cornell was already investigating the scientist's activities. The investigation, for example, says that Wonsink was deliberately looking for results with “viral” potential that might interest the media, and his co-authors mentioned statistical “magic” that would help them draw interesting conclusions. Wonsink and his co-authors in correspondence even discussed the need to artificially obtain a statistical significance of p below 0.05 (this is considered the standard for scientific research). As a result, the study, in which p in the final version magically decreased from 0.06 to 0.02, was withdrawn from the journal for another reason: Wonsink admitted that it was actually not done on schoolchildren 8-11 years old, but on preschoolers …
Wonsink, whose Hirsch index is 53 according to Scopus (meaning he has at least 53 papers cited at least 53 times; that's quite a lot), will retire on June 30, 2019, at the end of the school year. Until that time, he will neither teach nor conduct scientific work, but the university obliged him to help double-check his own previous research. So it is quite possible that the list of withdrawn articles of the scientist will be replenished: 19 articles are needed to get into the top thirty world "champions" in this indicator according to the Retraction Watch portal.
