The hunt for statistical significance in scientific research

Food researcher Brian Wansink, known around the world as a top eating behavior expert, has resigned from his position at Cornell University after 13 of his studies were retracted due to academic misconduct. Cornell announced that his academic misconducts included the misreporting of research data, problematic statistical techniques, failure to properly document and preserve research results and inappropriate authorship.

Wansink rose to fame at Cornell after finding that people who grocery shop hungry buy more calories, pre-ordering lunch can help people choose healthier food and serving food out of large bowls encourages people to serve themselves large portions. His work, which has been cited over 20,000 times and has influenced numerous dietary guidelines and eating programs in the US, was retracted after it was found that he illegally manipulated his results. Officials began investigating Wansink’s past studies when Wansink  published a blog post in which he encouraged graduate students to hypothesize after results are found instead of before. After further inspection, scientists discovered that Wansink was p-hacking, the misuse of data to find trends that are statistically significant when in reality, no significant result exists. In p-hacking, the p represents  a measure of statistical significance which must be less than 0.05 for researchers to call their results significant. Instead of testing a hypothesis and reporting on all findings, Wansink selected specific data to obtain more desirable resul.

In scientific fields, p-hacking is a huge issue. A 2012 survey of 2,000 psychologists conducted by Leslie John of Harvard Business School found that 50 percent of participants only reported studies that panned out, while 20 percent of participants stopped data collection after achieving the desired results. The regularity of p-hacking  and other incomplete research techniques raises the issue of the publish or perish mentality among researchers.

“Wansink  and  other  researchers  are  taking  the  easy  way  out  with these unethical tactics.  This  not  only can  reverse    previous  studies  but can  also  undermine  the  credibility  of  the  academic  field  and  misinform consumers,” Sophomore  Ryan  Vakhshoori  said.

In order to increase their chance of being hired at top universities or receiving grants, scientists have to publish statistically significant studies in respected journals. This pressurizes researchers to manipulate and modify data in order to get desired results.

“If scientists only get grants for publishing statistically significant results, then naturally they will place a higher focus on positive results in order to get published and paid,” Sophomore Esmeralda Cortez said.

However,  solutions  to  the  abundance  of  p-hacking  in   academia exist.  One  suggestion  is  to  make  researchers  publish   their  hypotheses and share their data online,   which prevents   scientists  from  altering them unethically to support questionable results.

“To minimize academic misconduct, scientists should be rewarded for their honest work even if their results are inconclusive, so that fewer researchers feel inclined to cheat their way to the top. It is important to prioritize ethics over results,” Sophomore Veronica Son said.