Two years ago, the star researcher Dan Ariely, a behavioral scientist, fell into the spotlight when a trio of data sleuths exposed problems with the numbers supporting a headline-making 2012 study he helped write. The study found people were less likely to lie on a form if they signed an honesty statement at the top, as opposed to the bottom.
On the blog Data Colada, the sleuths examined one of three experiments in the paper, which claimed to use data from an insurance company. Amid the fallout from their findings, Ariely claimed that the company “collected, entered, merged, and anonymized” the data before sending it to him. That company, The Hartford, said back then that it could not locate the data.
Now that’s changed, creating a new wrinkle in a much-watched case that has challenged the Duke University professor’s credibility — just as questions swirl around one of his co-authors on the same paper.
The Hartford told NPR on Friday that they found the data they contributed but discovered that after they sent it, the data was manipulated to support the findings of the study, which was retracted in 2021. The company also found differences in font between the data they sent and what shows up in the study, not to mention altogether different numbers: some inflated, some randomized.
Moreover, the company says Ariely violated their contract with him by failing to get their approval to publish the data.
“We cannot find any record of Dr. Ariely ever discussing the initial data set with us or collaborating with us on any analysis,” read the company’s statement to NPR. When asked for further comment, a spokesperson from The Hartford referred The Chronicle to the statement. In a statement of his own, Ariely said he denied the allegations.
“I have never manipulated or misrepresented data in any of my work and have never knowingly participated in any project where the data or conclusions were manipulated or misrepresented,” he said.
The experiment involving the insurance company isn’t the only one with problems. Earlier this year, Francesca Gino, a Harvard Business School professor who’s also an author on the paper, was placed on administrative leave. Later, the blog Data Colada found problems in three papers Gino helped write, which are being retracted.