Everyone involved in Dan Ariely’s fake data scandal now has an alibi—except for Ariely himself

In the fall of 2021, Dan Ariely, the most famous marketing professor in the world, was caught with fake data. A prominent study about how to make people more honest by signing at the top of the document included data from an insurance company that was obviously tampered with. This is a paper that was widely relied on for the implementation of expensive government policy changes around the world. The paper was later retracted and we more recently learned that another author, Francesca Gino, had separately faked up the data in the other studies in the paper. Two seemingly unrelated fraud cases in one paper. You can’t make this stuff up. It’s too far-fetched.

The Data Colada blog that first exposed Dan Ariely’s fake data made the following assessment about who could have logically faked up the data based on documentary evidence made available to them at the time:

the fourth author himself, someone in the fourth author’s lab, or someone at the insurance company.

Data Colada 98

Dan Ariely was the fourth author. So let’s look at these parties one at a time:

(1) The fourth author (Dan Ariely)

Still no alibi; claims it wasn’t him without offering to explain anything or provide any documentary evidence.

(2) Someone in the author’s lab (Dan Ariely’s subordinates)

Dan Ariely ruled out members of his own lab in an interview with an Israeli news outlet, claiming instead that it must have been someone at the insurance agency.

(3) The insurance agency

This is today’s news. Hartford did a thorough investigation and put out a statement about their findings that was published in NPR today (July 28, 2023). Here are some of their claims:

First, in May 2008 Hartford provided a 6,033 vehicle dataset to Dan Ariely (there were 20,741 rows in the published study). The data had been substantially altered after it left their hands–Hartford was able to find the emails. Here are parts of their statement:

While some of the information included in the Data Colada analysis appears to have come from the raw data we provided, our review shows that the data is not the same. There appear to be significant changes made to the size, shape and characteristics of our data after we provided it and without our knowledge or consent.

The number of policies and vehicles in our data is only a fraction of those in the published study data (3,756 policies/6,033 vehicles vs 13,488 policies/20,741 vehicles).

Though some of the data in the published study data is originally sourced from our data (i.e., certain columns of data match portions of the published study data), it is clear the data was manipulated inappropriately and supplemented by synthesized or fabricated data.

While the published data supports Ariely’s hypothesis, our data does not.

Hartford Insurance

Second, Ariely published the data in violation of his contract with Hartford

We cannot find any record of Dr. Ariely ever discussing the initial data set with us or collaborating with us on any analysis. Dr. Ariely never sought our approval to publish our data in a study, violating our contract with him, which ran from July 2007 through December 2008. Our last email to Dr. Ariely was in February 2009.

Hartford Insurance


According to Dan Ariely’s own “research,” it’s obvious that someone would fake up some data in order to become rich and famous, given the opportunity. He says everyone would do it–I don’t agree. Everyone he can personally relate to may do the same thing in his situation. What’s less obvious to me personally is why a supposedly prestigious institution like Duke University would enable that kind of behavior. What is Duke’s angle? Isn’t that the more important question? Also, what about the publisher of his forthcoming book, Harper Collins? Why would they publish what we now know is fiction, and try to pass it off as some sort of scientific knowledge? What is their incentive. Bad behavior is more often than not a story about the enablers.


Leave a Reply

Your email address will not be published. Required fields are marked *