Data Colada Accuses Duke’s Dan Ariely Of Tampering With Data In A 2nd Influential Study

Dan Ariely, the Duke Fuqua professor who has spent several years fending off accusations of research misconduct, is facing a fresh round of scrutiny after the academic fraud-detection blog Data Colada concluded that data underlying one of his most-cited studies was likely tampered with or fabricated.

The findings, first reported by the Duke Chronicle, focus on Study 2 of a 2002 paper Ariely co-authored with Klaus Wertenbroch of INSEAD, titled Procrastination, Deadlines, and Performance: Self-Control by Precommitment. Published in Psychological Science, the paper found that people completed tasks more successfully when given evenly spaced, externally imposed deadlines than when they set their own deadlines or faced a single deadline at the end.

According to Data Colada, it has been cited more than 2,100 times and remains standard reading in economics and psychology courses.

4 RED FLAGS

Ariely, 59, is a professor of business administration at Duke’s Fuqua School of Business, with joint appointments in psychiatry and behavioral sciences, the Sanford School of Public Policy, and Trinity College’s economics department. He earned his doctorate at Duke in 1998 and founded the university’s Center for Advanced Hindsight. He is the author of several bestselling books, including Predictably Irrational and The Honest Truth About Dishonesty, and wrote The Wall Street Journal’s “Ask Ariely” advice column from 2012 to 2022.

In an August 31 post, Data Colada founders Uri Simonsohn, Leif Nelson, and Joseph Simmons lay out four anomalies they say cannot be explained innocently. The reported effect size is implausibly large – larger, they note, than the effect of gender on height. Eighteen of the 20 participants in the Last Day Deadline condition have an exact “corrections twin,” another participant who logged identical results on all three tasks. Measures that should correlate strongly, such as how much a participant said they liked a task and how interesting they found it, show almost no relationship in the original data even though they correlate strongly in an independent replication. And self-reported time estimates, which people typically round to convenient numbers, were round barely more often than chance would predict.

The Data Colada team writes that it cannot construct an innocent explanation covering all of the anomalies, concluding that the data were “severely tampered with or fabricated to produce the desired results.”

HOW THE ANOMALIES SURFACED

According to Data Colada’s account, the episode traces back to a new replication attempt by Kyle Hyndman, who received the original study files from Ariely by email in 2006, and Alberto Bisin. Their replication of Study 2 failed.

A footnote in their forthcoming Psychological Science paper discloses that in 2024, at the request of the journal’s editors, the replication authors shared their own analysis of the original file with Ariely and asked to include a summary of it in their paper; Ariely denied the request, arguing the file might not contain authentic data. That disclosure, Data Colada says, is what prompted Simonsohn, Nelson, and Simmons to dig into the original files themselves.

A companion post published Wednesday (September 2) examines Study 1 of the same paper – a comparison of two sections of an MIT executive-education course in which students wrote three papers under either evenly spaced or self-chosen deadlines. Data Colada says the final grades of 13 of the 49 students in the self-chosen-deadline section had been altered by hand from what their component grades would produce, with 12 of those 13 changes moving in the direction that supported the paper’s finding that students who spread out their own deadlines performed better – while the section’s overall average grade was left unchanged by the alterations.

Using an earlier, pre-review version of the manuscript that it located in an INSEAD working-paper archive, Data Colada says it could reproduce the paper’s published statistics only after applying those alterations, and could reproduce the statistics reported in the earlier draft only after reversing them.

ARIELY & WERTENBROCH RESPOND

Wertenbroch, who says he has never had access to the underlying data for any of the paper’s studies, calls Data Colada’s forensic analysis “sophisticated and thorough” and told the blog it demonstrates that the results are false. He has asked Psychological Science to retract the paper.

Ariely has not addressed the specific findings directly with Data Colada or the Chronicle, but in a statement posted to LinkedIn and his personal website, he says he was recently made aware that the data “contained serious anomalies,” and that neither his current records nor his memory of experiments run more than two decades ago are enough to answer the questions that have come up. He says his focus now is cooperating with the journal’s review and retraction process.

Duke did not respond to Chronicle’s request for comment on whether the university was aware of the allegations or planned to investigate.

A LONGER RECORD OF SCRUTINY

This is not Ariely’s first entanglement with data-integrity questions – and not the first time Data Colada has been the source. In August 2021, the same trio published an analysis finding that a 2012 field study on dishonesty, which Ariely co-authored with four other researchers including Harvard Business School’s Francesca Gino and Max Bazerman, relied on data that appeared to have been fabricated. The Harvard Crimson reported at the time that the paper, which examined whether people were more honest when they signed an integrity pledge at the top of a form rather than the bottom, had been widely relied on by organizations trying to cut down on fraudulent reporting. The study was later retracted.

Ariely, the only co-author with direct contact with the insurance company that supplied the data, denied fabricating it. In a statement to Data Colada, he said his own role was limited to passing along the file he’d received from the insurer, and that he had no part in gathering or entering the underlying records himself. Gino would later face her own, separate misconduct findings at Harvard.

Ariely told The Chronicle of Higher Education in January 2024 that Duke had by then completed a confidential investigation into the 2012 paper, concluding that he had not knowingly falsified the data but should have more carefully vetted and preserved it. Duke has repeatedly declined to confirm or detail the findings of that review, citing faculty privacy.

THE EPSTEIN CONNECTION

The renewed attention to Ariely’s research record follows a separate controversy earlier this year. The Duke Chronicle first reported in January that a Justice Department disclosure of roughly 3.5 million pages of records tied to Jeffrey Epstein showed Ariely’s name appearing 636 times, including email exchanges in which he and Epstein discussed meeting up and, in one 2012 message, Ariely asked Epstein to help him reconnect with a woman he had met.

A Duke spokesperson told Poets&Quants at the time that the university was “carefully evaluating” the material but could not comment on any personnel actions. As the Chronicle noted, the records did not allege criminal or unethical conduct by Ariely. He addressed the emails in a February op-ed in the Chronicle, describing his contact with Epstein as infrequent and largely logistical, and saying he had no financial, professional, or ongoing relationship with him.

As of this writing, Psychological Science‘s review of the retraction request for the 2002 paper is ongoing, and Duke has not said whether it will open a new investigation.

DON’T MISS REPORT: DUKE PROFESSOR APPEARS PROMINENTLY IN NEWLY RELEASED EPSTEIN EMAILS

© Copyright 2026 Poets & Quants. All rights reserved. This article may not be republished, rewritten or otherwise distributed without written permission. To reprint or license this article or any content from Poets & Quants, please submit your request HERE.