{shortcode-302ecdce2304c88354697d7426276bdac06af697}

Quantitative analysis blog Data Colada published an analysis Saturday summarizing three exhibits in the defamation lawsuit filed against them by Harvard Business School professor Francesca Gino — the bloggers’ first public remarks on the substance of the lawsuit.

In June 2023, Data Colada published four posts alleging data fraud in four papers co-authored by Gino. Soon after the allegations, Harvard notified Gino of potential tenure revocation on July 28.

Just five days later, Gino took legal action — against Harvard, HBS Dean Srikant Datar, and the data investigation blog Data Colada — for alleged defamation and gender-based discrimination. Gino has accused Data Colada of conspiring with Harvard to defame her, citing the bloggers’ decision to bring their allegations directly to the Business School in June 2021.

Data Colada’s post Saturday walked readers through three exhibits from Gino’s lawsuit, which included retraction letters sent by Harvard to three academic journals. The letters contain findings by the bloggers and Maidstone Consulting Group, the external forensic firm hired by Harvard’s investigation committee in May 2022.

The post came just over two weeks after the bloggers announced they had retained legal counsel using funds raised by academics and supporters in response to the lawsuit. As of Wednesday evening, the fundraiser had raised more than $330,000 for the team’s legal fees.

Andrew T. Miltenberg, an attorney for Gino, wrote in a statement Wednesday that the exhibits should not be construed as evidence of fraud.

“It is essential to remember that the forensics firm hired by HBS was unable to conclude fraud,” Miltenberg wrote. “The retraction notices HBS sent to journals quoted selectively from its forensics consultants — and pulled in unattributed extended verbatim quotes from Data Colada. The resulting mash-up with no statement of authorship — no person or team taking either credit or responsibility.”

“This combination is especially misleading,” Miltenberg added. “This is the reason these were included as exhibits in a defamation case and should not be perceived as fact.”

A retraction email sent by Harvard to the Proceedings of the National Academy of Sciences reported that Gino’s 2012 study was “invalid due to alteration of the data that affects the significance of the findings.” Retraction notices addressed to two other journals — Psychological Science and the Journal of Personality and Social Psychology — included similar wording.

In each notice, Harvard investigation officers alleged discrepancies between public repository Open Science Framework data – published by Gino – and original data – collected by Qualtrics or research assistants.

In the final part of its examination of Gino’s work in June, Data Colada wrote that “we have received confirmation, from outside of Harvard, that Harvard’s investigators did look at the original Qualtrics data file and that the data had been modified.”

Gino, however, alleges this post defamed her by “falsely implying that there are additional, non-disclosed facts that irrefutably demonstrate Professor Gino’s malfeasance.”

She claims that Harvard could not prove the originality of the Qualtrics data over the published data — or of her involvement in modifying it.

The data bloggers had flagged eight participants in Gino’s 2012 paper — a study on honesty pledges — under suspicion that their responses had been tampered with, as the response IDs were either out of order or repeated.

In a separate investigation, the forensic firm uncovered 11 observations that appeared in Gino’s 2012 study but not in the original Excel files obtained from her research assistant — including the responses from all eight of Data Colada’s flagged participants.

“The probability of us getting this right by chance is about 1 in a billion,” Data Colada wrote in their post Saturday.

The forensic firm’s investigation also found that “52% of reported responses contained entries that were modified without apparent cause.”

In another of Gino’s papers — a 2015 study linking inauthenticity to feelings of impurity — Data Colada pointed out anomalies in participants’ demographic information. Several study participants incorrectly listed their graduation year as “Harvard,” and Data Colada observed that these specific responses were extraordinarily supportive of Gino’s hypothesis.

The firm found that 20 of the 24 responses listing their class year as “Harvard” were original, but 8 of these responses had been altered in the published dataset, along with several responses that listed actual class years.

According to the firm, 25 observations that appeared in Gino’s published data were absent in the study’s original Qualtrics files — all of which strongly supported Gino’s hypothesis. The firm noted that while Gino’s published dataset provides statistically significant support for her claim, the study’s original dataset does not.

“Exhibit 4 documents much more evidence of tampering than we had uncovered, with many other (non-‘Harvard’) observations having been altered (or selectively included/excluded), in ways that contributed to the published findings,” Data Colada wrote in their post Saturday.

The lawsuit also includes the retraction letter for a 2014 study by Gino, which found that individuals who cheated on a task were more creative on subsequent unrelated tasks.

In this study, Data Colada flagged 13 out-of-order observations under suspicion of manipulation — the dataset was otherwise almost perfectly sorted.

Published excerpts of the firm’s investigation did not examine these data points but did find that some of the study’s independent variables (specifically, whether or not a person had cheated) were altered from the original data.

Exhibit 5 focuses on a 2020 study on mindset and networking, where Gino reported that being in a “prevention-focused” mindset rather than a “promotion-focused” mindset makes people more likely to view a networking event as morally impure. In their original post, Data Colada hypothesized that some of the study’s “promotion” responses had been changed to reflect lower levels of perceived moral impurity, and some “prevention” responses had been changed to reflect higher levels of perceived moral impurity.

On Saturday, Data Colada wrote that the firm’s findings referencing in Exhibit 5 reference evidence of tampering that “seems to confirm” Data Colada’s earlier observations. Moreover, Harvard’s investigation found far more potentially altered responses than Data Colada had — flagging 168 answers as possibly tampered with.

Currently, Data Colada is in the process of responding to Gino’s lawsuit itself. The bloggers wrote in a Sept. 1 post that their employers — the University of Pennsylvania, University of California, Berkeley, and ESADE — have indicated that they will support Data Colada at least through the initial stages of the legal process.

Data Colada must file a motion in response to Gino’s lawsuit by Nov. 8. If the case is not dismissed, Gino’s suit against the bloggers could last anywhere from a few months to several years.

In the meantime, the bloggers write that Data Colada will continue mostly unchanged.

“We have some posts in the works, some of which are relevant to this ongoing crisis, and some of which are not,” said Data Colada in their Sept. 1 post. “We don’t know what is going to happen next, but we are confident that ultimately we will win this case.”

—Staff writer Adelaide E. Parker can be reached at adelaide.parker@thecrimson.com. Follow her on X @adelaide_prkr.

—Staff writer Jennifer Y. Song can be reached at jennifer.song@thecrimson.com.