Data Colada

Around 2011, Simmons, Nelson and Simonsohn "bonded over the false, ridiculous, and flashy findings that the field [of behavioral sciences] was capable of producing", such as a paper by Cornell psychologist Daryl Bem that had supposedly found evidence for clairvoyance.[2] They reacted by publishing an influential[1] 2011 paper about false positive results in psychology, illustrating the problem with a parody research finding that supposedly showed that listening to the Beatles song "When I’m Sixty-Four" made experiments subjects one and a half years younger.[2]

The "Data Colada" blog was launched two years later, in 2013, carrying the tagline "Thinking about evidence, and vice versa", and "became a hub for nerdy discussions of statistical methods — and, before long, various research crimes and misdemeanors" (New York Times).[1]

In particular, the three researchers objected to what the then widespread practice of cherry-picking data and attempts to make insignificant results appear statistically credible, especially an approach for which they coined the term p-hacking in a 2014 paper.[2][3][4]

Apart from calling out faulty but presumably well-intended research practices, Data Colada has also published evidence of data manipulations and research misconduct, such as in the studies about the concept of moral high ground by psychologist Lawrence Sanna, and in research by Flemish psychologist Dirk Smeesters.[2] Following the findings of Data Colada, "[t]he two men’s careers came to an unceremonious end" (according to The New Yorker).[2]

An August 2021 post by Data Colada found that data from a field study in a 2012 PNAS paper[5] by Lisa L. Shu, Nina Mazar, Francesca Gino, Dan Ariely, and Max H. Bazerman had been fabricated.[6][7] All of the 2012 study's authors agreed with this assessment and the paper was retracted.[7] The study's authors also agreed that Dan Ariely was the only author to have had access to the data prior to transmitting it in its fraudulent form to Nina Mazar, the analyst.[6] Dan Ariely denied manipulating the data prior to forwarding it on to Mazar[8] but Excel metadata showed that he created the spreadsheet and was the last to edit it. He also admitted to having mislabeled all of the values in an entire column of the data in e-mail communication with Mazar that took place shortly after he initially sent her the data.[6][9] Dan Ariely has stated that someone at the insurance agency that provided the data must have fabricated it.[10][11][2]

The work of Data Colada has been credited with contributing to the start of the replication crisis in social sciences,[2] and with helping to establish better research practices such as the sharing of replication data.[3]

The Nobel-prize winning psychologist Daniel Kahneman described Data Colada in 2023 as "heroes of mine" and expressed his regret about having previously endorsed some research findings that the blog later showed to be faulty.[2] Brian Nosek of Center for Open Science applauded Data Colada for "hav[ing] done an amazing job of developing new methodologies to interrogate the credibility of research."[3]

On the other hand, as summarized by The New Yorker, "Data Colada’s harshest critics saw the young men as jealous upstarts who didn’t understand the soft artistry of the social sciences", such as psychologist Norbert Schwarz, who accused them and other reformers of engaging in a "witch hunt," or psychologist Daniel Gilbert who denounced the "replication police" as "shameless little bullies".[2]

In June 2023, Harvard Business School placed Francesca Gino on unpaid administrative leave after an internal investigation determined she had falsified data in her research.[12][13][14] Gino subsequently filed a defamation suit against Harvard, Harvard Business School Dean Srikant Datar, and the three members of Data Colada for $25 million, alleging that they had conspired to damage her reputation with false accusations, and that the penalties against her amounted to gender-based discrimination under Title IX.[14] The suit does not contest or refute the defendants' scientific findings. It asserts that since the researchers used inference and probability to argue that the anomalies between the original dataset and the data used in Gino's analyses could not have been due to random chance or benign error, and therefore occurred from fraudulent manipulation, the defendants cannot prove she in particular committed the scientific misconduct.[15]

• http://datacolada.org/