Court Clears Researchers of Defamation For Identifying Manipulated Data (arstechnica.com) 21
An anonymous reader quotes a report from Ars Technica: Earlier this year, we got a look at something unusual: the results of an internal investigation conducted by Harvard Business School that concluded one of its star faculty members had committed research misconduct. Normally, these reports are kept confidential, leaving questions regarding the methods and extent of data manipulations. But in this case, the report became public because the researcher had filed a lawsuit that alleged defamation on the part of the team of data detectives that had first identified potential cases of fabricated data, as well as Harvard Business School itself. Now, the court has ruled (PDF) on motions to dismiss the case. While the suit against Harvard will go on, the court has ruled that evidence-backed conclusions regarding fabricated data cannot constitute defamation -- which is probably a very good thing for science.
The researchers who had been sued, Uri Simonsohn, Leif Nelson, and Joe Simmons, run a blog called Data Colada where, among other things, they note cases of suspicious-looking data in the behavioral sciences. As we detailed in our earlier coverage, they published a series of blog posts describing an apparent case of fabricated data in four different papers published by the high-profile researcher Francesca Gino, a professor at Harvard Business School. The researchers also submitted the evidence to Harvard, which ran its own investigation that included interviewing the researchers involved and examining many of the original data files behind the paper. In the end, Harvard determined that research misconduct had been committed, placed Gino on administrative leave and considered revoking her tenure. Harvard contacted the journals where the papers were published to inform them that the underlying data was unreliable.
Gino then filed suit alleging that Harvard had breached their contract with her, defamed her, and interfered with her relationship with the publisher of her books. She also added defamation accusations against the Data Colada team. Both Harvard and the Data Colada collective filed a motion to have all the actions dismissed, which brings us to this new decision. Harvard got a mixed outcome. This appears to largely be the result that the Harvard Business School adopted a new and temporary policy for addressing research misconduct when the accusations against Gino came in. This, according to the court, leaves questions regarding whether the university had breached its contract with her. However, most of the rest of the suit was dismissed. The judge ruled that the university informing Gino's colleagues that Gino had been placed on administrative leave does not constitute defamation. Nor do the notices requesting retractions sent to the journals where the papers were published. "I find the Retraction Notices amount 'only to a statement of [Harvard Business School]'s evolving, subjective view or interpretation of its investigation into inaccuracies in certain [data] contained in the articles,' rather than defamation," the judge decided.
More critically, the researchers had every allegation against them thrown out. Here, the fact that the accusations involved evidence-based conclusions, and were presented with typical scientific caution, ended up protecting the researchers. The court cites precedent to note that "[s]cientific controversies must be settled by the methods of science rather than by the methods of litigation" and concludes that the material sent to Harvard "constitutes the Data Colada Defendants' subjective interpretation of the facts available to them." Since it had already been determined that Gino was a public figure due to her high-profile academic career, this does not rise to the standard of defamation. And, while the Data Colada team was pretty definitive in determining that data manipulation had taken place, its members were cautious about acknowledging that the evidence they had did not clearly indicate Gino was the one who had performed the manipulation. Finally, it was striking that the researchers had protected themselves by providing links to the data sources they'd used to draw their conclusions. The decision cites a precedent that indicates "by providing hyperlinks to the relevant information, the articles enable readers to review the underlying information for themselves and reach their own conclusions."
The researchers who had been sued, Uri Simonsohn, Leif Nelson, and Joe Simmons, run a blog called Data Colada where, among other things, they note cases of suspicious-looking data in the behavioral sciences. As we detailed in our earlier coverage, they published a series of blog posts describing an apparent case of fabricated data in four different papers published by the high-profile researcher Francesca Gino, a professor at Harvard Business School. The researchers also submitted the evidence to Harvard, which ran its own investigation that included interviewing the researchers involved and examining many of the original data files behind the paper. In the end, Harvard determined that research misconduct had been committed, placed Gino on administrative leave and considered revoking her tenure. Harvard contacted the journals where the papers were published to inform them that the underlying data was unreliable.
Gino then filed suit alleging that Harvard had breached their contract with her, defamed her, and interfered with her relationship with the publisher of her books. She also added defamation accusations against the Data Colada team. Both Harvard and the Data Colada collective filed a motion to have all the actions dismissed, which brings us to this new decision. Harvard got a mixed outcome. This appears to largely be the result that the Harvard Business School adopted a new and temporary policy for addressing research misconduct when the accusations against Gino came in. This, according to the court, leaves questions regarding whether the university had breached its contract with her. However, most of the rest of the suit was dismissed. The judge ruled that the university informing Gino's colleagues that Gino had been placed on administrative leave does not constitute defamation. Nor do the notices requesting retractions sent to the journals where the papers were published. "I find the Retraction Notices amount 'only to a statement of [Harvard Business School]'s evolving, subjective view or interpretation of its investigation into inaccuracies in certain [data] contained in the articles,' rather than defamation," the judge decided.
More critically, the researchers had every allegation against them thrown out. Here, the fact that the accusations involved evidence-based conclusions, and were presented with typical scientific caution, ended up protecting the researchers. The court cites precedent to note that "[s]cientific controversies must be settled by the methods of science rather than by the methods of litigation" and concludes that the material sent to Harvard "constitutes the Data Colada Defendants' subjective interpretation of the facts available to them." Since it had already been determined that Gino was a public figure due to her high-profile academic career, this does not rise to the standard of defamation. And, while the Data Colada team was pretty definitive in determining that data manipulation had taken place, its members were cautious about acknowledging that the evidence they had did not clearly indicate Gino was the one who had performed the manipulation. Finally, it was striking that the researchers had protected themselves by providing links to the data sources they'd used to draw their conclusions. The decision cites a precedent that indicates "by providing hyperlinks to the relevant information, the articles enable readers to review the underlying information for themselves and reach their own conclusions."
God bless America (Score:5, Insightful)
In America, the truth is an absolute defense against defamation.
True, but misleading (Score:5, Informative)
In America, the truth is an absolute defense against defamation.
True, but misleading in this case. They gave no absolute proof that Gino modified the data, only strong statistical evidence that the data was modified, in a way that validates the conclusions of the paper (unmodified data shows no such validation).
What the Data Colada people did was analyze the data and show a lot of anomalies, the anomalies show strong evidence of scientific malpractice, and they published this evidence (and the analysis) on their website and alerted Harvard of the issues.
Their website is an interesting read, and gives a good introduction to some really good applications of the scientific method.
And of note, on their website they analyze several other papers as well. In other examples the author will cooperate with the Data Colada people giving them raw data, agreeing with them that the data was mamipulated or the result calculated incorrectly, that sort of thing. The DC people are quick to point out when authors show good integrity during this process. (IIRC, in one instance it was a lab assistant that was modifying the data, not the lead author.)
There was one study, very *very* highly cited, where the researchers sent out resumes with black-sounding names and white-sounding names, counted the number of callbacks for each, and found that applicants with white-sounding names got more callbacks than black-sounding names.
Everyone and their dog cited this study as evidence of racism in America.
A 2nd study did not reproduce result.
The DC people set out to find why, and discovered that there are poor-sounding names and affluent-sounding names (for both black and white applicants), figured out a way to measure this on a per-name basis, and discovered that the original study had inadvertently used rich-sounding names for whites, and poor-sounding names for blacks.
When this feature is accounted for, the apparent racism (in this specific measure) disappears (their analysis [datacolada.org].
All of this is explained in clear and interesting detail on their website [datacolada.org].
Highly recommended reading!
Re:True, but misleading (Score:4, Interesting)
The lower callback rates for Jamal and Lakisha in the classic 2004 AER paper, and the successful replications mentioned earlier, are as consistent with racial as with SES discrimination. The SES account parsimoniously also explains this one failure to replicate the effect. But this conclusion is tentative as best, we are comparing studies that differ on many dimensions (and the new study had some noteworthy glitches, read footnote 4). To test racial discrimination in particular, and name effects in general, we need the same study to orthogonally manipulate these, or at least use names pretested to differ only on the dimension of interest. I don't think any audit study has done that.
You are not quite right with your conclusion. It's just that the earlier study did not look into the socioeconomic status (SES) as well, hence it did not catch this correlation.
Re: True, but misleading (Score:2)
Re: (Score:2)
They didn't say that the data was falsified by her, only that the data looked falsified
In the real world, what is the difference? Isn't it splitting hairs?
Re: True, but misleading (Score:3)
Re: (Score:1)
So it's not racism, it's just classism? Is there a meaningful difference?
It reminds me of the literacy test that disproportionately failed a certain racial demographic, [slate.com] proving that written tests can be racist.
Or zoning laws that replaced [kqed.org] redlining, with similar results.
So in the end, it seems pointless to adjust for SES when picking names. Just randomly select a bunch of black people and a bunch o
Re:God bless America (Score:4, Insightful)
In America, the truth is an absolute defense against defamation.
Don't forget you also need boatloads of money to hire lawyers to defend you in court for the court to recognize you were telling the truth in the first place.
The average Joe on the street, who most likely cannot come up with $1000 in an emergency, won't get such chance when sued.
Start looking at grievence study journals (Score:1, Troll)
Need to look at many scientifically invalid "self-reported survey of X women" get passed of as research in grievance journals.
And the 'taken as gospel fact' opinion pieces from social sciences agenda pushing books where large amounts of 'data' is just self-reported surveys of 25 people and/or 'took 0.0000001% of tweets from 10,000,000 tweet database' and therefore 'proved' that the basic group sending the tweets is 'racist', 'misogynist', 'radicalized', dangerous, etc.
Or using google sentiment analysis serv
End goal (Score:2)
Keeping one half of the population scared, a dependable voting block for more social spending, and a championing force for 'shopping is therapy' and overconsumption.
I don't understand (Score:2)
Re: (Score:3)
This is exactly how it should have been.
Re:I don't understand (Score:5, Insightful)
Re: I don't understand (Score:2)
Re: (Score:2)
We are issuing this expression of concern in consultation with the publisher to fulfil their reporting obligation regarding the publication [1] mentioned above.
One of the authors has raised concerns regarding the methodology employed in the study, the conclusions drawn and the insufficient consideration of laboratory staff and resources.
In order to keep the highest scientific standards, an in-depth investigation is initiated by the responsible editors together with the journal’s editorial office in collaboration with the editorial board, and in accordance with the Committee on Publication Ethics (COPE) guidance. The article will be updated and any necessary corrections made at the conclusion of the investigation process.
Maybe could've sued for defamation (though good luck going against Big Pharma)? Or maybe when expressing concern they shouldn't
Re: (Score:2)
In science, there's often two opposing viewpoints, the covid vaccine for example. If we go down the path that you cannot say someone's specific work is wrong, how are we ever going to arrive at the truth? In the past, if a study is called into question, someone will submit another journal article questioning it ("expression of concern"). Like this https://www.ncbi.nlm.nih.gov/p... [nih.gov]
We are issuing this expression of concern in consultation with the publisher to fulfil their reporting obligation regarding the publication [1] mentioned above.
One of the authors has raised concerns regarding the methodology employed in the study, the conclusions drawn and the insufficient consideration of laboratory staff and resources.
In order to keep the highest scientific standards, an in-depth investigation is initiated by the responsible editors together with the journal’s editorial office in collaboration with the editorial board, and in accordance with the Committee on Publication Ethics (COPE) guidance. The article will be updated and any necessary corrections made at the conclusion of the investigation process.
Maybe could've sued for defamation (though good luck going against Big Pharma)? Or maybe when expressing concern they shouldn't use aggressive words like manipulated?
In the current, very few researchers have the time, money, or expertise to issue such questioning papers. Hell, "peer review" becomes just grammar/spell check because this research is so far down a rabbit hole that THERE ARE NO PEERS.
Why it Pays to Break the Rules at Work and in Life (Score:1)
The professor's 2018 book is pretty clear: "Rebel Talent: Why it Pays to Break the Rules at Work and in Life Paperback." Got away with it for six years!
Re: (Score:3)
The corporate overlords told us to read the book but I didn't, because, well, following their corporate instructions wouldn't be very rebellious, would it?
Sigh (Score:2)
"The truth cannot be libelous" is a pretty well-held facet of any English law-based system.
This was pretty much just a vexatious lawsuit.
Re: (Score:3)
IIUC, not in Britain.
This is a miracle (Score:2)
> [s]cientific controversies must be settled by the methods of science rather than by the methods of litigation
This should be headline news.