Criminology professor leaves after being accused of falsifying data

[marshwiggle]

Florida State University criminology professor leaves after being accused of falsifying data

A criminology professor at Florida State University suddenly left his $190,000-a-year position after being accused of fudging data to make racism seem more common.

Eric Stewart has had six of his studies - one of which dates as far back as 2006 - retracted amid allegations that he fabricated data by altering sample sizes.

Stewart has denied such criticisms but after the sixth incident in 2020 an FSU committee reconvened to address his record and findings. 

Since last month, the professor has been absent, perhaps marking the end of his 16-year career at the school and a years-long investigation into his academic malpractice, the Florida Standard reported.

[Parasaurolophus]

There's a strong irony here, given all the junk science in criminology.

[Sun_Worshiper]

The pressure to publish is intense and some people cross ethical lines. Good example of why replication studies, data sharing, and pre-registration are important.

[marshwiggle]

The pressure to publish is intense and some people cross ethical lines. Good example of why replication studies, data sharing, and pre-registration are important.

Publishing should require the same amount of effort regardless of what story the data tell. "Crossing ethical lines" is about making the published results sensational.

[Puget]

Something is clearly broken at FSU-- it isn't clear exactly when the co-author first raised concerns, but it was about a paper published in 2011, so probably shortly thereafter. This wasn't some vague allegation-- the co-author KNEW what the actual data were, yet the case was heard by a group of faculty that included 3 co-authors with the accused in direct violation of CoI policies, and nothing was done until 6 papers were retracted!

The pressure to publish is intense and some people cross ethical lines. Good example of why replication studies, data sharing, and pre-registration are important.

I agree these are all super important, but data sharing and pre-registration, while they guard against other bad research practices, aren't usually able to detect outright data fabrication, at least if the fabricator is sophisticated about it and generates plausible artificial data.

The best guard against outright fabrication is that, since research involves a lot of people, one or more of the others involved are likely to notice anomalies or suspicious behavior at some point and raise questions (group frauds seem surprisingly rare). Yet in this case that did happen, but nothing was done for more than a decade. And in other famous cases of fraud, it took a long time because the others involved were junior (grad students, postdocs) and afraid of questioning their PI.

[Sun_Worshiper]

The pressure to publish is intense and some people cross ethical lines. Good example of why replication studies, data sharing, and pre-registration are important.

Publishing should require the same amount of effort regardless of what story the data tell. "Crossing ethical lines" is about making the published results sensational.

Maybe publishing should require the same amount of effort, but the reality is that null findings are difficult to publish in top journals.

[marshwiggle]

The pressure to publish is intense and some people cross ethical lines. Good example of why replication studies, data sharing, and pre-registration are important.

Publishing should require the same amount of effort regardless of what story the data tell. "Crossing ethical lines" is about making the published results sensational.

Maybe publishing should require the same amount of effort, but the reality is that null findings are difficult to publish in top journals.

That has nothing to do with the effort required to do the research. And if "null findings" in certain fields are indications that we're not living in a dystopian nightmare, then that says more about those fields than it about society.

[ab_grp]

I remember when this story broke a couple years ago, and his former colleague and co-author Pickett was treated pretty poorly for daring to question the data (and Stewart).  The Chronicle article goes into detail about all of these shenanigans and the complete lack of accountability.  https://www.chronicle.com/article/the-criminologist-accused-of-cooking-the-books   Pickett's lengthy explanation of the issues is also included in the Daily Mail article in the original thread post.  The EIC of the journal Criminology apparently did not think they were worth pursuing and, further, that Pickett was acting in bad faith out of a personal vendetta against Stewart.  Possibly there was less scrutiny applied or more hesitancy to do so because of the subject of the research and races of the accused and accuser.  I hope it's not the case at this point that Stewart will just disappear into the night and face no further repercussions.  I would think this would affect his career, but nothing prior has, apparently.

[Sun_Worshiper]

The pressure to publish is intense and some people cross ethical lines. Good example of why replication studies, data sharing, and pre-registration are important.

Publishing should require the same amount of effort regardless of what story the data tell. "Crossing ethical lines" is about making the published results sensational.

Maybe publishing should require the same amount of effort, but the reality is that null findings are difficult to publish in top journals.

That has nothing to do with the effort required to do the research. And if "null findings" in certain fields are indications that we're not living in a dystopian nightmare, then that says more about those fields than it about society.

My point isn't about how much effort it takes to do research, it is about the pressure to publish leading people to manipulate their data or just p-hack it to death in search for statistical significance - the file drawer problem.

[Mobius]

Can't top Alice Goffman's dissertation/book.