(TL;DR - the rationale for retraction was indeed dubious.)
Here’s what happened. Lots of people want to assess the probability of getting shot by a cop based on your race. The problem is, this calculation involves both shootings and uneventful encounters, but only the former have solid data.
Johnson et al, using this better data, try to predict the race of the cop from all other aspects of the shooting. Basically, checking if unarmed black men are getting shot mostly by white cops. Turns out that’s not the case.
Is this “flawed?” No, it’s answering another question, trying to make good with the available data. Problem is, their abstract said “White officers are not more likely to shoot minority civilians than non-White officers”, which is imprecise wording. If it said, “White officers are not more likely to
be the shooters of minority...” then they’d be fine.
Opponents pounced on this, and made it sound like the entire paper is engulfed in some grave mathematical error. A few months ago, the authors went ahead and
corrected their abstract. From a few days ago, their mealy-mouthed
statement of retraction cites misuse of their work, but finds blame only in the already-corrected material.
Meanwhile, Dr. Joseph Fair, self-styled
@curefinder, virologist and epidemiologist, appeared on NBC multiple times to discuss his recovery from COVID-19. Except, by his own admission, all his PCR tests were negative, as was his antibody test.
Somehow, the statistics paper is retracted, while Fair’s blatant lies are not.