Information Processing: Wall Street Journal: Moral Panic at MSU

Thursday, July 09, 2020

Wall Street Journal: Moral Panic at MSU

Professor Cesario and co-authors have retracted their PNAS paper Officer characteristics and racial disparities in fatal officer-involved shootings. I know that they had a difficult time with this decision. See further below for WSJ coverage.

My understanding from communication with Cesario is that he and his co-authors stand by the data and statistical analysis used in their paper. He also stands by the remarks in this Manifold (podcast) interview: Joe Cesario on Police Decision Making and Racial Bias in Deadly Force Decisions.

See this statement about the retraction from Cesario and co-author:

... One problem with such benchmarking approaches is that debate arises about whether it is more informative to compare the number of civilians shot to overall population proportions or to proxies for violent crime proportions. Indeed, one will obtain different results depending on what one thinks is the relevant comparison group: calculating P(shot|race) for the entire population will likely show evidence of anti-Black disparity, whereas calculating P(shot|race) for the pool of civilians who have violent interactions with the police will likely show no evidence of anti-Black disparity.

It is in this context that Johnson et al. (2019) was produced. Rather than debating which pool of civilians is the correct comparison group, we tested a less broad question that could be answered with current data: Is there a relationship between the race of officers and the civilians they fatally shot? However, this does not address the larger question of how race impacts the probability of being shot by police.

Cesario et al. maintain that the results described in the second paragraph above still stand: i.e., race of officer does not affect race of civilians shot. The issue is whether the result can be used to infer something about the conditional probability P(shot|race) discussed in the first paragraph. Other papers by Cesario, and his own broad conclusions from years of research in this area, suggest that P(shot|race) does NOT show the level of bias sometimes claimed by activists or in the media.

Note:

1. The PNAS paper that has been retracted is one out of several produced by Cesario and collaborators. In the podcast show notes we linked to Is There Evidence of Racial Disparity in Police Use of Deadly Force? Analyses of Officer-Involved Fatal Shootings in 2015–2016, which appeared in the journal Social Psychological and Personality Science. Cesario tells me that the following language from that paper still stands -- it is unaffected by the PNAS retraction:

When adjusting for crime, we find no systematic evidence of anti-Black disparities in fatal shootings, fatal shootings of unarmed citizens, or fatal shootings involving misidentification of harmless objects. Multiverse analyses showed only one significant anti-Black disparity of 144 possible tests. Exposure to police given crime rate differences likely accounts for the higher per capita rate of fatal police shootings for Blacks, at least when analyzing all shootings.

2. Some of their work is based on statistical analysis of police incident data, but other work is based on simulator studies of actual police officers under stress.

The small amount of funding that Professor Cesario received from my office in 2016 was for simulator studies (specifically, to cover production costs for realistic video of police stops). The request for funding was strongly endorsed by the Associate Dean for Research in the College of Social Science at MSU, and the College of Social Science matched the contribution from my office. The project was interdisciplinary, in collaboration with researchers in our School of Criminal Justice. Professor Cesario has also received National Science Foundation grants to study this topic. This 2018 award, Understanding Race Bias in the Decision to Shoot with an Integrated Model of Decision Making, is for $620k.

Obviously this is very important research and should not be politicized.

WSJ: The Proceedings of the National Academy of Sciences is a peer-reviewed journal that claims to publish “only the highest quality scientific research.” Now, the authors of a 2019 PNAS article are disowning their research simply because I cited it.

Psychologists Joseph Cesario of Michigan State and David Johnson of the University of Maryland analyzed 917 fatal police shootings of civilians from 2015 to test whether the race of the officer or the civilian predicted fatal police shootings. Neither did. Once “race specific rates of violent crime” are taken into account, the authors found, there are no disparities among those fatally shot by the police. These findings accord with decades of research showing that civilian behavior is the greatest influence on police behavior.

In September 2019, I cited the article’s finding in congressional testimony. I also referred to it in a City Journal article, in which I noted that two Princeton political scientists, Dean Knox and Jonathan Mummolo, had challenged the study design. Messrs. Cesario and Johnson stood by their findings. Even under the study design proposed by Messrs. Knox and Mummolo, they wrote, there is again “no significant evidence of anti-black disparity in the likelihood of being fatally shot by the police.”

My June 3 Journal op-ed quoted the PNAS article’s conclusion verbatim. It set off a firestorm at Michigan State. The university’s Graduate Employees Union pressured the MSU press office to apologize for the “harm it caused” by mentioning my article in a newsletter. The union targeted physicist Steve Hsu, who had approved funding for Mr. Cesario’s research. MSU sacked Mr. Hsu from his administrative position. PNAS editorialized that Messrs. Cesario and Johnson had “poorly framed” their article—the one that got through the journal’s three levels of editorial and peer review.

Mr. Cesario told this page that Mr. Hsu’s dismissal could narrow the “kinds of topics people can talk about, or what kinds of conclusions people can come to.” Now he and Mr. Johnson have themselves jeopardized the possibility of politically neutral scholarship. On Monday they retracted their paper. They say they stand behind its conclusion and statistical approach but complain about its “misuse,” specifically mentioning my op-eds.

The authors don’t say how I misused their work. Instead, they attribute to me a position I have never taken: that the “probability of being shot by police did not differ between Black and White Americans.” To the contrary, I have, like them, stressed that racial disparities in policing reflect differences in violent crime rates. The only thing wrong with their article, and my citation of it, is that its conclusion is unacceptable in our current political climate.

This retraction bodes ill for the development of knowledge. If scientists must disavow their findings because they challenge reigning orthodoxies, then those orthodoxies will prevail even when they are wrong. Political consensus will drive scholarship, and not the reverse. The consequences for the policing debate are particularly dire. Researchers will suppress any results that contravene the narrative about endemic police racism. That narrative is now producing a shocking rise in shootings in American cities. The victims, including toddlers, are almost exclusively black.

Ms. Mac Donald is a fellow at the Manhattan Institute.