Scientific versus public debates: A PNAS case study

Unfortunately, scientific communication and the resulting public use of research often do not reflect the painstaking and sometimes imperfect process of peer review, and in a hyperpartisan landscape invalid conclusions can acquire a tendentious life of their own. Here we review the process by which a controversial PNAS paper was published, the correction to errors that were uncovered after publication, and how the scientific process missed the original error but ultimately led to a Correction.*We’re providing this explanation now because the July 2019 paper (1), which purportedly focused on the degree of racial bias documented in officer-involved shootings, has garnered new attention in light of the recent tragedy involving George Floyd and the massive protests that have followed. A June 2, 2020, Wall Street Journal op-ed (2) cited the PNAS article but failed to provide context or note the original paper’s unjustified supposition. Elucidating the genesis of this PNAS paper, and its misstatements, should help to set the record straight. We think doing so is essential so that future scholarship—and future efforts in science communication related to data on police activities—can avoid similar pitfalls on these sensitive and important issues.Scientific publication operates via a peer review process in which submitted papers are assessed by knowledgeable specialists who critically evaluate the validity of the research, the reliability of the data, and the soundness of the conclusions drawn from them. As in any human endeavor, the peer review process is not perfect, and sometimes ill-founded results and conclusions are published. However, publication in a peer-reviewed journal subjects the findings to even wider critical scrutiny by scientists who are free to challenge the paper’s methods and conclusions and correct the scientific record.Last July, PNAS published an article titled “Officer characteristics and racial disparities in … [↵][1]1To whom correspondence may be addressed. Email: mcw{at}wjh.harvard.edu. [1]: #xref-corresp-1-1