With a broad brush: a response to Reingle Gonzalez et al., 2018

We recently read a study¹ that presents an analysis of officer-involved shootings in relation to military history and deployment. Although the results presented are not incorrect, per se, they are limited by a lack of inclusion of confounding variables and mediators and result in an overstating of the conclusions by authors. To state simply that officers with a military history are more likely to be involved in discharging their weapons on the job (and providing a review of literature on Veterans and violence as the introduction) mischaracterizes this nuanced relationship and purports a homogeneity among Veterans that does not exist. There is a concern that the conclusion would be read as generalizable and that the limitations to this study would not have been made clear; indeed, only twelve days after this study was published, The Marshall Project ran a piece citing this study titled ‘Police with Military Experience More Likely to Shoot’².

At the outset, the introductory paragraph is concerning in its references to violence and aggressive behavior without contextualizing prior findings regarding military history and adverse outcomes. This is a gross oversimplification of the relationship between Veterans and violence and presents an assumed and implicit linkage between officer-involved shooting and violence. Yes, there has been public scrutiny around use of force and officer-involved shootings, but are all cases of shootings unjustified? Are there not instances when discharging one’s weapon would be warranted? The authors themselves state that in the majority of these cases, ‘officers were likely justified in using deadly force’^1(p2) – and this study’s operationalization of ‘discharging a firearm in the line of duty’ does not make clear that all instances are the same and should therefore warrant being the same outcome to be predicted.

The authors later state the need for identifying specific aspects of exposure that may put individuals at higher risk of shootings, and correctly identify that this would provide critical direction for future research, policy, and practice. However, subsequent analyses reveal that only the law enforcement job description (e.g. SWAT) and military status (Veteran [non-deployed, deployed]) are examined, and these are conflated with measures of exposure. The DD214 provides additional information which could be pertinent predictors, including MOS/AFSC, character of service, time since discharge, and paygrade (not rank type; as there are qualitative and quantitative differences between E1-E4 and E5-E9), but these variables are not included in the analyses.

Equally problematic, given the mention of hiring benefits and Veterans’ preference legislation, is that only ten percent of deployed Veterans in the study were post-9/11 (a total of 5), but this was not contextualized more explicitly for the reader. Post-9/11 veterans stand to be the most impacted given the number transitioning from service. Finally, the authors correctly state that the Pew results are expanded upon by controlling for law enforcement job description. However, the tripartite definition of military experience is still a crude estimate, resulting in a reification of inaccurate stereotypes. We implore the scientific community and public at large to carefully consider the potential for negative consequences when sharing results, particularly pertaining to groups who are already facing a host of stigmas and challenges like our military Veterans and their families.