Bad apples and incredible certitude

Chalfin and Kaplan attend to the problem of police misconduct with a series of simulation analyses that leverage data on complaints and uses of force in the Chicago Police Department. They conclude that incapacitating officers has minimal effects on misconduct and that, given political constraints, policy makers may prefer broader reforms around accountability and management to removing "bad apples". In this comment, we argue that this conclusion and its policy implications are characterized by "incredible certitude" driven by a selective focus on a subset of their full simulation results and inadequate incorporation of network spillovers into their analysis.

Contrary to CK's conclusions, their preferred estimates of the effect of incapacitating "bad apples" on misconduct are squarely within the range of other interventions aimed at reducing police complaints and use of force. Once network spillovers are accounted for, estimates are up to five times as large. We conclude with a discussion of how even small reductions in misconduct can have outsized benefits as measured in both dollars and human suffering, and argue that the removal of problem officers is a normative good that should be pursued on moral grounds.