A test for implicit bias in discretionary criminal justice decisions.

Objective: Our goal was to develop a framework to test for implicit racial bias in discretionary decisions made by community supervision agents in conditions with increasing information ambiguity.

Hypotheses: We reasoned that as in-person contact decreases, community supervision officers' specific knowledge of clients would be replaced by heuristics that lead to racially disproportionate outcomes in higher discretion events. Officers' implicit biases would lead to disproportionately higher technical violation rates among Black community corrections' clients when they have less personal contact, but we expected no analogous increase in nondiscretionary decisions.

Method: Using data from Black and White clients entering probation and postrelease supervision in North Carolina from 2012 through 2016, we estimated the difference in racial disparities in discretionary versus nondiscretionary decisions across five levels of supervision. We evaluated the robustness of our main fixed-effects model using an alternative regression discontinuity design.

Results: Racial disparities in discretionary decisions grew as supervision intensity decreased, and the bias was larger for women than men. There was no similar pattern of increased disparity for nondiscretionary decisions.

Conclusions: Criminal justice system actors have a great deal of discretion, particularly in how they deal with less serious criminal behavior. Although decentralized decisions are foundational to the function of the criminal justice system, they provide an opportunity for implicit bias to seep in. Shortcuts and mental heuristics are more influential when the decision-maker's mental resources are already strained-for instance, if someone is tired, distracted, or overworked. Therefore, limiting discretion and increasing oversight and accountability may reduce the impact of implicit bias on criminal justice system outcomes. (PsycInfo Database Record (c) 2023 APA, all rights reserved).