Aggregate racial disparities in prosecution can reverse direction when filtered by offense category

Evidence

This overall pattern favoring minority defendants does not hold so well when examining data by offense categories. For example, the favorable case filing rate for Black defendants (34 few cases per 1,000) disappears when we focus on “person offenses” only. Also, while overall Hispanics were more advantaged at charge reduction at filing (8 more cases per 1,000), for “person offenses” they were least likely to experience this outcome (33 fewer cases per 1,000). Similarly, while Hispanics were least likely to experience a custodial sentence (18 fewer cases per 1,000) for all offense categories, as shown in figure 4.1, for “person offenses” they were most disadvantaged (34 more cases per 1,000 received a custodial sentence as compared to Whites). — Joseph Avery, Jim Sidanius, Joel Cooper, Gregory Davis, Amanda N., Bias in the Law, 584

Categorization

• SEED: An article angle: ‘Simpson’s paradox in the criminal justice system’ — how aggregate statistics on racial bias can hide or even invert the true pattern within specific offense categories, and what that means for policy debates built on headline numbers.