On the Validity of Arrest as a Proxy for Offense: Race and the Likelihood of Arrest for Violent Crimes

Arrest data may not accurately reflect actual crime rates: The paper finds that arrests for violent crimes don't directly measure whether someone actually committed an offense, since not all crimes result in arrest. This is critical because risk assessment tools used in courts rely on arrest records as a proxy for actual criminal behavior.

Racial disparities in arrest rates vary significantly by crime type and location: The study of 16 US states (2007-2016) reveals that the gap between Black and White arrest rates for violent offenses is not consistent—it depends on the specific type of crime and geographic location. This means a single risk assessment tool may produce different racial biases depending on where it's used.

Using arrest data in risk assessment tools can amplify existing criminal justice biases: Because arrest rates themselves show racial disparities, algorithmic risk assessment instruments that use arrests as their primary measure may systematically overestimate risk for Black individuals, potentially leading to harsher decisions at bail, sentencing, and parole stages.

Current validation practices for risk assessment tools may be insufficient: Even though researchers have focused on violent crime arrests as supposedly more reliable indicators than other arrests, this paper demonstrates that violent arrest data still contains significant racial disparities and geographic inconsistencies that undermine their validity as objective measures.