Quantifying Bias in Driving‐Under‐The‐Influence Enforcement

Abstract

As the marginal suspect and propensity to commit crime are unobserved across racial categories, it is difficult to quantify racial bias with law enforcement outcomes data. We test for racial bias in driving‐under‐the‐influence of alcohol enforcement. The assessment outcome variable, blood alcohol content (BAC), provides information about motorist's choices and allows for a more refined test for racial bias compared with other law enforcement outcome data. We find no evidence of racial bias in the relevant range where our model applies. However, we do find differences in find rates at the lowest levels of BAC, where there should be no impairment.

JEL Classification: J7, K42