“officers were 12% more likely to stop a car occupied by a person of color during the day”

Link. A clever way to differentiate racism from differential arrest rates not necessarily directly related to racism.