Another example of a bad policing data. This number was generated by treating each category as mutually exclusive but they shouldn’t be. For example, a police-related death can simultaneously be a “traffic stop,” a “violent crime,” and someone with a “mental illness.” 1/2 https://twitter.com/roderickgraham/status/1382051014518435845

2/2 If you assign all the categories that apply and don’t treat them as mutually exclusive the number of deaths that begin as traffic stops is much, much higher. Not Rod’s data or fault - most of the published research I read that includes this variable makes the same mistake.

Using this logic, none of the incidents coded as “violent crime” involved a “person with a weapon” and none of the incidents coded as “person with a weapon” involved a “violent crime.”