Read on Twitter
Sometimes @FiveThirtyEight, we go looking for data about a problem and end up finding an entirely different story ... because the data that's tracked turns out to not actually be the data we need.
That happened with this story on police reform ... https://fivethirtyeight.com/features/is-police-reform-a-fundamentally-flawed-idea/
That happened with this story on police reform ... https://fivethirtyeight.com/features/is-police-reform-a-fundamentally-flawed-idea/
I was trying to figure out what police do well (i.e., if we break emergency response apart to diff. jobs, what should be left to police?). @ameliatd was trying to figure out which police reforms worked.
We both ran into the same problem ... what is "well"? What is "works"?
We both ran into the same problem ... what is "well"? What is "works"?
Turns out, the only stats we have to work with are crime reports and formal police complaints. Neither of which tell anyone much about how police are treating residents of highly policed neighborhoods, whether those citizens feel their needs are met, whether/who feels safe, etc..
Even the crime stats are biased based on what the police see and what gets reported to them. And that really is a bias. When researchers in Seattle compared drug abuse hotspots from police reports vs. those identified by EMT response ... there wasn't a ton of overlap.
The data doesn't tell us what we actually want to know. It's not designed to. It's designed around assumption that what police care about and define as "safety" is the same as the people they serve. So what do you do when you know that's not true? https://fivethirtyeight.com/features/is-police-reform-a-fundamentally-flawed-idea/
