zlacker

I think many of the mathematicians would agree with:

>This "racial bias" is composed of data collected from crime statistics.

They disagree that crime statistics are reflective of where crime occurs. They are more reflective of where police officers are and where crimes are most easily spotted. For example, the racial disparities in policing crack cocaine versus powder cocaine.

Predictive systems that read in biased data will produce biased data.

>>wdevan+(OP)
> They are more reflective of where police officers are and where crimes are most easily spotted.

The article says: MacDonald argues that PredPol uses only crimes reported by victims, such as burglaries and robberies, to inform its software. “We never do predictions for crime types that have the possibility of officer-initiated bias, such as drug crimes or prostitution,” he says.

>>wdevan+(OP)
I would guess that the murder rate data is pretty complete.

>>lazyjo+J1
That doesn't mean reports are completely independent. Isn't it obvious that the perception of whether the police will do anything is going to affect the reporting?

>>perl4e+Ua
Wouldn't that affect it in an inhibitory way, though? Black areas would be reporting less crime than actually occurs- so if reported crime rates there are still higher, that tells us something?

>>antepo+ka1
Just as a thought experiment, it could be that lesser crimes are not reported because the police don't take them seriously, which leads to more criminals committing major crimes, which are mostly reported.

By lesser crimes I don't mean trivial things that provide an excuse to harass non-criminals and give "broken window" policing a bad name, but actual crimes which aren't major violent crimes.