Quote (thesnipa @ 29 Apr 2021 14:34)
are you saying controlled for the proportions of violent crime arrests*, divided by the deaths by police, the resulting fractions are similar?
if so doesnt that make sense in a population who's over-policed and profiled being compared to one that isnt?
cops find maybe 10% of criminals, its about where they look. to simplify it.
problem is that cops can overpolice a certain neighborhood for two different reasons which are really hard to disentangle: bias against the residents of the area, and the area being a genuine crime hotspot.
it's like... yeah, no shit, cops will look more closely in the ghettos than in a posh suburb with accurately cut lawns.
From the top of my head, the most sensible approach would be the following: set up a regression model which predicts the number of violent crime arrests in a place based on socio-demographic variables like education level, income, urban vs rural and so on. Train this model with data from countries with a more homogeneous population and/or less history of racial discrimination than the U.S. to get an estimate which is as free from racial and classist bias as possible. Then use this model to predict the number of violent crime arrests we would ordinarly expect in certain places in America and compare this baseline against the actual number and quality of reported crime. If necessary, we can add a global multiplier to account for the "arrest-happiness" of various countries' respective police force... The disparity should give us a crude estimate about the magnitude of bias in U.S. policing.