I think there are a few ways to design studies to try to address this, depending on what data is available and what we’re willing to treat as proxies for what:
- If we have access to a lot of body camera footage, we could try score incidents to determine if suspect behavior really does account for the difference. If we want to be super rigorous about it, we could replace people with wire-frames or otherwise mask race prior to scoring. Either way, this would be hard and incredibly time consuming. It also depends on the quality of body cam data, which right now doesn’t seem that great; officers often have discretion to turn off their body cameras, and footage seems to go missing when convenient (but here I may be overrelying on anecdote).
- We could look at how often resisting arrest charges are filed when a body camera is present vs. when one is not. We could also look at racial disparities in cases where a body camera is present vs. when one is not. We could look at trends in arrest statistics in areas as they roll out body cameras and that shapes officer behavior. We could look at how often cases are dismissed when a body camera is present, and any racial disparity there.
- We could compare resisting arrest statistics between wealthy and poor areas, and (if data is available), between wealthy and poor defendants. There, wealth or income would act as a proxy for self-control, and we could see what if any difference that makes on race disparities.
- We could control for intoxication, since intoxication is likely to be a good proxy for lack of self-control. If it’s an issue of behavior and cooperation, we should expect racial disparities to be reduced when comparing intoxicated black suspects to intoxicated white suspects, since we’d expect those populations to be both pretty uncooperative.
But I think a better question is what to do since we don’t have any of these studies in front of us. I’m saying, let’s turn to the studies we do have, and try to make sure our hypotheses are consistent.
What question are you looking to answer here? If a black person murders someone and a white person doesn’t, you can bet there will be some disparate police treatment, since murder is illegal and not-murder isn’t. That doesn’t tell us anything about race. Again, we need to compare like to like.
The “first paragraph” is the abstract, you need to look at that part of the report. From page 17:
I don’t agree that that’s how the report should be interpreted. The report is trying to compare like to like, and it recognizes that the criminal justice system as experienced by women is different from the criminal justice system as experienced by men. That’s clear just from the proportion of incarcerated people who are men (more than 80%, according to this report), but it is probably also reflected in the types of crimes, the lengths of sentences, and the factors that are considered. There may not be enough data on women to compare like to like for black women and white women. And in any case, looking at how race affects men is looking at how race affects the substantial majority of prisoners.