Police oversight is my research area, and none of the data is good. Like I said, even when it *is* captured on film, prosecutions and subsequent convictions are extremely rare. The issue isn't so much that there is no evidence, it's that the public refuses to convict cops when their cases make it to trial (hardly ever). Video evidence isn't enough to sway that, particularly when you factor in the testilying of additional officers.
We have four major interventions right now - BWC, anti-bias training, deescalation training, and civilian oversight. None of them are research or evidence-based and they produce mixed results at best and negative results at worst.
Anti-bias training often results in a rebound effect where what was once implicit becomes explicit.
The only one that shows the smallest bit of promise is civilian oversight. But that kind of relies on them being fully independent and autonomous. Here in NYC, ours is not. The police union has had a lot of sway in establishing regulations and funding for the CCRB. They have an overwhelming caseload with low-paid positions usually filled by people fresh out of college who stand no chance when interviewing police and facing off against their union lawyers.
The issue with most policing interventions is that they are "common sense." None of them are based on research and data. People are trying to implement common sense policies to a highly illogical situation instead of taking the time to research what will *actually work*.
This is why you see the push to de-fund and abolish policing as we know it. It is a wholly failed social experiment and the only way to reform it is to rebuild it from the ground up.