Moskos vs. Pro Publica: Deadly Force In Black and White

   Rosenthal accuses Moskos of using bad data and engaging in statistical shenanigans. He says that the Pro Publica analysis is better. Moskos responds...and I think he's clearly right. Pro Publica has, for one thing, either cherry-picked data or accidentally chosen the same data that someone would choose were they cherry-picking it.
   Independently of what the actual data look like, the two sides of this dispute ought to be able to come to some agreement about what kinds of adjustments are reasonable here. I'm in no way confident that Moskos's approach is right--but there can be no doubt that his arguments provide very strong support for the conclusion that the unadjusted rates do not prove what they allege to prove. At minimum, his arguments show that it is an error to appeal to the unadjusted rates as proof of something like the prejudice hypothesis.