There's a new study that's made the rounds of major papers over the last week, because it seems to show that there's racial bias in policing outcomes in departments across America, *except* in shootings. However, as the author of this blog post points out (someone who actually read the paper, unlike most of the journalists who seem to have reported on it), the shooting data was not nationally representative (having been taken only from Houston), unlike the data used for the rest of the study. (And even then, as Sethi notes, the Houston data seems to be a little weird.)
And it's worth pointing out (whatever one's assumptions) that this is just a single study. Real knowledge is gained bit by bit over time, with a lot of different eyeballs and brains looking at and pondering an issue (blog articles included, as it may give someone else an idea for another study which at looks at better/different data).
Update: another, more in-depth look: [50] Teenagers in Bikinis: Interpreting Police-Shooting Data