Over ten years ago, one of the first articles I ever wrote was about the limits of mass incarceration as a crime-fighting strategy, and in the course of speaking to criminologists about the issue I had one of my first great reporting epiphanies — most crime statistics are garbage, they told me, because cops can make crimes go away by reclassifying them.
Since different departments are managed differently and face different incentives, they all apply different standards at different times, and comparing the robbery rate in San Francisco in 2002 to the robbery rate in New York in 1997 just gets you gibberish. That's why you want to look at murder. If a guy shows up at the morgue full of bullets, that's a murder. Murders you can count.
I repeated this conventional wisdom offhand in a blog post in May, but David Bernstein and Noah Isackson have taken a deep dive into the Chicago Police Department for Chicago Magazine and shown that if you push hard enough you can juke the murder stats if you want to.
Specifically, they find evidence that while Chicago reported 414 murders in 2013 they really had at least 432 through what, if you read the article (and you should) looks like pretty deliberate malfeasance.
The big problem with making murders disappear (apart from it being illegal and immoral) is that, compared to other crimes, it's much more likely that you'll be caught. For example, Patrick Walker was found injured and unresponsive in a crashed automobile, taken to a hospital where he died, then during an autopsy they found a gunshot wound in his head and a bullet casing in the back seat of the car. Owing to the fact that he was clearly shot and killed, this was filed on his death certificate as a murder. But the Chicago Police Department has classified the case as a death investigation, making it disappear from their stats.
Since murders leave bodies and bodies leave paper trails, it is possible for journalists like Bernstein and Isackson to catch the cops in the act in a way that's not possible for lesser crimes.
But what the people I spoke to over a decade ago weren't really thinking about — and what I should have thought about in the intervening years — was the collapse of locally focused reporting and time-consuming investigative journalism. As locally based journalism declines, local corruption is easier to get away with (see Glentzkow, Glaeser, and Goldin and Campante and Do for some more rigorous looks at this) and there's less and less we can trust.
As best as one can tell, the truth about the Chicago murder stats doesn't actually change the broad picture — the murder is rate has fallen a lot in the past 25 years, but not as steeply as in other major American cities — but the lying about the stats has serious human consequences. Among other things, as Bernstein and Isackson write, when a murder isn't classified as a murder it doesn't get investigated like a murder. So the effort make the fight against crime look more successful can end up making it easier to get away with murder.