Thursday, February 5, 2015

Truly Too Hot For Jacobin!

The other day on Facebook, my Jacobin colleague Connor Kilpatrick posted an excerpt from an interview with prison scholar Marie Gottschalk. Gottschalk has a new book out called Caught: The Prison State and the Lockdown of American Politics that looks to be very much worth reading.

In the interview, she points out how, despite large racial disparities in incarceration rates, even whites in America are imprisoned at far higher rates than the populations of most other rich countries.

Gottschalk is right. Which reminded me of something that came up during the Charlie Hebdo debate. Some people at the time were circulating articles reporting that 60% or 70% of French prisoners are Muslims. In fact, these were basically guesses since France doesn't collect data on the religious affiliations of detainees. (Which doesn't necessarily mean they're wrong).

The argument went that by comparing those stats with analogous U.S. figures it could be said that the "racial" bias against Muslims in the French criminal justice system is even worse than the bias against blacks in the U.S.

So I went looking for actual stats on French prisoners, and compared them with U.S. data. Here's what it looks like:

It turns out it's true -- the black-white incarceration ratio in the U.S. is lower than the North African immigrant-native ratio in France. But that's only because so few native French are in prison. 

Which raises a larger issue.

I remembered finding something surprising in the U.S. stats when I looked at them a while ago. It turns out that the smallest racial disparities in U.S. imprisonment rates are in the Deep South, while the largest are in states like New Jersey and Connecticut. Not quite what you'd expect, right?

What to make of that surprising fact? I have no problem believing that the New Jersey and Connecticut justice systems are racist. What I find hard to believe is that those in Alabama and Mississippi are far less racist.

So after looking at the French numbers, I decided to do a little statistical analysis. I found that the degree of racial disparity in U.S. states' incarceration rates is almost entirely a function of how low the white rate is. It's completely unrelated to how high the black rate is. (R-squared is 54% for the white rate, 5% for the black rate.)

Racial disparity in overall incarceration, it seems, is a pretty useless way to measure the bias of a criminal justice system. What seems to be the case, rather, is that the more punitive a justice system gets, the more the experience of incarceration starts to affect people outside the very lowest ranks of society.

The result is a paradox: the higher a state's overall incarceration rate, the smaller the racial disparity. Here's what that looks like:

So suppose, tomorrow, the government were to blindly adopt an across-the-board cut in statutory sentencing standards for every crime. The result would surely be a massive drop in incarceration, for both blacks and whites. But also, it seems, a big increase in the level of racial disparity. Food for thought.


  1. Very interesting, thank you! Would you post a link to the data set? It looks like there is actually a strong non-linear effect (ie the comma-shapes distribution of points) and Id love to tinker w the model. It looks like the relationship is actually much stronger than a linear model R.sq is giving you credit for.

    1. Here they are:

  2. Thank you! Sorry it took me a minute there to realize you'd responded. I'm deep in dissertation land.