Black and white behind bars

I really like the Journal Sentinel's new Proof and Hearsay blog written by Derek Nunnally, so I hate my first reference to it to include the picking of a nit. But precision with numbers is important and the error that Nunnally made is repeated in the document that he refers to.

Nunnally refers to the recent report of the Sentencing Project. which finds that blacks are incarcerated at 5.6 times the rate of whites nationally and at over ten times the rate in Wisconsin. Nunnally writes "for every 100,000 people living in Wisconsin, 4,416 black people and 415 white ones are in prison." Although I suspect that he means to say the right thing, this is wrong. Indeed, I don't blame him because the text of the report sloppily refers to "rate per 100.000 population" but that's not quite what these numbers are.


What the report actually found is that there are 4416 blacks incarcerated for every 100000 black people and 415 whites incarcerated for every 100000 white people. If Nunnally were literally correct, that would mean there are over 260000 black and white people incarcerated in Wisconsin (the real number for inmates of all ethnicities is less that 10% of that) and there would be ten black inmates for every one white inmate. (It's closer to 1:1) If there were 4416 blacks incarcerated in Wisconsin for every 100000 residents, 80% of the black population would be behind bars. The black incarceration rate would be much greater than 10x the white rate. It would, in fact, be over 175 times the white rate.

Of course there is still a big discrepancy between the white and black incarceration rate, but the Sentencing Project report, although it contains policy recommendations (largely of the "turn 'em loose" variety), does not attempt to figure out why this is.

Even if you concede that there must be a reason rooted somewhere in America's racial history, that doesn't tell you much about what's going on today since we know that there can be substantial differences in, for example, the incidence of arrests among racial groups. Let's take one that is easy to find and simple because I really need to get back to work

We know that roughly the same number of whites and blacks were arrested for murder and nonnegligent manslaughter in 2005. Yet there are approximately six times as many white people as black people in the United States so blacks are apparently six times as likely as whites to be arrested for these crimes. This is unlikely to be explained by bias in the criminal justice system.

As I said, I am perfectly prepared - even eager - to believe that the reason for this is rooted, to a significant degree, in our sad racial past and that poverty plays a role. This seems indisputable.

But that doesn't really tell us what to do about it today. It doesn't mean, for example, that crime has not become a major factor in the perpetuation of poverty and that we'll never do much about poverty until we do something about crime.

The numbers in the Sentencing Project report are important but they don't explain themselves and their cause is not self-evident. They can begin a conversation but they don't carry it very far.

There is, by the way, an interesting example of the fallacy of correlation that can be illustrated by the report. Eleven of the 15 states with the highest discrepancy between black and white incarceration rates were carried by John Kerry. Fourteen of 15 with the lowest discrepancy were carried by George W. Bush. There is, it seems, a statistically significant correlation between Republican voting and greater racial parity.

Nb: It's probably God's sweet justice that I had an omission on the first draft of this post.