Let's not confuse outselves with facts

I am a bit surprised by Paul Soglin's response to John McAdams study on racial disparities in incarceration rates in Wisconsin. I want to think its entirely tongue-in-cheek and not an anti-intellectual refusal to understand what the study was trying to say, but it's a powerful hard thing to do.

First, let's look at the study. This is how Mayor Paul (that's how you address mayors in Madison, right?) sums up the study:

The McAdams report suggests that Milwaukee homicide numbers are within reasonable expectations. In fact more murders can be tolerated. McAdams's methodology changes the way we analysis all crime data including homicides.

But, as Hizzoner must know, the McAdams report does not suggest, imply or even cast a sidelong glance at any such thing. Professor McAdams does not say that the number of homicides in Milwaukee is what we would expect or that it is acceptable. In fact, strictly speaking (always a good thing), the study says nothing about the homicide rate. What he was looking at is disparities in what happens to blacks and whites after a crime has been committed.

In connection with that, he is trying to get under numbers that show that the disparity in black and white incarceration rate is higher here than it is elsewhere. One thing that you need to know in order to assess why that is so is to look at whether the black population here commits crimes at a rate that is higher than elsewhere.

That's not easy to do. If you use conviction and arrest records, you might just be reflecting biases in the system that occur before those stages in the process. So John wonders whether the nature of the black population in Wisconsin differs from those states where the incarceration disparity is lower. He hypothesizes that states where the black population is more likely to be urban and poor may have a larger disparity than in states where it is less likely to be urban and poor. He constructs a model, runs the numbers and - whallah!- there seems to be precisely the correlation that he hypothesized.

This isn't airtight but it is instructive. Maybe what Paul is attacking is the notion that a social scientist might try to control for potential explanatory factors as a way of testing accusations of discrimination. If that's the case, then I shouldn't think he will have many friends on Bascom Hill.

In any event, he is conflating the idea of what those explanatory factors predict with a value judgement about whether the results are acceptable. Studies may show that people with cholestorol in excess of a given number can be "expected" to have a higher incident of heart disease. That doesn't mean that the fact that they have that rate of disease is fine and dandy. But it does tell us that the problem of heart disease might have something to do with cholesterol and not eye color, just as McAdams' numbers suggest that the disparity might be related to something other than racist Badgers.

One possibility is that urban, poor blacks are more likely to commit crimes than non-poor, rural blacks. That's not an outlandish hypothesis. It even scans with some liberal notions about the root causes of crime. Whatever its value, it is not the same as saying that the disparate crime rate can be tolerated.

Another possibility is that poor urban blacks get locked up more often than whites or nonpoor and rural blacks for reasons that have nothing to do with a greater propensity to commit crimes. The numbers suggest that, if this is true, Wisconsin may be no worse than anywhere else but, once again, that's not the same as saying that, for example, nationwide bias against poor blacks in cities is OK.

Professor McAdams acknowledges this second possibility, so he goes on to look at other numbers to see if he can find any evidence of that. I plan to post on that later, but I am really surprised to see this anti-science attitude from a denizen of the Party of Reason.