Marquette Warrior: Black Incarceration in Wisconsin: More

Saturday, September 29, 2007

Black Incarceration in Wisconsin: More

Our article on the supposed “disproportionate incarceration of blacks” in Wisconsin seems to be getting some attention.

Some leftist moonbat bloggers have taken a swing at it, and mostly proved they have no idea about the issues involved.

But some of the response has been more serious.

Wisconsin Public Radio is preparing a segment on the issue, and asked Prof. Pamela Oliver, at the UW-Madison, to reply to our essay. She did so, and gave us permission to publish what she wrote in a private e-mail to a WPR producer. So we are publishing her entire critique of our article here, along with a “preface” she sent us in a later e-mail. We are publishing the “preface” below the main e-mail, since we think it makes more sense that way, but in fairness to Oliver, do read it.
. . . this is a long complicated essay that makes many different points and responding to it point by point would take a week. Here are some responses.

(1) Racial disparity refers to the statistical pattern that Blacks (and to a lesser extent Latinos and American Indians) are arrested and incarcerated at higher rates than Whites. This is indisputably true.

The Commission is charged with trying to reduce these disparities.
But does one want to reduce these disparities? If they result from disparities in committing crimes, we don’t think so.

Note that some of the statements below suggest that the Commission is thinking of ways to reduce crime in the black community. If black crime rates decline (relative to white crime rates) we certainly would want disparity in incarceration to decline.
(2) At this juncture, 40% of the Black male population nationally is under the supervision of the correctional system and about a third of young Black men are expected to spend some time in prison. McAdams seems to think this is an acceptable state of affairs if he can show that it is “as predicted” by control variables. Other people think this is a terrible problem with enormous consequences for whole Black communities.
But is the problem that the system discriminates against blacks, or that blacks commit too many crimes? It makes a huge difference. One is on a fool’s errand trying to fix the criminal justice system if the problem is not with the criminal justice system.
(3) Any reasonable person who has dug into the data recognizes that there are two components of the statistical disparities. Some of the difference is coming from differences in the rate at which Whites and Blacks commit crimes, and some of the difference is coming from differences in the way people are treated in the system given that they have done something wrong. Roughly half of the presentations the Commission has heard and roughly half of the Commissioners have an orientation that is primarily focused on reducing crimes committed by Black people. That is, they have raised concerns about funding for education, about programs to strengthen families and reintegrate fathers into communities, about jobs programs and job discrimination, about bringing people to God, etc. In short, a lot of the people on the Commission think this is a terrible crisis that should be addressed by dealing with underlying problems of crime. Where they differ from McAdams is that they think there is more to reducing crime than incarcerating an ever-growing share of the Black population.
Some of those proposed ideas for reducing black crime sound good to us (strengthen families and reintegrate fathers into communities, bringing people to God), and some sound like more of the same things that have failed (more spending on education, jobs programs). But even the things we like aren’t easily manipulated by public policy.

Are the commissioners saying that locking up black criminals is not part of the solution? We hope not.

Anybody with even a casual interest in crime in the Milwaukee area is aware of the reality shown on a map provided annually in the Milwaukee Fire and Police Commission Public Safety Report.

So far as we are concerned, anybody is welcome to go find ways of reducing crime with this or that warm fuzzy-sounding liberal program. But until they can fix the crime problem that way (and we aren’t holding our breath), cops need to catch criminals, and prosecutors and courts need to punish them.

That is one social program that actually works, in contrast to all the programs that the liberals are always touting.

We’ve probably all heard the story about the man sitting in a bar over his drink, lamenting the fact that he drinks too much. A fellow patron asks “why do you drink so much?” The man replies “to forget my problems.” The fellow patron asks “what problems.” The man replies “I drink too much.”

Thus we find liberals blaming crime on the problems of the inner city. Fair enough. But what is the chief problem of the inner city? Crime.
(4) What McAdams leaves out of his statement -- because that would contradict his arguments -- is that the time trends for crime and incarceration don’t support his arguments that this is just tracking crime. Crime generally went down while incarceration skyrocketed. So it just is not true that incarceration is simply tracking crime.
Liberals often see it as a problem that, in the last couple of decades, as the incarceration rate has gone up, crime has gone down. Actually, the way they always say it is that “as the crime rate has gone down, the incarceration rate has gone up.”

Somehow it never occurs to them that crime went down because of increased incarceration!

Oliver’s statement, by the way, doesn’t really address the point that we made, which is that higher incarceration rates of blacks reflect higher rates of offending by blacks.
A more complicated problem to dig into is the possibility of an incapacitation effect, and I’m now persuaded that there is some incapacitation effect in the data, for property crime but not for violent crime. Most criminologists now think that drug incarcerations have absolutely zero impact on reducing illegal drug use or sales, and some argue that they actually increase it through indirect effects. But really digging into these issues is to complex for a short response.
If somebody wants to dig into the issue, they might check the works we cite on incapacitation in our article. There seems to be good evidence for an effect both for violent crimes and property crimes.
(5) The Sentencing Commission report clearly found a substantial racial difference in the probability of being sentenced to prison rather than probation for drug offenses and for lower-level violent and property offenses. This analysis controlled for prior record and controlled for multiple offenses at the trial, as well as for other factors available in the record.
That is indeed pretty much what the Sentencing Commission found, but they admitted to real problems with their data. Quoting their report:
. . . many legitimate bases for distinguishing among defendants, such as prior criminal record, fewer educational opportunities, and erratic work history, are likely correlated with being Black, which also systematically disadvantages Black offenders (p. 4).
And this:
Ideally, a report of this scope would rely upon data tracking offenders by race from the initial police call through investigation, arrest, the prosecutor’s charging decision, plea negotiations, trial, conviction, and finally, sentencing. Data would include, not only the race of the offender, but the race of other actors, including the responding officers, victim, prosecutor, judge, and jurors, as well as whether the offender was assigned a public defender or retained private counsel. Beyond race, numerous socioeconomic factors may also contribute to differential involvement in crime and differential treatment by the criminal justice system. All of these data points, collected in a consistent and integrated system, would contribute to a fuller understanding of the role of race in sentencing decisions and throughout the criminal justice system.

Currently, Wisconsin does not collect this data in a coordinated way conducive to analysis by race.

[. . .]

For instance, the Commission collects data regarding the factors that influence a judge’s sentencing decision for sentencing guideline offenses when judges submit sentencing worksheets. These worksheets, when submitted and fully and properly completed, contain information about factors, such as employment history and family support, that might influence a judge’s sentencing decision. These factors are likely correlated to race and thus, might help explain the role race may or may not play in sentencing decisions. However, the Commission recently found that judges only submitted worksheets for 23% of the guideline offenses for which worksheets should have been submitted. Furthermore, worksheets were not submitted proportionately from all parts of the state, resulting in overrepresentation of Milwaukee County cases in the available data. Finally, even when judges do submit worksheets, the worksheets often are not filled out completely or correctly. As a result, the data collected by the Commission is less reliable, and our ability to use the available sentencing data to explain the role race may or may not play in sentencing decisions decreases significantly (pp. 5-6).
In fact, we think that, by the usual social science standards, the Sentencing Commission data is pretty good. But it frankly omits some factors that might legitimately affect sentencing, and it only deals with one point in the whole process. It can’t touch the issue of whether black offenders are more or less likely to be arrested, more or less likely to be charged, more or less likely to have their cases dropped by the district attorney, and so on.
It found no difference in sentence lengths. The findings of the Sentencing Commission for Wisconsin are exactly parallel to what has been found in other states. No difference in sentence length, a big difference in the in/out decision for lesser offenses and drug sentences but not for the most serious offenses. McAdams stresses the sentence length part of the analysis but ignores the in/out part of the analysis. While many of the actors on the sentencing commission wanted and got language that stressed that maybe something else would explain away the in/out result, anyone familiar with the logic of statistical analysis would recognize that the effects are large enough and that enough other factors have been controlled that they are probably “real.”
We are not at all sure of this. We have seen quite large estimated effects in a statistical model entirely go away when the proper controls were put in place.

Oliver goes on:
This seems especially plausible as the pattern -- no difference in sentence length, big difference in the in/out decision for lessor offenses -- is exactly what is found over and over in by other studies in other jurisdictions. (It is interesting that McAdams is sure the sentence length part of the analysis is definitive, while he dismisses the in/out part of the analysis as inconclusive.)
The differences shown by the Sentencing Commission may very well be real, although we have to note that they only claim racial disparity in sentencing for lessor crimes -- which get shorter sentences and contribute less to overall racial disparity ratios.

But again, see above for the limitations of their study.
(6) I did a detailed analysis comparing aggregate counts of arrests and prison admissions broken down by race and UCR crime categories for Wisconsin counties the late 1990s -- I have not yet been able to get the data to replicate for more recent years. Basically what I found parallels again what is found in other studies. First, the biggest disparities are at the point of arrest and these disparities are very high for serious crimes -- this is consistent with seeing differential crime as a big component of the problem (although there are issues with arrests, see below).
OK, we think she is saying that there are disparities in arrests, which probably parallel disparities in committing crimes. Which they should.
Second, for the most serious crimes, there is very little or no racial difference in the ratio of prison to arrest. Third, for the less serious crimes, there is a significant racial difference in the ratio of prison sentences to arrests, with this ratio being in the ballpark of 1.5 to 2 times as high for Blacks as for Whites. (And in these data, Hispanics are counted as White, because of the way arrests are recorded.)
This is all interesting, but to show discrimination it is necessary that the black arrestees are in fact comparable to the white arrestees.

Doing that, unfortunately, would require data that nobody has.

It does seem, however, that Oliver has pretty much conceded that there is no racial discrimination where the most serious crimes (those most likely to draw prison sentences and indeed long prison sentences) are involved.
Fourth, for low level drug possession offenses, the ratio of prison sentences to arrests was 9 times as high for Blacks as for Whites. In short, there is substantial prima facie evidence of something happening post arrest; it cannot all be just a matter of differential crime.
We are not sure precisely what data she is talking about, but would point out that the Sentencing Commission, in the report she cites, doesn’t even bother to analyze drug possession, looking only at drug trafficking.

We conceded in the article that drug enforcement is particularly tough on the black community. We would just make two additional points:

1.) There is a rational basis for tough drug enforcement in the inner city, since that is where drug dealing and the visible effects of drug use are most severe.

2.) Tough drug enforcement in the black community is not something that racist whites imposed. It has had the strong support of both black leaders (the Congressional Black Caucus) and rank and file black citizens.

Neither point proves that current policy is good. But let’s be clear on why we have it.
(7) Arrests as a proxy for crime is a tough nut. Most criminologists believe that arrests track crime pretty well for the most serious offenses. That is, yes, there is a serious problem with serious crime among African Americans, and no reasonable person thinks there isn’t.
But the problem is, there are a lot of unserious people around!

You might talk to some faculty members at Marquette. No, not the criminologists, who are sensible on the issue. But some other professors (especially in the humanities) who haven’t realized this.
I have certainly never talked to an African American person who did not think there was a problem with crime in poor Black communities, and a lot of the talk on the Commission, especially by the Black members, has been about problems with crime and violence. As I said, about half of the Commission’s time has been spent with people talking about programs to reduce disparities in criminal justice by reducing the rates at which people commit crimes.
This is good.
It’s at the lower end of the crime scale and the drug crimes where there is evidence of a difference in treatment in the system. Some people are much more likely to get caught and prosecuted for what they do than others.
However, our analysis of data from the Incident Based Reporting System showed that, in the city of Milwaukee, blacks commit violent crimes at a rate eight times that of whites.

Now, since the data on property crimes are pretty much useless where race of offender is concerned, it might be that the disparity in offending is much less where that is concerned. But we have trouble seeing why this would be so. Are vandals more likely to be caught in a black neighborhood?

The virtue of the incident data is that it doesn’t matter if the offender is arrested or not. So differential rates of arrest (or prosecution or conviction) don’t matter.
(8) Public health data tell us that young White people use all illegal drugs -- including cocaine and crack cocaine -- at higher rates than Blacks do, but White kids have a very low chance of getting caught and prosecuted for their drug use, while policing patterns pick up huge numbers of young Black kids on drug charges. Studies of “driving while Black” drug interdictions find that Whites who are stopped are just as likely (in some studies more likely) to be found carrying illegal drugs than Blacks who are stopped, but because Blacks are so much more likely to be stopped, they account for the overwhelming majority of drug arrests. McAdams tries to get around this by minimizing White kids’ drug use (they are just puffing marijuana while the Black kids are doing the really bad stuff) but this just isn’t what the data show. When I talk to White kids, they don’t dispute that White kids do a lot of illegal drugs, they just think Blacks kids do drugs as much as they do. It is true that the vast majority of drug prison sentences are for “possession with intent to deliver” but most of those are cases involving less than 5 grams of crack cocaine, many involving less than 1 gram that were still prosecuted as delivery charges. A very odd thing about cocaine arrests and incarcerations is that it is a market that -- according to arrest data -- has many more dealers than users.
Which might suggest that a lot of cocaine users escape detection and don’t get arrested.

Also note that “dealer weight” is a legally defined concept. It means you are found with a greater quantity of the drug than an individual might have in a personal “stash,” and are presumed to be selling it. Unless one wants to posit that cops are planting drugs on a lot of black people, the argument seems very odd.
This makes no economic sense. Prosecutions of people as presumptive dealers in the possession of very small amounts of the drug are highly indicative of patterns of over-charging, and the absence of a pool of customers in the arrests are highly indicative of policing patterns focusing on some people and not others. Marijuana arrests, for example, do not have this pattern -- there you see what you would expect in a market, a lot more customers (users) than dealers.
See our discussion of drug crimes (above).

We do note that Oliver is here relying heavily on anecdotal data. Note that she has also ignored our key arguments about drug crimes. It’s not that white kids don’t use drugs. It’s that their use is seen as a personal vice and not a threat to the community.

We lean toward the view that drugs should be legalized. But few in the black community (nor among whites) agree with us.
(9) In a small exploratory study of juvenile court cases in Madison, I found that Whites first appearance in court records was for a felony, while Blacks first appearance in court records was for a low-level misdemeanor. Are we to draw the conclusion that White kids who commit felonies have never done anything bad before? Or that White kids who commit low-level misdemeanors are less likely to end up in the court records? Most of us in the system think it is the latter -- White kids either don’t get caught, or get handled informally, or are sent to suburban municipal courts, not to county court. The net effect is that a White kid presenting a felony has a shorter record than a Black kid presenting the same felony, so everyone thinks it is fair for the Black kid to get a worse sentence because he has a longer record. But that longer record is, itself, a product of differential enforcement or responses to low level crime. The Commission has been struggling with ways to get a handle on the policing of White kids in the suburban jurisdictions. Again, not denying differences in serious crime, but the policing of low level crime is part of what is feeding into the disparities, as it produces differences in criminal records.
This may well be true, although we would hope that black juveniles would “get the point” and learn something that would carry into adult life. It’s also the case that juvenile records don’t get carried into adult sentencing decisions.
(10) Another thing that people express concern about that is harder to get a handle on is charging by police and DAs. We certainly hear stories that sound like massive over-reaction and over-charging for a given offense. Complaints like two kids get in a fight, one White one Black, the White kid gets sent home with a warning, the Black kid is arrested on a assault charge. Or a Black woman being picked up and held in a suburban jail without bail on a 10-year-old shoplifting warrant from when she was 17. It’s very hard to get systematic data on this kind of thing, but there are lots of stories people tell.
Unfortunately, this is admittedly anecdotal data.

If social psychology has taught us one thing, it is that people are more attentive to accounts (stories, tales) that fit into their existing cognitive framework.
(11) Aggregate studies of states. I have done a lot of these and we could do dueling studies for weeks. A couple of comments. I have done the % urban studies. It turns out that %Black is the strong predictor -- states with higher % Black have lower incarceration rates. %Black and % of Blacks who are urban are highly negatively correlated and cannot be taken apart, but I have compared them, and %Black is the stronger more consistent predictor, not % of Blacks who are urban. I.e. McAdams’ dismissive “well they are just rural” analysis does not hold up to deep scrutiny. I have also compared Black and White imprisonment rates to Black and White poverty rates, and to crime rates. Basically, White imprisonment rates are correlated with White poverty rates, but Black imprisonment rates are not. Racial disparities are driven mostly by having a low White imprisonment rate, which in turn is highly correlated with the White population not being very poor. So the patterns are really complicated if you do the analysis in depth. In addition, incarceration rose steeply while poverty and unemployment were declining in the 1980s, and in the 1980s, Black incarceration rates were highest where Black poverty was lowest. In the 1980s, incarceration for everyone (Black and White) was higher in states where crime was higher, but in the 1990s, Black incarceration rates are uncorrelated with crime rates, while White incarceration rates are positively related to crime rates. The strongest predictor of White incarceration is the White homicide victimization rate, but the Black homicide victimization rate is uncorrelated with other measures of crime and with the White homicide rate, and is uncorrelated with Black incarceration. Etc. etc.
Frankly, we haven’t following all the twists and turns of Oliver’s attempts to model racial diaparity rates, since they seem arcane and convoluted. We started with the plausible notion that blacks in the central cities of Metropolitan Statistical areas are more likely to commit crimes than blacks in suburban, small town or rural settings. This idea best explains the low diaparity ratio of Southern states. They still have substantial rural black populations. Thus in states where a large proportion of blacks live in such central cities, the black crime rate (and thus the disparity rate) will be higher. Contrary to what Oliver asserts, the relationship was quite strong.

We then added the poverty rate among blacks (which we believed would tend to drive up black crime), and the poverty rate among whites (which we believed would increase crime among whites and drive down the disparity ratio). Voilà! The model was a very good predictor of disparity ratios.

This is the way graduate students are taught to model things. At least in political science.
In short, if you treat the data honestly, you see lots of different patterns that cut lots of different ways. The Commission is staffed by people who agree that mass incarceration of Black people is a very serious problem but who understand that there are many different sources of the problem that will require many different solutions.
Oliver then added the following in an e-mail to us:
I wrote the response quickly with only a quick skim of what you wrote because I did not have time for a long response.
This was fair enough. When dealing with the media, one simply has to respond quickly.
There are points where, on a second read, I can see that you acknowledge some of the complexity that I don’t give you credit for. Each of the specific points is a complex issue that has a lot of nuance to really get it right in the statistics. We both agree that differential crime rates are part of the picture. Where I think we disagree is that I think there is substantial evidence that differential treatment within the criminal justice system also plays a role. And we seem to disagree about whether extraordinarily high rates of incarceration are a good thing or a bad thing as a way to deal with the problems of poverty, racial discrimination in job markets, weak schools, and the other factors that play into crime rates.
We certainly do seem to disagree about this. We think the evidence is highly robust: incarceration deters crime, and incarceration incapacitates criminals, keeping them in prison when they would be out doing nasty things otherwise.

And since black criminals overwhelmingly prey on black people, to fail to incarcerate black criminals is to fail to protect black people.

We frankly think that a “law and order” policy is a precondition for doing anything else that might help the black community.
I assume we also disagree about whether the drug war is being fairly prosecuted and about whether there ought to be a supply side drug war.
We in fact, probably don’t disagree. See above.

Summarizing

So what can we conclude.

First, Oliver has essentially conceded that, for the most serious crimes (generally, Uniform Crime Reports “violent” crimes) there is virtually no evidence of racial disparity. Given that these are the kinds of crimes that are most likely to result in incarceration, and likely to produce long prison sentences, that’s quite a lot to concede.

Second, we’ll happily concede that the enforcement of drug laws is particularly tough on the black community. (Indeed, we said this in our article.) This isn’t any sort of evidence of white racism toward blacks. But neither is it a benign situation.

Finally, an open question conserns “mid-level” offenses: felonies which are not the most serious felonies. The Sentencing Commission has produced some good, but hardly decisive evidence of racial discrimination. However even the liberal Journal-Sentinel, in discussing the Sentencing Commission study, said that:
Poor data collection and sentencing factors make it unclear, however, what role race and ethnicity play, the commission concludes. . . .

. . . [A]nswering the question on the role of race and ethnicity - without leaving any room for doubt - is what it will take for the right people to take ownership of the right changes necessary. In other words, using true, straight-up, apple-to-apple comparisons. . . .
Our own data show that among persons imprisoned for Uniform Crime Report “property” offenses blacks serve shorter sentences.

What is going on with all this is a puzzle, but we have a considerable problem seeing how a criminal justice system that is pretty clearly not discriminating against blacks where the most serious offenses are concerned would be doing so with lesser offenses.

Perhaps some good will come out of the Racial Disparity Commission. Oliver’s statements are, in fact, hopeful.

But let’s be clear on what the danger is. A lot of loose rhetoric about racial bias could lead to a sort of affirmative action program in which, to make the numbers “look better,” black offenders who ought to be charged, convicted and sent to prison aren’t. No change in the law would be necessary for this perverse result. Pressure on cops, prosecutors and judges could cause it to happen.

And we know who will be the victims.

Labels: , , , ,

1 Comments:

Blogger John McAdams said...

Pamela Oliver sends the following comment:

We don't need to get into a dual on your blog, but one clarification.

The term "racial disparity" is a statistical concept as I and most others are using it. It is NOT a synonym for "racial discrimination."

Racial discrimination may cause a racial disparity, but there may be other causes. This matters because you are distorting what people say by mis-interpreting their language. It is also irrelevant to the question of disparity (and even to the issue of differential impact of enforcement) whether Black people support tough on crime policies.

Apart from that, we disagree a lot about the importance of the treatment of people for "lesser offenses." I think it is very important because it has the potential to artificially inflate prior records at the time of adjudication, and because the majority of sentences to prison are relatively short sentences for lesser offenses. (The majority of people in prison are there for serious crimes because they spend more time there, but the majority of people impacted by having a prison record go through prison on short sentences or revocations.)

8:16 PM  

Post a Comment

<< Home