Tracking discretionary bias in Cook County

The State’s Attorney’s Office in Cook County, Illinois have done something really remarkable: they have made available their data set for the treatment of criminal cases in their jurisdiction dating back over five years. Chicago is in Cook County, and this makes the release of this data set a Very Big Deal: The Cook County State’s Attorney’s Office is the second-largest prosecutor’s office in the country. Kimberly Foxx, the Cook County State’s Attorney, deserves immense credit for her willingness to make this data available, as does Matthew Saniie, the Chief Data Officer.

This data set gives us the tools to more carefully assess how well our justice system is working. One of the questions I’ve been interested in for a while has been the question of whether allowing for certain kinds of discretionary judgment on the part of actors in the criminal justice system allows for more or less justice overall. To be clear: it is not possible to get rid of every exercise of discretion. I think it would be an unquestionably terrible thing to do if it were possible, but it’s not possible, and so it’s hard for me to even imagine what the proposal would be in the first place. Police officers exercise discretion whenever they initiate a stop. Prosecuting attorneys exercise discretion when they decide whether to approve of charges being filed, and which charges to file, and how to handle plea bargains. Judges exercise discretion in their management of their courtroom and in their sentencing judgments. Corrections officers exercise discretion in their treatment of offenders in their care, including decisions they make about sanctions which will affect the offender going forward. The same is true for probation and parole officers, who exercise discretion in deciding whether parolee or probationer behaviors require sanctions up to and including referral back to the court system (such as requesting an arrest warrant or setting up a formal judicial hearing). Human actions will always involve discretion of some kind.

So we’re never getting rid of every exercise of discretion. Still, there is a debate about whether exercises of discretion should be limited in order to promote fairness. Most notably, people discuss whether structured sentencing promotes fairness by limiting the discretion of judges in sentencing an offender. Mandatory minimum sentences are examples of structured sentences; more draconian versions of mandatory minimums are “three strikes” laws, which require severe penalties on people convicted of a third serious felony with no possibility for mitigation by the judge or prosecuting attorney.

Structured sentencing is odd, in part because the vast majority of actors in the criminal justice system seem to hate them. Judges dislike structured sentencing laws; prosecuting attorneys dislike them; defense attorneys dislike them; citizens going through the court system dislike them. Most judges and states’ attorneys hold the view that they need the flexibility to determine the appropriate sentence for each offender, and that justice is promoted by tailoring a sentence to the specifics of each case. So where do structured sentencing laws come from? There are two groups of people that like structured sentencing a lot: “tough on crime” politicians, and the voting electorate. The implicit (or explicit) suggestion: we have crime because judges and prosecuting attorneys, when left to their own devices, coddle offenders. (This is implicit because politicians rarely attack judges directly; instead, the suggestion is that it’s the entire criminal justice system’s fault that people are able to be let off so easily.)

Now, it’s a fact that “tough on crime” politicians are often despicable human beings who at best are willfully and woefully underinformed about the causes and drivers of human behavior, and at worst are cynical, craven, politically self-interested, openly racist and xenophobic bullies (who are also willfully and woefully underinformed about the causes and drivers of human behavior). And so I’m tempted to think that’s the end of the story: we should give the relevant actors in the criminal justice system broad discretionary power, and then we should take steps to ensure that those positions are only occupied by people who are capable of using those discretionary powers justly and wisely.

However! Unfotunately, people aren’t always wise and just, and that’s why we can’t have nice things. We can all easily imagine what might happen if we gave judges broad discretionary powers in sentencing, and prosecutors broad discretionary powers in bringing charges, and police broad discretionary powers in stopping and arresting, because that’s pretty much the world we actually live in: it would result in some people getting screwed over, namely black and brown people. (And I don’t mean to pass over the experiences of Latinx and Native people, but our criminal justice has a unique relationship to black people that it just doesn’t have even to other socially oppressed minority groups in this country.) Part of me has always thought “well, structured sentencing is clearly terrible in a lot of ways, but one good thing is that it forces a certain amount of fairness into the system.” Everyone getting screwed equally is fairer than just some people being screwed. (And yes, yes, I know, there is still lots of discretion in the system that might make any such fairness irrelevant.)

So, here’s my question: is it possible to use the Cook County data set to determine how justly discretion is being exercised? I’m going to stipulate right now that a “just” exercise of discretion is one that is not sensitive to factors that ought to be irrelevant from the standpoint of a system of justice. If a judge who was a Cubs fan sentenced every offender wearing Cubs apparel less harshly, that would be an unjust exercise of discretion; one’s choice of clothing is irrelevant from the standpoint of justice (certainly so long as the clothing is inoffensive).

In the Cook County data set, the obvious candidate for an irrelevant factor that might contribute to a discretionary judgment is race. For example, if judges have discretionary power to sentence, and they regularly sentence black offenders more hashly for the same offense than offenders who are not black, that would be an unjust use of their discretionary power; being black is not relevant from the standpoint of justice. I want to look at the three main data sets that we have from Cook County, and say something about how we might use those data sets to determine whether discretionary judgments are being exercised justly. If it turns out that the data set supports the view that discretion is exercised justly, then great! That’s at least one way that the justice system is working. If, however, it turns out that discretion is not being exercised justly, that is a sign that something needs to change. It’s not necessarily an argument that we need to limit the relevant discretionary powers; we might instead train the criminal justice agents better, or replace them with criminal justice agents who can exercise their discretionary powers better. Still, we shouldn’t dismiss the idea that discretionary powers might need to be limited, if it turns out that police or prosecuting attorneys or judges systematically exercise those powers unjustly.

Turning, then, to the first data set area from Cook County: Initiation. The initiation data is, essentially, data about arrests. I think it’s difficult to say anything about whether the discretionary powers of the police are exercised justly based on initiation data, primarily because the data set does not include the context in which the arrests were made. I can speculate about something we might say, though. The bulk of cases that the State’s Attorney’s office handles in Cook County are narcotics cases. We can estimate the rates of narcotics usage across different races using data of various kinds; we can also estimate the rates at which people are in possession of narcotics using information about e.g. how often drivers who have their vehicles searched are found with narcotics. (We have some data on this in the state of Vermont; similar data might have been collected elsewhere.) Those both require using data sources outside of the Cook County data set, but if it turns out that black people are disproportionately arrested on narcotics charges, that might be a sign that there is a failure in exercises of discretion somewhere. (I’m ignoring matters related to probable cause right now.)

(A quick word about something that I might talk about more in a different post: suppose you have community X, which has an elevated violent crime rate, and community Y, which does not. Suppose also that community X has a high percentage of black people living in it, and Y does not. Suppose also that the rate of non-violent offenses—e.g. narcotics use—is the same across X and Y. Then if the police deploy in larger numbers in X because of the higher violent crime rate, that might result in arrests in X at a higher rate than in community Y for narcotics use, even though the rate of narcotics use is the same. Even if every narcotics arrest in X and Y is justified in probable cause, the overall pattern seems unjust to me. Overpolicing certain communities seems unjust because something which should be irrelevant from the standpoint of justice—which neighborhood I live in—becomes determinant of whether I get arrested for a particular crime. I don’t know what the solution is, though; like I said, maybe I’ll talk about that in another post.)

The second data set area: Dispositions. Disposition refers to the way the case is resolved. The case might be resolved with no conviction, as through a dismissal through finding no probable cause for the initial arrest, or nolle prosequi (i.e. a decision to not prosecute), or through a verdict or finding of not guilty at trial, or through a striking of the case with leave to reinstate at later date. The case might also be resolved with a conviction, either through a plea of guilty, or a verdict or finding of guilty at trial. I again think that it’s difficult to say much about the use of discretionary judgment in how a case is disposed; we lack useful information about the factors that contribute to the exercise of discretion in the first place. Still, that’s the benefit of large data sets: we sometimes don’t need information about specifics to identify patterns in the whole. If it turns out that cases are dismissed more often for black defendants than for non-black defendants, then that might be a sign that police officers are exercising their discretionary powers unjustly. Likewise, we should look at the data for the rates at which people are found guilty, both in bench trials (where the judge determines guilt) and in jury trials (where the jury determines guilty); if race is a strongly influential factor in whether someone is found guilty or not, or if it is a strongly influential factor in whether someone is found guilty in a bench trial versus a jury trial, that might be a sign that there are unjust exercises of discretion somewhere in the system.

Finally, the third and most significant data set: Sentencing. People who have been convicted can either be detained (in jail, in prison, or in boot camp), or they can be under the authority of the court outside of detention (on probation, under supervision, or on a conditional discharge). In addition to having information about the type and nature of the sentence (including whether there was a plea of guilty of a finding of guilt), the Cook County data set also includes information about the length of sentence, as well as about the general offense type (narcotics, burglary, aggravated battery with a firearm) and the specific offense title with the legal chapter, act, and section for the charge. Finally, the Cook County data set includes information about the judge who presided over the sentencing. There are therefore multiple ways to evaluate whether discretion is being exercised justly with respect to sentencing:

• First, we want to know whether the kind of sentence an offender receives (e.g. detention vs. probation) in each category (e.g. plea of guilty for narcotics) is strongly influenced by the race of the offender. Cook county tells us that 63.24% of black defendants are sentenced to jail or prison, compared to 45.15% of white defendants; at the same time 46.72% of white defendants receive probation compared to 32.09% of black defendants. This certainly seems significant, but that data is aggregated across every offense category. We could get a clearer idea if we looked at the impact of race on each offense category.

• Second, we want to know whether the length of sentence in each category depends on irrelevant factors like the race of the offender. If, for example, black defendants averaged longer prison sentences for the same offense category as white defendants did, that would be an indicator that the exercise of discretion might be based on race.

• Third, we might want to know whether different judges exercise their discretionary powers in wildly inconsistent fashion. The idea behind sentencing guidelines is that consistency is important in the law: a defendant should not be unduly punished merely because they had the misfortune of drawing Judge P rather than Judge Q. Now, on any rehabilitationist view of justice, consistency can be overrated; what the rehabilitationist cares about is whether a sentence best promotes rehabilitation in a particular case, not whether the sentence is the same as that in another case. Even on this view, however, if the data set is large enough, the difference among cases should wash out: it’s unlikely that Judge P always drew the defendants best served by 20 years in prison while Judge Q always drew the defendants best served by 2 years in prison.

If we care about how discretion is exercised in the criminal justice system, those are some questions I think that the Cook County data set can help answer. I might even bug some of the other people involved in the WokeData project to do the analysis and see what we get. But it’s worth commending Kim Foxx once again for making this data set available; as she notes, these are issues that touch on matters of clear public importance, and we cannot determine what is happening or make wise choices in the public interest without knowing what is happening.