Taking a fresh approach to a long-standing problem in the study of crime, a University of Pennsylvania Law School professor and colleagues at the University of Chicago and Harvard University have demonstrated conclusively for the first time that racial bias affects judicial sentencing decisions.
In a new paper entitled “Do Judges Vary in their Treatment of Race?” forthcoming in the Journal of Legal Studies, David S. Abrams, assistant professor of law, business and public policy at Penn Law, and his co-authors, Marianne Bertrand of the University of Chicago and Sendhil Mullainathan of Harvard, offer statistical proof that the answer is “yes.”
Whether racial discrimination affects criminal sentencing is a matter of contentious debate among scholars and the general public alike, in part, because it contradicts a bedrock constitutional commitment to equal protection of the law.
Social scientists have studied the issue for decades, but the seemingly simple question “Does race affect sentencing?” is surprisingly difficult to answer on the basis of empirical evidence.
Abrams explains: “The most straightforward way you might look at it is to say, ‘Let’s look at what sentences people get and see whether sentence length varies by race. If it looks like people of one race receive longer sentences than another, that might indicate that the criminal justice system is unfair. But the shortcoming to that approach is that it’s also possible that sentences can differ for many reasons. For example, it’s possible people of different races might have different criminal histories on average, and that could also explain the difference in sentence length.”
To address that difficulty social scientists have traditionally applied control variables to standard regression equations, a statistical method for identifying significant correlations between observed events. For instance, controlling for the type of crime committed or for the defendant’s criminal history, researchers look to see whether the results of their equation still show racial disparity.
“The problem with that is you still leave the possibility that any differences you see are due to unobserved variables, differences that might be there but that you can’t control for,” said Abrams. “That might be demeanor in the courtroom, it might be the quality of the attorney you can afford, it might be some details about the crime that you might not capture in your data. If those things are correlated with race, which they probably are, you’re not going to know whether the effect you think you’re detecting is really race or is something else.”
To avoid this problem of “unobservables,” Abrams, an empirical economist by training, and his colleagues take a new approach to the question.
In their study, they take advantage of the fact that the criminal justice system randomly assigns cases to judges.
“What that means,” said Abrams, “is that if you have a large enough sample of cases, on average across judges they’re going to get the same types of cases, meaning the same mix of race of the defendant, the same mix of crimes. Everything about the cases, assuming they’re randomly assigned, should be the same on average for each judge, including the variables we can’t observe.”
Working with a large data set of felony cases from Cook County, IL, the investigators use a statistical technique called a Monte Carlo simulation to show that judges take race into account in their sentencing decisions.
They look at what they call the “racial gap” in sentencing — the difference between sentences for black defendants and white defendants — and find that it varies across judges, showing that race is affecting sentencing decisions.
“We find evidence of significant inter-judge disparity in the racial gap in incarceration rates, providing support for the model where at least some judges treat defendants differently based on their race,” Abrams and his co-authors note in the study. “The magnitude of this effect is substantial. The gap in incarceration rates between White and African-American defendants increases by 18 percentage points (compared to a mean incarceration rate of 51% for African-Americans and 38% for Whites) when moving from the 10th to 90th percentile judge in the racial gap distribution.”
While the technique eliminates the problem of “unobserved variables” and demonstrates the salience of race, it has limits of its own. The results don’t show in which direction racial preferences tilt, whether in favor of blacks or whites.
“One reason I don’t think you necessarily have to figure that out is because to the extent you think people’s faith in the criminal justice system is dependent on their belief in its fairness, the results are disturbing. Regardless of which way it’s unfair, it’s a bad thing to have race enter into sentencing. So in that way, I think the desire to improve the system is independent of which way it goes,” said Abrams.
What can be done to make sentencing fairer? Sentencing guidelines have been one approach.
“Guidelines are a way to try to constrain variation and this racial gap in sentencing. But, truthfully, guidelines have a downside,” said Abrams. “They are a way to try to constrain variation in sentencing, but they also limit a judge’s ability to be fairer in particular cases. So I’m not sure that’s the only or the best solution.”
The research itself points to another approach.
“We can educate judges and make them aware of this result,” said Abrams. “No judge is likely to acknowledge, on his or her own, ‘Well, of course, I take race into account.’ And likely they don’t in any explicit way. But they probably do implicitly, because we take all kinds of things into account implicitly. And I think making judges aware of it could potentially help going forward.”
In fact, there is evidence for that belief from the world of medicine. Social science research there has shown that physicians’ immediate, unconscious reactions to racial minorities lead them to undertreat black patients. As reported recently in The New York Times, about a quarter of the doctors in one such study suspected that it was designed to test racial bias. The researchers found that this “aware” group of physicians did not treat patients differently. Once they realized that race was an issue, it was no longer a factor in their treatment decisions.
The hopeful and potentially useful message of Abrams’ research is that judges could react the same way: If they can observe the racial bias of their sentencing decisions, they will work to correct it.