At his confirmation hearing in 2005, soon-to-be Chief Justice John G. Roberts Jr. famously likened his duty to that of a baseball umpire. “My job is to call balls and strikes and not to pitch or bat,” he said, continuing a centuries-old tradition of judges portraying themselves as perfectly impartial arbiters of the law.
That may be the ideal, but a growing body of research shows that judges suffer from all-too-human foibles. They get hangry, and sometimes their fatigue makes them treat cases unfairly. They misunderstand basic facts about probability, allowing some people to get lucky based on the order of their cases.
A new study points to an even more disturbing pattern of haphazard justice: In Louisiana, juvenile court judges appear to have issued harsher punishments following an unexpected loss by the Louisiana State University football team. And disproportionately, those longer sentences fell on black children.
LSU economists Ozkan Eren and Naci Mocan came to these conclusions after studying more than a decade of state court data. In a draft of their paper, released this week by the National Bureau of Economic Research, they show that the Louisiana judges sentenced children differently during the football season, depending on how well LSU had played the previous weekend.
The researchers used betting data from Las Vegas to classify each LSU victory or loss. They focused on losses where LSU had been predicted to win by four points or more. Between 1996 and 2012, these surprise defeats happened 14 times out of the 179 games that there was information on.
Neither upset wins, nor the outcome of close games seemed to measurably affect the outcome of court cases. But upset losses by LSU hardened how judges behaved the following week in court.
In those downtrodden weeks, judges tended to recommend sentences for children that were about a month longer, or about seven percent longer than average. The foul mood seemed to last all of Monday through Friday, but disappeared the following week. Judges were particularly severe following an upset loss when LSU was ranked in the top 10 — probably because these were high-stakes games that affected LSU’s chances at a national championship title, as Eren and Mocan argue in the paper.
When the economists looked more closely at the data, they found that the football bias was only measurable among those who had gotten their bachelor’s degrees at LSU. These alumni, who accounted for about one-third of the judges in juvenile court at the time, appeared to be responsible for the bulk of the problem — which makes sense, because these would be LSU’s most hardcore fans.
The rest of the judges, who either hadn’t gone to LSU or who had only gone to LSU for law school, didn’t seem to be influenced by the LSU football season.
The most striking result from the paper — which will require further investigation as it undergoes peer review — is that black defendants suffered the most after an LSU loss.
During regular weeks, the researchers found black and white defendants in juvenile court were more or less treated the same after controlling for the type of the crime, the child’s age, and other factors.
But during the weeks following an LSU upset, sentences seemed to lengthen only for the black defendants. The researchers could not discern any increase in the sentence lengths for whites.
“These results suggest that the brunt of judges’ emotional reaction [to LSU losses] is borne mostly by black defendants,” they write. “This disparity in sentencing following an upset loss implies unequal treatment of black defendants, triggered by an outside event, unrelated to the merits of the case.”
Of course, many people in Louisiana are LSU fans — not only the judges in these cases but also the prosecutors and the defense attorneys and the defendants themselves. Any of these people could have affected the lengths of the sentences. Perhaps prosecutors were also in a foul mood and sought harsher punishments, for instance, or perhaps defense attorneys were less likely to do a good job after an LSU loss.
But the researchers said that judges are the main culprit. Lawyers make the biggest difference on trial day, but here it didn’t seem to matter when the case was argued. It only seemed to matter when the sentencing date was — whether the judge made that sentencing decision during the week following an LSU upset or not.
In total, the researchers estimate that each LSU upset resulted in more than 1,300 extra days of probation or detention time for the unlucky defendants who were on the docket that week. “This finding underscores the importance of emotions in decision-making even in a high-stakes environment,” they write. Judges are highly-educated professionals who take their jobs seriously. But even they seem to underestimate how their own moods interfere with their duties.
In this light, it starts to seem as if it's a good idea for judges to use data and computers to manage their own biases, to provide a safety net in moments when they aren’t thinking clearly. We know that algorithms for sentencing and risk prediction are flawed, but at least they are flawed consistently and predictably.
Human judges are flawed, too, but in messier ways that we are only now starting to document. Sometimes judges make decisions for reasons they don’t fully understand — because they’re hangry or because they’re bummed over their football team’s latest loss.