Those non-random appellate panels? Study finds they’re (almost) everywhere

December 4, 2014

The random assignment of federal circuit court judges to the three-judge panels that hear appellate arguments is a bedrock assumption about the U.S. judicial system. That’s why it’s notable when panel assignments appear to be the result of something other than sheer fate. I told you yesterday, for example, about a law review article by a University of Georgia law professor who hypothesized that two judges at the 7th U.S. Circuit Court of Appeals have taken advantage of an idiosyncratic policy that permits 7th Circuit judges serving on motions panels to claim authority over the merits of appeals they’ve allowed to proceed. When Edward Whelan of the Ethics and Public Policy Center, who writes the Bench Memos blog at the National Review Online, recently raised questions about assignment procedures at both the 7th and 9th Circuits, he drew responses from judges at both circuits.

Appeals courts want their assignment processes to appear blind and balanced. But according to a working paper by law professors Adam Chilton of the University of Chicago and Marin Levy of Duke, we shouldn’t be so quick to assume they are. The data-rich and statistically sophisticated article says that evidence indicates the assumption of randomly configured appellate panels is false.

I’ve worded the paper’s conclusion carefully, following the lead of Chilton and Levy. Their paper, posted at SSRN in November and entitled, “Challenging the Randomness of Panel Assignment in the Federal Courts of Appeals,” emphasizes that the study was designed only to test the assumption that panels are randomly assigned, not to explain why assignments may not be random or even to posit that random assignment is a desirable goal. The professors repeatedly caution that their findings don’t establish any court’s intention to skew panel composition.

Despite the caveats, their analysis is startling. The professors compiled a database of the composition of all oral argument panels at the 12 geographical circuit courts between September 2008 and August 2013. (As they describe, just gathering that data from the circuit courts was Herculean.) They constructed a computer program to generate random panel assignments in each circuit, taking into account factors such as the number of panels assembled by each appellate court and the number of judges who sat on panels within each circuit. Using that program, they ran 100,000 simulations, generating 3 billion artificial panels. When they compared the artificially generated panels to reality, they concluded (with 95 percent statistical confidence) that they “would not have discovered as much evidence of non-randomness as we did if all the circuits were actually using a random process.”

Chilton and Levy looked specifically at the randomness of three factors: ideology (using the proxy of whether a judge was appointed by a Democratic or Republican president), gender and race. They found evidence that the ideological balance of appellate panels was non-random in five circuits – the 2nd, 4th, 8th, 9th and District of Columbia. The evidence in the D.C. Circuit was especially stark. In their simulations, the mean number of panels with only one Republican appointee was 102 of the 510 panels in the 2008 to 2013 time frame. In reality, there were 119 such panels.

Evidence of gender and racial non-randomness was less pervasive in the study’s findings. The 3rd, 4th and 5th Circuits showed evidence of non-random distribution of female judges. The 2nd, 3rd, 4th and 5th produced non-random distribution results for minority judges.

It’s worth pointing out that the Chilton and Levy study showed no statistically significant evidence of non-random panel assignments in the 7th and 9th Circuits, whose assignment practices have received anecdotal attention (including from me). The 1st, 6th, 10th and 11th Circuits also stayed within the professors’ randomness parameters. The 2nd, 3rd, 4th and 5th Circuits, on the other hand, were at odds with random distribution in two categories each.

Chilton and Levy point out all sorts of variables that could account for non-random distributions, including recusals and judges’ scheduling conflicts. Perfect randomness may be impossible to achieve, they say, and non-randomness might be simply “an attempt to balance a series of otherwise benign factors.”

But as the professors point out in a separate section of their paper, appellate judges and court scholars have long taken for granted that judges are selected to hear cases randomly. The study concludes that reality is more complicated.

For more of my posts, please go to WestlawNext Practitioner Insights

Follow me on Twitter

No comments so far

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see