A fundamental academic assumption about the federal courts of appeals is that the three-judge panels that hear cases have been randomly configured. Scores of scholarly articles have noted this “fact,” and it has been relied on heavily by empirical researchers. Even though there are practical reasons to doubt that judges would always be randomly assigned to panels—such as courts might well want to take into account the scheduling needs of their judges—this assumption has never been tested. This Article is the first to do so, and it calls the assumption into question.
To determine whether the circuit courts utilize random assignment, we have created what we believe to be the largest dataset of panel assignments of those courts constructed to date. Using this dataset, we tested whether panel assignments are, in fact, random by comparing the actual assignments to truly random panels generated by code that we have created to simulate the panel generation process. Our results provide evidence of nonrandomness in the federal courts of appeals.
To be sure, the analysis here is descriptive, not explanatory or normative. We do not ourselves mean to suggest that strict randomness is a desirable goal and indeed note that there are many good reasons for departing from it. Our aim is to test an existing scholarly assumption, and we believe our findings will have implications for the courts, court scholars, and empirical researchers.
To read the complete Article, click “VIEW PDF” below.