Document Type


Publication Date

Winter 2002


Federal courts, Appellate courts, Plaintiffs, Appeals, Harry Edwards


Civil Procedure | Litigation


Judge Harry Edwards dislikes empirical work that is not flattering to federal appellate judges. A few years ago Dean Richard Revesz published an empirical study of the United States Court of Appeals for the D.C. Circuit providing further support for the rather tame proposition that judges’ political orientation has some effect on outcome in some politically charged cases. A year later Judge Edwards published a criticism phrased in extreme terms. Dean Revesz then wrote a devastating reply by which he demonstrated that Judge Edwards “is simply wrong with respect to each of the numerous criticisms that he levels.” We believe that Judge Edwards, when he commented on our presentation at a recent conference, preserved his batting average. Giving no forewarning and employing an unjudicial tone, he lambasted a single paragraph of our thirty-five page paper, a paragraph to which we had not so much as alluded in our oral presentation. That paragraph just happened to have summarized a series of our earlier articles that had raised some doubts about the evenhandedness of federal appellate courts’ treatments of plaintiffs and defendants. Even after delivering his oral comments, he refused to let us see the written version prepared by him and his co-author. Now, upon its publication, we have finally been able to read his written remarks. Respond to that argument we must, because allowing it to stand would undermine not only our appellate research but also much of current empirical work on the law.

To focus the debate, the premise for our notorious paragraph was the direct observation in federal civil cases “that defendants appealing their losses after trial obtain reversals at a 33% rate, while losing plaintiffs succeed in only 12% of their appeals from trials.” We had obtained this result from a database that we had created from data of the Administrative Office of the U.S. Courts on all federal appeals from fiscal year 1988 through fiscal year 1997. Judge Edwards does not question these reversal-rate statistics showing an antiplaintiff effect—indeed, he expresses the surprising (and somewhat atypical) reaction that he finds these asymmetrical results completely unsurprising. What he chooses to attack instead is our subsequent speculation that appellate courts seem to be favoring the defendant, perhaps because like the rest of society they view the trial courts as being proplaintiff, or perhaps because their distance from the particular case’s facts inclines them to discount harms to the plaintiff. To destroy our speculation, he maintains that everyone who knows anything about the legal system—especially his fellow appellate judges but notably not us—knows that plaintiffs appeal on the facts (a sure way to fail), while defendants appeal on the law (a good-percentage bet, according to Judge Edwards’ oral comments, because district courts often wing it by leaving the law to appellate courts). To his mind, he thereby provides an alternative explanation of our data that condemns, by its very existence, our speculation on favoritism.

In sum, we can say that Judge Edwards is operating under a misperception that generally plaintiffs appeal the facts while defendants appeal the law. We can speculate that his court’s heavy exposure to jobs cases has inordinately shaped this mistaken view of the legal world. Regardless of cause, appellate misperceptions exist, and they may affect appellate outcomes. The motive behind statistical analysis of case outcomes is to enable seeing patterns invisible to those who study appellate opinions. Opinions may look unassailable when read one-by-one, but overall patterns of data may reveal judicial decisionmakers’ systemic attitudes, misperceptions, or biases—which, after all, can be seriously pernicious. That is, opinion-reading and data-mining can reveal different things, and both are independently worth doing.

Publication Citation

Published in: Washington University Law Quarterly, vol. 80, no. 4 (Winter 2002).