Does the Judge Matter? Exploiting Random Assignment on a Court of Last Resort to Assess Judge and Case Selection Effects


  • We thank Na'ama Schlam and Noam Guttman for invaluable research assistance, for their insightful comments, and for coordinating the work with the students in a superb manner. We are also grateful to Efrat Zilberbush, Na'ama Daniel, Nitzan Ilani, and Gadi Ezra for their assistance in collecting the data. We thank participants at faculty workshops at Bar Ilan University, Haifa University, and at the Fourth Annual Taubenschlag Lecture at Tel Aviv University Buchmann Faculty of Law for comments. This study was supported in part by a research grant from Cegla Center for Interdisciplinary Research of the Law at Tel Aviv University, and by support from Robert B. Diener and David S. Litman, Cornell Law School class of 1982, founders of the Cornell Law School-Tel Aviv University Exchange Initiative.

Theodore Eisenberg, Henry Allen Mark Professor of Law, Cornell Law School, and Adjunct Professor of Statistical Sciences, Cornell University, Myron Taylor Hall, Ithaca, NY 14853; email: Fisher is Senior Lecturer at the Tel Aviv University Buchmann Faculty of Law; Rosen-Zvi is Senior Lecturer at the Tel Aviv University Buchmann Faculty of Law.


We study 1,410 mandatory jurisdiction and 48 discretionary jurisdiction criminal law case outcomes in cases appealed to the Israel Supreme Court in 2006 and 2007 to assess influences on case outcomes. A methodological innovation is accounting for factors—case specialization, seniority, and workload—that modify random case assignment. To the extent one accounts for nonrandom assignment, one can infer that case outcome differences are judge effects. In mandatory jurisdiction cases, individual justices cast 3,986 votes and differed by as much as 15 percent in the probability of casting a vote favoring defendants. Female justices were about 2 to 3 percent more likely than male justices to vote for defendants but this effect is sensitive to including one justice. Defendant gender was associated with outcome, with female defendants about 17 percent more likely than male defendants to receive a favorable vote on appeal. Our data's samples of mandatory and discretionary jurisdiction cases allow us to show that studies limited to discretionary jurisdiction case outcomes can distort perceptions of judges' preferences. Justices' ordinal rank in rate of voting for defendants or the state was uncorrelated across mandatory and discretionary jurisdiction cases. For example, the justice who sat on the most criminal cases was the fourth (of 16 justices) most favorable to the state in mandatory jurisdiction cases but the 12th most favorable in discretionary jurisdiction cases. This result casts doubt on some inferences based on studies of judges on discretionary jurisdiction courts, such as the U.S. Supreme Court, in which only discretionary case outcomes are observed.