Some of the below are works in progress; others are essentially finished products that I like more than editors/reviewers did.
The Importance of Importance in Certiorari (with Gregory Caldeira)
We analyze all paid certiorari petitions in the 1939 Term of the U.S. Supreme Court, and demonstrate that a case’s importance and novelty—as indicated by measures we introduce here—each affect the probability that a petition is granted, net of covariates that have been previously demonstrated to be influential. Then, we present two empirical exercises that explore how the Court understands case importance. Finally, we consider the conceptual relationship between importance and salience, and related measurement challenges.
Version presented at 2022 APSA Conference is available here (pdf).
Writing Style and Persuasion on the U.S. Circuit Courts of Appeal (with Jeffrey Budziak)
Judges regularly try to shape legal policy by influencing the behavior of others. The primary way they can do so is through their legal opinions. We investigate whether an opinion’s writing style influences a judge’s subsequent decision to treat the opinion. We theorize that opinions written in a style conforming to elite norms should be more influential. To test this expectation, we consider the universe of federal appellate court treatments of the 2160 cases in the Update to the Appeals Courts Database, creating a dataset of over 9 million treatable-treating dyads. The results of our analyses suggest judges’ ability to influence others through writing style is limited. Only one element of writing style—certainty—is consistently related to treatment decisions, and this effect is limited to in-circuit treatments. In conclusion, we discuss the consequences of these findings for our understanding of the importance of opinion writing style.
Version presented at 2022 APSA Conference is available here (pdf).
Effect Sizes for Calculation of Second Differences After Logistic and Probit Regression: Alternative Quantities of Interest
Social scientists are often interested in how the effect of some covariate on an outcome varies across groups. A common approach is to calculate a second difference, the difference between two group-specific effect sizes. I propose four alternative quantities of interest for calculating second differences after estimation of logistic or probit regression. While the conventional approach is based on setting covariate values that are constant across groups, the alternatives seek to compare effects that are more directly comparable across groups, in two specific senses. The approaches allow different substantive questions to be addressed and reduce dependence on the nonlinear functional form of logit and probit. I compare the proposed alternatives with the conventional approach in an application testing whether effects of election day registration on turnout are conditional on education, showing that the method chosen to calculate second differences can substantially affect the inferences drawn.
Draft being updated. An earlier (rough) draft is here (pdf).
Agenda Control in the Hughes Court, October Term 1939 (with Gregory Caldeira)
We examine the composition of the set of cases discussed and the U.S. Supreme Court’s collective decisions on writs of certiorari through an analysis of all paid petitions voted on in OT 1939.
Update of version presented at the 2017 MPSA Conference is available here (pdf); supplementary info here (pdf).
The Minimal Effect of Opinion Language on Review in a Judicial Hierarchy (with Matt Hitt)
Does the Supreme Court seek out shoddily-written opinions for correction? Does it see well-written opinions as better vehicles? We construct an original dataset that contains information about textual features of a representative sample of published U.S. Courts of Appeals opinions from 1949-2002, as well as all published opinions for which cert was sought in OT 1982. We investigate whether textual properties are associated with discretionary review by the U.S. Supreme Court. We advance a multifaceted understanding of judicial writing quality, encompassing readability, cognitive complexity, affective language, and new to the judicial politics literature, informality. We find that all of these properties fail to exhibit a robust statistically significant association with review. These effects, furthermore, are uniformly negligible in magnitude. We conclude that, for the period we study, the Supreme Court does not use its discretionary docket to correct poorly-written legal doctrine, nor does it seek better-written opinions as vehicles.
Update of version presented most recently at the 2018 APSA Conference is available here (pdf).
Innocuous Time-saver or Counter-majoritarian Loophole? The Cert Pool and Policy on the U.S. Supreme Court
Since 1972, the certiorari petitions that arrive at the Supreme Court are randomly assigned for review to one of the five to eight justices who are members of the Supreme Court cert pool. Justices who are not members of the pool still review each petition. Thus, some petitions for review are evaluated by as few as two justices (or their clerks), and the rest of the Court relies on the recommendations of these justices. This practice has been criticized for its potentially counter-majoritarian implications. Formalizing the communication between the better-informed justices who review a petition and the rest of the Court as a sender-receiver game, I assess the circumstances under which the Court median’s preferred policy outcome is subverted. Then, I analyze the decision to opt out of the cert pool. In some equilibria, the cert pool can move policy away from the median justice; in others (specifically, those in which one justice on each side of the median opts out of the cert pool) it is predicted to have no influence. Empirical patterns of cert pool membership in the last eleven natural courts are consistent with equilibrium predictions.
Available here (pdf).
Testing Theories of Strategic Decision-Making on the U.S. Supreme Court (dissertation)
The dissertation addresses competing theories of influence on the U.S. Supreme Court. In essence, the competing theoretical models differ over the location of Supreme Court policy. The Court median model proposes that the median justice sets policy; the agenda-setter model suggests that assignment of the majority opinion is an agenda-setting power that can draw policy away from the Court median; the majority median model indicates that Court policy reflects the median member of the majority coalition. All three models have theoretically compelling elements; however, there have been few systematic assessments of their empirical relevance, most of these have been limited to the final opinion-writing stage, and the evidence on their relative explanatory power is mixed. I analyze these three formal models of Court policy-making, and derive predictions that each makes for four stages of Court decision-making. Then, I test the predictions at the three non-final stages, using archival and original data for the Court terms 1946-1985. I find conditional support for the Court median and agenda-setter models.
Available here (external site).