Assessing and Predicting Who Wins Federal Tax Trial Decisions

Abstract

In "Assessing and Predicting Who Wins Federal Tax Trial Decisions," Professor Daniel Schneider uses empirical research to examine the interplay between who wins in federal tax decisions and the social backgrounds of the judges who decided the cases. While tax scholarship engages in traditional legal reasoning about tax cases, it has not empirically analyzed who wins. This article uses statistics, especially logistic regressions, to engage in such an analysis, taking judges' backgrounds into account.

Professor Schneider's database consists of decisions from both the Tax Court and the district courts sitting in the Southern District of New York, the Northern District of Illinois, and the Central District of California that were rendered between 1979 and 1998. He has assembled data about both the decisions - the court where the decision was made, the year of the decision, the primary Code section at issue, and who won the case - and the social backgrounds of the judges who rendered the decisions - each judge's gender, race (for the district court judges) educational background, political affiliation, prior professional experience, and length of the judge's tenure when he rendered the decision.

Some of Professor Schneider's descriptive statistics include his observations that the taxpayer won more in the Tax Court than in the district courts, before judges who had gone to more elite colleges, before judges who had sat on the bench less time when rendering the decision, and before judges who had come from a prior professional experience primarily in private practice.

All of the tested variables in judges' social backgrounds were correlated with who won, with strong associations between background and the taxpayer winning occurring when a judge was a woman, had a more elite education, had a prior primary work experience in private practice, or was appointed by a Democratic president.

Keywords: federal taxation, who won, statistics, empirical research, logistic regression

JEL Classification: H29, K19, K41