The empirical study of judicial behavior continues to grow and mature. The live challenges include specification, such as constructing useful conceptions and measures of ideology, mapping particular domains in which identifiable forces influence decisions, and quantifying the magnitudes of those influences. To make progress on these challenges, we roll out new and expanded datasets that build on the work of Cass Sunstein, Lee Epstein, Gregory Sisk, and others, and we report on the character of constitutional litigation today. Our datasets cover U.S. Court of Appeals decisions in five domains: (1) commercial speech, (2) gun rights, (3) abortion rights, (4) establishment clause claims, and (5) anti-affirmative action claims. The bulk of the data reaches into 2016. Part of the data collection was automated, but all judge votes were coded by at least one law professor. Our vote coding allows judges to support claims in part or in full. We then deploy three proxies for judge ideology, including a new variable designed by Adam Bonica and Maya Sen that relies on judges’ pre-appointment campaign contributions. In our regression models, we introduce both standard and novel independent variables, such as three measures of procedural and substantive law.
Commercial speech cases are the focus of this Article. We find no evidence of ideological influence within the full set of those cases, in the sense of judge votes tracking ordinary policy disagreements. The results make commercial speech cases look like gun rights cases—and unlike abortion rights, establishment clause, and affirmative action cases, which are consistently ideologically charged in our models. The differing magnitudes of ideological influence across case sets are presented numerically and visually. However, when commercial speech cases are limited to post-2000 decisions, to cases involving disclosure requirements, or to cases involving “right-wing advertising,” some results do change. Our variable for “big business” claimants is statistically significant in the post-2000 cases but not in the full sample of cases. Also, subtle ideological rifts seem to emerge in the disclosure and rightwing advertising cases, with some judges apparently migrating toward or away from supporting commercial speech claims in part or in full. Some of our findings are preliminary and warrant further research. Regardless, our data and analysis should cast more light on contemporary constitutional litigation as it now stands, after the close of the Obama administration and at the beginning of the Trump presidency.