Confirmation Bias in the United States Supreme Court Judicial Database

Abstract

We ask whether the widely used direction of decision and direction of vote variables in the United States Supreme Court Judicial Database (USSCJD) are contaminated by confirmation bias, or have been affected by expectations about the likely effects of judicial preferences on case outcomes. Using a sample of generally comparable cases, we find evidence that the assignment of issue codes to these cases, codes that govern the subsequent assignment of “direction” to the Court's judgments, is conditional on both case disposition and the known preferences of the deciding court, in the direction predicted by the hypothesis of confirmation bias. We also find that the USSCJD direction variables overstate the effect of judicial preferences and understate the effect of congressional preferences on case outcomes, relative to objectively coded measures of the Court's judgments.