Appellate court assignments as a natural experiment: Gender panel effects in sex discrimination cases

This paper argues that estimating causal effects on US Appellate Court panels can be advanced by analyzing the data as a series of natural experiments, fully exploiting the as-if random assignment of judges to cases. As a template, this paper reanalyzes Boyd et al.'s data on sex-discrimination cases. The question is the impact on the votes by male judges from having a female judge on their panel. Leverage from as-if random assignment can be exploited only by restricting comparisons of treatments cases (in the example, female co-panelist) exclusively to control cases (all-male panels) from the same period and time period from which the treatment cases are drawn. With as-if random assignment reducing the possibility of a biased estimate, the results confirms a gender panel effect similar in size to the claim by Boyd et al. Restricting comparisons to within the same circuit and time period further advances understanding of the causal mechanism. When male or female judges side with female plaintiffs, the females are more persuasive at swaying the votes of their male co-panelists' votes.