In Causal Inference in Civil Rights Litigation, Professor James Greiner argued that regression analysis is ill-suited and overused as a tool to prove causation in civil rights cases, suggesting that potential outcomes, another statistical technique, should be used instead. In this response, Dean Willborn and Professor Paetzold, although recognizing the limitations of regression analysis, argue that potential outcomes does not necessarily provide a superior way of examining data in civil rights cases. Indeed, they demonstrate how most of the problems Professor Greiner attributes to regression also apply to the potential outcomes approach. This is almost inevitable since potential outcomes would use some form of regression as part of the analysis in most civil rights cases. Furthermore, they disagree with Professor Greiner’s claim that regression lacks an adequate framework for making causal inferences. Although regression cannot prove causation, no statistical method can do that, including the potential outcomes approach. Instead, regression calls on the same basic causal framework used throughout discrimination law and, indeed, the same basic causal framework used by potential outcomes. Dean Willborn and Professor Paetzold argue that the best approach is to allow experts to use the full toolbox of statistical techniques so that a case can be viewed from multiple angles. Statistics will be more valuable in civil rights cases when properly treated as a plural word.
Executive Order Prohibits Federal Government and Contractor Employment Discrimination on the Basis of Sexual Orientation or Gender Identity.