Representation and re-presentation in litigation science.

Federal appellate courts have devised several criteria to help judges distinguish between reliable and unreliable scientific evidence. The best known are the U.S. Supreme Court's criteria offered in 1993 in Daubert v. Merrell Dow Pharmaceuticals, Inc. This article focuses on another criterion, offered by the Ninth Circuit Court of Appeals, that instructs judges to assign lower credibility to "litigation science" than to science generated before litigation. In this article I argue that the criterion-based approach to judicial screening of scientific evidence is deeply flawed. That approach buys into the faulty premise that there are external criteria, lying outside the legal process, by which judges can distinguish between good and bad science. It erroneously assumes that judges can ascertain the appropriate criteria and objectively apply them to challenged evidence before litigation unfolds, and before methodological disputes are sorted out during that process. Judicial screening does not take into account the dynamics of litigation itself, including gaming by the parties and framing by judges, as constitutive factors in the production and representation of knowledge. What is admitted through judicial screening, in other words, is not precisely what a jury would see anyway. Courts are sites of repeated re-representations of scientific knowledge. In sum, the screening approach fails to take account of the wealth of existing scholarship on the production and validation of scientific facts. An unreflective application of that approach thus puts courts at risk of relying upon a "junk science" of the nature of scientific knowledge.