The "reference class problem" is a serious challenge to the use of statistical evidence that arguably arises every day in wide variety of cases, including toxic torts, property valuation, and even drug smuggling. At its core, it observes that statistical inferences depend critically on how people, events, or things are classified. As there is (purportedly) no principle for privileging certain categories over others, statistics become manipulable, undermining the very objectivity and certainty that make statistical evidence valuable and attractive to legal actors. In this paper, I propose a practical solution to the reference class problem by drawing on model selection theory in statistics. The solution has potentially wide-ranging and significant implications for statistics in the law. Not only does it remove another barrier to the use of statistics in legal decisionmaking, but it also suggests a concrete framework by which litigants can present, evaluate, and contest statistical evidence.
coming soon: the law of evidence on Spindle Law