Empirical Methodology and Legal Scholarship

Abstract

In The Rules of Inference' (Rules), Lee Epstein and Gary King argue that all empirical and causal assertions in all legal scholarshipincluding doctrinal, normative, and interpretive scholarship that asserts subsidiary empirical and causal claimsshould follow a unitary approach to inference, one that is grounded in the logic of statistics. From that premise, they advance a sweeping indictment of legal scholarship and its supporting institutions, claiming that every article in their large data pool violates at least one rule of inference2 and urging a radical reorientation of legal education.3 At some level of generality, it is hard to disagree with the spirit of Epstein and King's complaints. Scholarship generally, not only in law or political science, should ground its empirical assertions in warranted inferences from sound evidence, should admit to causal and empirical uncertainty where it exists, should avoid tendentiousness and selection bias, and should follow the best statistical practices when making statistical claims. Lawyers and judges, and indeed educated people generally, would benefit from a basic knowledge of statistical methods. Putting aside the opportunity costs of statistical training (an issue to which we return below), greater attention to the rules of inference would, when appropriate, improve legal scholarship. We nonetheless reject much of Epstein and King's indictment of legal scholarship. We begin in Part I by identifying a broad domain of legal scholarship for which the rules of inference lack obvious relevance. As is unsurprising for outsiders, Epstein and King overlook that legal scholarship frequently pursues doctrinal, interpretive, and normative purposes rather than empirical ones. Legal scholars often are just playing a different game than the empiricists play, which means that no amount of insistence on the empiricists' rules can indict legal scholarship any more than strict adherence to the rules of baseball