The conjunction paradox has fascinated generations of scholars, primarily because it brings into focus the apparent incompatibility of equally well accepted conventions. On the one hand, trials should be structured to reduce the total number, or optimize the allocation, of errors. On the other hand, burdens of persuasion are allocated to elements by the standard jury instruction rather than to a case as a whole. Because an error in finding to be true any element of the plaintiff's cause of action will result in an error if liability is found, errors on the overall case accumulate with errors on discrete issues. This, in turn, means that errors will neither be minimized nor optimized (except possibly randomly). Thus, the conventional view concerning the purpose of trial is inconsistent with the conventional view concerning the allocation of burdens of persuasion. Two recent efforts to resolve this conflict are examined in this article. Dean Saul Levmore has argued that the paradox is eliminated or reduced considerably because of either the implications of the Condorcet Jury Theorem or the implications of super majority voting rules. Professor Alex Stein has constructed a micro-economic explanation of negligence that is also offered as resolving the paradox. Neither succeed, and both fail for analogous reasons. First, each makes a series of ad hoc adjustments to the supposedly formal arguments that are out of place in formal reasoning. The result is that neither argument is, in fact, formal; both arguments thus implicitly reject the very formalisms they are supposedly employing in their explanations. Second, both articles mismodel the system of litigation they are trying to explain in an effort to close the gap between their supposedly formal models and the reality of the legal system; and when necessary corrections are made to their respective models of litigation, neither formal argument maps onto the reality of trials, leaving the original problem untouched and unexplained. These two efforts are thus very much similar to the failed effort to give a Bayesian explanation to trials and juridical proof, which similarly failed due to the inability to align the formal requirements of subjective Bayesianism with the reality of modern trials. We also explore the reasons for this consistent misuse of formal arguments in the evidentiary context. Rationality requires, at a minimum, sensitivity to the intellectual tools brought to a task, of which algorithmic theoretical accounts are only one of many. Another, somewhat neglected in legal scholarship, is substantive explanations of legal questions that take into account the surrounding legal landscape. As we show, although the theoretical efforts to domesticate the conjunction paradox fail, a substantive explanation of it can be given that demonstrates the small likelihood of perverse consequences flowing from it. The article thus adds to the growing literature concerning the nature of legal theorizing by demonstrating yet another area where legal theorizing in one of its modern conventional manifestations (involving the search for the algorithmic argument that purportedly explains or justifies an area of law) has been ineffectual, whereas explanations that are informed by the substantive contours of the relevant legal field have considerable promise.
Date of this Version
Ronald Jay Allen and Sarah Lively, "Burdens of Persuasion in Civil Cases: Algorithms v. Explanations" (January 2004). Public Law and Legal Theory Papers. Working Paper 17.