The Law's Aversion to Naked Statistics and Other Mistakes

Abstract

A vast literature has developed probing the law's aversion to statistical/probability evidence in general and its rejection of naked statistical evidence in particular. This literature rests on false premises. At least so far as US law is concerned, there is no general aversion to statistical forms of proof and even naked statistics are admissible and sufficient for a verdict when the evidentiary proffer meets the normal standards of admissibility, the most important of which is reliability. The belief to the contrary rests upon a series of mistakes: most importantly, mismodeling of the structure of legal systems and the nature of common law decision making. Contributing to these mistakes is the common methodology in this literature of relying on weird hypotheticals that mismodel the underlying legal relations and contain impossible epistemological demands. Collectively, these phenomena have distracted attention from issues that actually affect real legal systems.

Preponderance, proportionality, stepwise liability

“…5 Knowing that the law is like this, companies could strategically exploit the preponderance rule by maintaining market shares not greater than 50%, in order to avoid liability at the expense of victims. (However, the scenario in which there is only ‘naked statistical evidence’ is rather unlikely—see Allen and Smiciklas, 2022. )…”

“…Suppose, to elaborate on the story, that one of the companies had 80% of the market; everything else being equal, one could infer from this statistical information that, with .8 probability, plaintiff took the pills of that company. But this would merely be an inference from 'naked statistical evidence', which has its own problems, as is well known (see Allen and Smiciklas, 2022;Schauer, 2003). Note that, if that statistical information were accurate and accepted as sufficient evidence, plaintiff would be entitled to full compensation (100%) from that single company-which seems to be unfair for rather obvious reasons.…”

“…5 Knowing that the law is like this, companies could strategically exploit the preponderance rule by maintaining market shares not greater than 50%, in order to avoid liability at the expense of victims. (However, the scenario in which there is only 'naked statistical evidence' is rather unlikely-see Allen and Smiciklas, 2022. ) These complications indicate that one must consider specifically what statistical and probability values apply to (more on this below under 'Conceptual concerns: Probability and burdens of proof').…”