Various real and imagined criminal law cases rest on “naked statistical evidence”. That is, they rest more or less entirely on a probability for guilt/liability derived from a single statistical model. The intuition is that there is something missing in these cases, high as the probability for guilt/liability may be, such that the relevant standard for legal proof is not met. Here we contribute to the considerable debate about how this intuition is best explained and what it teaches us about evidential reasoning in the legal setting. We part ways with the recent scholarship, however. Unlike most others, our diagnosis is not that there is an important qualitative property that some evidence or bodies of evidence have, and that naked statistical evidence, most strikingly, lacks. Rather, we see cases resting on naked statistical evidence as lying at the extreme of a continuum of cases that are vulnerable to challenge due to “meta-uncertainty” about the underlying model.

各种真实和想象的刑法案件都依赖于“赤裸裸的统计证据”。也就是说，它们或多或少完全依赖于从单一统计模型得出的犯罪/责任概率。直觉是这些案件中缺少某些东西，犯罪/责任的可能性可能很高，因此不符合相关的法律证明标准。在这里，我们参与了关于如何最好地解释这种直觉以及它对我们在法律环境中进行证据推理的教导的大量辩论。然而，我们与最近的奖学金分道扬镳。与大多数其他人不同，我们的诊断并不是某些证据或证据体具有重要的定性特性，而赤裸裸的统计证据，最引人注目的是，缺乏。相当，