Review Articles · Review R3

On Limiting the Use of Bayes in Presenting Forensic Evidence

Norman Fenton, Martin Neil

Forensic Science Seminar · 2014 · Volume 4 · Issue 1 · pp. 8-23

Original Abstract

Abstract

A 2010 UK Court of Appeal Ruling (known as “R v T”) asserted that Bayes theorem and likelihood ratios should not be used in evaluating forensic evidence, except for DNA and ‘possibly other areas where there is a firm statistical base’. The potential impact of this ruling is enormous and it has drawn fierce criticism from expert witnesses, academics and lawyers, who have identified various weaknesses and fallacies in the ruling. This paper focuses on the strategic and cultural challenges that the ruling raises to ensure that the role of Bayes is better understood and exploited in the presentation of forensic evidence. We provide a simple unifying way of describing all probabilistic forensic ‘match’ evidence; this enables us to easily identify and avoid the kind of common misunderstandings and fallacies that have afflicted probabilistic reasoning about evidence, including especially why it is irrational to assume that some forensic evidence is ‘statistically sound’ whereas other less established forensic evidence is not. But these misunderstandings are not restricted to lawyers, since we show that both forensic scientists and even Bayesian experts have consistently failed to include all relevant information in their evidence, such as error probabilities, and this applies to DNA as much as any other forensic science. We also show that there are severe limits of the extent to which the results of Bayes can be presented in purely intuitive terms; we show that the scope in forensics is even narrower than previously assumed. Hence, there are two major challenges facing the opponents of the R v T ruling: First, there must be much greater awareness of the need to improve Bayesian forensic arguments (before they are even presented in court) in order to avoid the common errors and omissions that are made. Second, there must be a radical rethink on the strategy for presenting the results of Bayesian arguments in court. Resorting to the formulas and calculations in court is a dead end strategy since these will never be understood by most lawyers, judges and juries, but the intuitive presentations simply do not scale up. Ultimately this means getting the lay observers to ‘accept’ that they need only question the prior assumptions that go into the Bayesian calculations and not the accuracy or validity of the calculations given those assumptions. Bayesian networks may provide a suitable mechanism for performing these calculations.

Keywords: Forensic science; Bayes; Evidence

Authors & Publication

Article Details

Citation Information

Metadata Quality