Sally Clark: Trial and error by Judge and Jury

Sally Clark's 1999 conviction for the murder of her two infant children represented one of the most conspicuous miscarriages of justice in modern English criminal law. A central component of the prosecution case was statistical evidence presented by Professor Sir Roy Meadow, which suggested that the probability of two sudden infant deaths occurring naturally within the same family was approximately one in 73 million. This paper examines three methodological limitations that affect the evidential value of that estimate. First, the family characteristics used to reduce the estimated probability of natural death were not considered in relation to the probability of the alternative explanation, thereby distorting comparative inference. Second, sudden infant death syndrome is a diagnosis of exclusion; population frequency data derived from such a category cannot determine causation in an individual case and are contingent on the limits of postmortem investigation. Third, the probability estimate was presented without reference to the likelihood of competing explanations, limiting its interpretive value. The paper further argues that the significance of the Clark case lies not only in the arithmetical error in the one in 73 million figure but in the broader evidential process by which probabilistic evidence was admitted, framed and interpreted within a criminal trial. The paper also examines the empirical and unstructured approach of English criminal courts to expert evidence. It considers the extent to which subsequent developments, including the Criminal Practice Directions 2023, respond to these difficulties, and compares that position with the contrasting procedural framework associated with the Daubert standard in the United States. The analysis situates the Clark case within broader issues concerning the interpretation, contextualisation and scrutiny of expert evidence in legal proceedings.