Judges are the gatekeepers of evidence. Arguably, the most difficult duty for a judicial gatekeeper is to screen the reliability of expert opinions in scientific fields such as medicine that are beyond the ken of most judges. Yet, judges have a duty to scrutinize such expert opinion evidence to determine its reliability and admissibility. In toxic tort cases, the issue of causation-whether the alleged exposures actually caused the plaintiffs injury-is nearly always the central dispute, and determining admissibility of expert causation opinion is a daunting challenge for most judges. We present a comprehensive review of the courts' struggles with the screening of scientific evidence in such cases. In addition, we propose an approach to the screening of causation opinions based on probability science and logic. Central to this approach is Bayes' Law, a statistical tool that courts can use to analyze the extrinsic reliability of proffered causation testimony. We explain Bayes' Law and illustrate its potential application for evaluating the reliability of medical and scientific causation testimony.
All evidence is probabilistic. There are uncertainties attending all testimony, not only because the honesty or objectivity of witnesses may be doubtful, but also because even honest and unbiased witnesses may be mistaken in their perceptions. Reliability of causation evidence depends on both sensitivity and specificity of the tests used to determine causation. Highly sensitive tests of causation reflect an ability to identify a high percentage of those with the agent-induced disease, whereas highly specific tests of causation reflect an ability to reject a high percentage of those who have the disease, but not induced by the agent at issue. According to Bayes' Law, the reliability of causation opinion depends not only on the sensitivity and specificity of the tests employed by the causation expert, but also on the base rate of the agent-induced disease in the population. Bayes' Law dictates that the lower the rate of the agent-induced disease in the population, the less reliable the opinion that the agent at issue in fact caused the plaintiffs disease given certain levels of sensitivity and specificity. The base-rate problem and its effect on reliability of causation opinions are overlooked by judges when scrutinizing the reliability of proffered causation evidence. In this Article, we encourage courts to consider a Bayes' Law approach to screen out, at an early stage, those claims of injury lacking reliable evidence that an injury was more likely than not caused by exposures to toxic agents.
The goal of our Article is to provide a framework that helps the gatekeeper to screen out toxic tort claims insufficiently substantiated by the underlying scientific and medical data, and allow the factfinder to decide only those toxic tort claims for which there is reliable and relevant scientific support for each link of the causal chain, from subject exposure to the injury Scientific substantiation of each causal link determines the reliability of an experts opinion that the exposure more likely than not caused the plaintiffs injury.
Neal C. Stout & Peter A. Valberg,
Bayes' Law, Sequential Uncertainties, and Evidence of Causation in Toxic Tort Cases,
U. Mich. J. L. Reform
Available at: https://repository.law.umich.edu/mjlr/vol38/iss4/3