Remedying Daubert's Inadequacy in Evaluating the Admissibility of Scientific Models Used in Environmental-Tort Litigation*
Swinehart, Matthew W., Texas Law Review
Is farmed salmon safe to eat?1 How can we best improve the water quality of an estuary?2 What are the health benefits-and corresponding economic trade-offs-of improving air quality in the developing world? Does emissions trading actually improve air quality?4 What role do humans play in global climate change?5 How can we prevent another Hurricane Katrina from ravaging the Gulf Coast?6 These questions and countless others have prompted scientists, politicians, regulators, and industry experts to turn to environmental models for answers.
An "environmental model" is a model-a mathematical representation of an object or process-that seeks to describe physical and natural systems, such as the fate and transport of contaminants in groundwater.7 When used appropriately, models produce meaningful solutions to problems that are otherwise seemingly incomprehensible.8 Successful applications of environmental models can provide invaluable assistance to researchers and policy makers in solving complex problems and striking delicate balances between adverse constituencies.9 Models are valuable, perhaps invaluable, tools to understand cause and effect, particularly in the data-deficient world of environmental protection. However, when models go wrong, they can have devastating real-world consequences. Take, for example, the models that some say caused undue reliance on New Orleans's (inadequate) levee and floodwall infrastructure, perhaps dramatically increasing the destruction caused by Hurricane Katrina in 2005.10
But models are not just for scientists and governmental agencies. Litigants use them every day in environmental-tort litigation to tackle issues involving harmful exposure to carcinogens, groundwater contamination, and even global warming." The Supreme Court's decision in Dauben v. Merrell Dow Pharmaceuticals, Inc}2 is charged with the perhaps impossible task of evaluating all types of scientific evidence with a single approach.13 Unfortunately, it has proven especially inappropriate and inadequate in evaluating model reliability. The problems that plague all models-namely uncertainty and a lack of transparency-are exacerbated in environmentaltort litigation. The adversarial nature of litigation means models are often used to obfuscate rather than clarify empirical evidence. And, because modeling is a particularly technical scientific field, it is even more challenging for judges and juries who may be unfamiliar with even the most basic scientific issues posed by use of a model.
Daubert is not the appropriate way to evaluate model reliability, or at least it requires significant amendment to provide a credible analytic process. This Note analyzes Daubert's treatment of environmental models, finds it ultimately lacking, and suggests reformulating the inquiry to address modeling's unique challenges. It presents a newly tailored checklist for judges and juries to ensure that: (1) the modeler is an expert in the modeling methodology employed; (2) the correct model was chosen and applied within an appropriate scope; (3) the real-world system is not too complex for legitimate modeling given the current level of theoretical and empirical scientific understanding; (4) actual and potential conflicts of interest are identified and mitigated; (5) the model considered all available and relevant empirical evidence; (6) the modeler's results are capable of being communicated effectively; and (7) judicial decision makers look beyond the constituent parts of a model and base their decision on the weight of the evidence.
Part II of this Note examines the incredible societal value models provide, particularly in the environmental sciences. However, as we will see in Part III, this value is not unmitigated because the use of models poses significant risks and problems for decision makers. Part IV discusses the Supreme Court's admissibility guidelines for scientific evidence in Daubert and demonstrates that it provides inadequate guidance for judges and juries in discerning useful models. …