Defining and applying Causal Criteria in the environmental sciences

Our ability to identify causal relations in environmental science is limited by an inability to implement strong experimental study designs, and also by weak theoretical frameworks. In such cases, Causal Criteria assessment – originally developed in epidemiology – is a consistent and robust approach to judging the available causal evidence drawn from multiple studies. We review the definition and application of Causal Criteria in epidemiology and environmental science. We propose standard definitions of the criteria we see as being principally applicable to environmental studies. These criteria are:

• Plausibility: there is a theoretical explanation for the causal link
• Consistency of Association: multiple studies show the same association
• Strength of Association: the magnitude of effects is large relative to background variation
• Coherence of Association: variability in effects is consistent with what would be expected if the association is causal (e.g. dose-response relation)

Predictive performance (deductive predictions based on a putative causal link that are later confirmed) is an additional criterion that will be rarely applicable in environmental studies, but is potentially powerful. The application of Causal Criteria in environmental sciences will reduce issues of weak inference and lead to the better utilisation of the large body of existing work.