Presenting Scientific Theories Within Risk Assessments

ABSTRACT

We propose a structure for presenting risk assessments with the purpose of enhancing the transparency of the selection process of scientific theories and models derived from them. The structure has two stages, with 7 steps, where the stages involve two types of theories: core and auxiliary, which need to be identified in order to explain and evaluate observations and predictions. Core theories are those that are “fundamental” to the phenomena being observed, whereas auxiliary theories are those that describe or explain the actual observation process of the phenomena. The formulation of a scientific theory involves three constitutive components or types of judgments: explanative, evaluative, and regulative or aesthetic, driven by reason. Two perspectives guided us in developing the proposed structure: (1) In a risk assessment explanations based on notions of causality can be used as a tool for developing models and predictions of possible events outside the range of direct experience. The use of causality for development of models is based on judgments, reflecting regulative or aesthetic conceptualizations of different phenomena and how they (should) fit together in the world. (2) Weight of evidence evaluation should be based on falsification principles for excluding models, rather than validation or justification principles that select the best or nearly best-fitting models. Falsification entails discussion that identifies challenges to proposed models, and reconciles apparent inconsistencies between models and data. Based on the discussion of these perspectives the 7 steps of the structure are: the first stage for core theories, (A) scientific concepts, (B) causality network, and (C) mathematical model; and the second stage for auxiliary theories, (D) data interpretation, (E) statistical model, (F) evaluation (weight of evidence), and (G) reconciliation, which includes the actual decision formulation.

KeyWords:

• falsification
• causality
• core and auxiliary hypotheses
• evaluative
• explicative
• regulative.

ACKNOWLEDGMENTS

We thank the reviewers and editor for the many helpful comments that improved the paper.

Notes

Jardine (2003) is an article that discusses Gerd Buchdahl's interpretation of Kant's philosophy of science. What we are calling regulative or aesthetic is included under a heading of “systematic.” What we are calling “evaluative” is labeled “probative.”

One account of a well-known incident (Edmonds and Eidinow 2001, p 18), where Ludwig Wittgenstein is said to have waved a poker in a threatening fashion toward Karl Popper, has that Wittgenstein lifted the poker to use as a tool to make a point about causation.

This do function was defined in contrast to conditional probabilities, to reflect the effects of specific actions. Thus, the conditional probability, prob(x | Y = y), is not necessarily equal to the prob(x | do(Y) = y).

Lakatos (1974, p 123) gives a quote from Leibniz: “It is the greatest commendation of an hypothesis … if by its help predictions can be made even about phenomena or experiments not tried.”

Lewis (1983) provides the following formal definition of causality: “Let E be the proposition that e exists or has occurred, and C be the proposition that c exists or has occurred, then c causes e if and only if, 1) C and E are true; and 2) For some non-empty set, L, of true-law propositions and some set, F, of true propositions of particular fact, L and F jointly imply C ⊃ E, although L and F jointly do not imply E, and F alone does not imply C ⊃ E.” Note X = A ⊃ B (material implication) means that if A and B are true, X is true; if A and B are false, X is true, if A is true and B is false, X is false; and if A is false and B is true, X is true.

Popper (1959) gives a simple example showing that a simple conditional probability given evidence is not a good measure for support of hypotheses. Assume: h₁ = {1}; h₂ = {1, 3, 4}; and e = {1, 2, 4}, where it is assumed that the probabilities of each integer individually is ¼. Thus, p(h₁|e) = 1/3 > p(h₁) = ¼, so e would support h₁, and p(h₂|e) = 2/3 < p(h₂) = ¾, so e would not support h₂, but p(h₂|e) > p(h₁|e). Note that p(e) is relatively large, whereas for most statistical applications p(e) can be assumed to be very small.

Additional information

Notes on contributors

Harry Marks

This article is not subject to U.S. copyright law. The opinions expressed herein are those of the authors and may not reflect in any way opinions or policy of the U.S. Department of Agriculture