Model evidence-based judgment

II True judgments Model seen to describe true risk Expert judgment as truth-approaching Uncertainty sporadically mentioned Judgment aiming at truth Risk is a property of the world. Based on hard evidence and judgment Uncertainly related to imprecision of underlying true risk... 

VI Argument view Model is seen as an expression of argument Quantitative and qualitative Judgment Risk results from an interpretation. Based on hard evidence and judgment. Adversarial approach. 

In the above risk description, the assessor makes a judgment regarding his degree of belief P, about the occurrence of the event A based on the evidence available to him, which enables him to provide a justification J, i.e. P(A J). This evidence may consist of data, models, theories and judgments. As discussed in Section 3, there exists a variety of propositional attitudes regarding evidential propositions. Evidence can be believed, known or accepted. 

When human data are unavailable or limited, the risk assessment must be based on extrapolation from animal data, short-term tests, and other information. Under these circumstances, the evidence must be weighed with reference to its predictive accuracy, and expert judgment must be exercised in selecting the appropriate model for extrapolation toman. 

Knowledge-based computer systems can be used to assist the scientist by providing an inferential framework and access to relevant databases, including results of previous cases. It would be very difficult to construct a comprehensive statistical model to deal with all cases involving glass evidence. A compromise approach that combines statistical methods with expert experience and judgment is likely to be the most effective. 

Furthermore, there often is a form of evidential uncertainty for the assessor in relation to evidential propositions. Finally, it is possible that the assessor has accepted data, models or judgments the justification of which is (partly) based on values. Hence, the justification may not only of Epistemic nature (Ej), but also may involve of Non-Epistemic nature (NEJ). Hence, the risk description is not necessarily value-free, depending on the nature of the reasons because of which a particular propositional attitude is adopted regarding the evidences. A qualitative assessment of evidential biases can make such value-laden considerations explicit. 

The example shows that uncertainty can be minor even if only one type of evidence is available. In case no models or data is available and the assigned probability is purely based on expert judgment, the level of expertise and the degree of consensus may still lead to the conclusion of relatively minor uncertainty related to the probability value. 

---