Adjusting Belief Structures
针对一组随机量，当获得部分观测值时，如何调整对其他量的不确定性。提出信念结构方法，通过观测调整信念，并推导性质，示例包括回归诊断和局部回归设计。
SUMMARY Suppose that we have made various belief statements, but not a full probability specification, about a collection of random quantities. How do we modify uncertainties over one subset of the quantities, when given the values of another subset? We suggest an approach in which belief statements are organised into belief structures which are adjusted by observation on related belief structures and derive various properties of such adjustments. We give two examples. In the first example, belief adjustments are related to case statistics for regression diagnostics. In the second example adjusted structures are used to suggest designs for the localised regression model.