Abstract

Given data, central issues that any theory of statistical inference has to handle, concerning a quantity of interest
prescribed by the application, are the following:

(i) What does the evidence in the data say is the best choice of a value for a quantity of interest (estimation)?
(ii) Does a quantity of interest take a particular value (hypothesis assessment)?
(iii) Furthermore, any such theory should also say something about how strong the evidence is. Just as an estimate
without an assessment of its accuracy is useless, hypothesis assessment calls for the quantification of evidential strength.

Given the centrality of the evidence it seems natural that a characterization of how statistical evidence is to be measured
should play a primary role in determining the theory of inference. The purpose of this session is to consider recently developed
approaches to the development of a theory of statistical inference that include an explicit characterization of how statistical
evidence is to be measured. Such a theory has the potential to remove many of the ambiguities/paradoxes that currently cause
problems for the application of statistical methodology.