ABSTRACT
This chapter discusses a variety of approaches that have been considered for char-
acterizing statistical evidence and reasons why it is believed that these are not fully
successful. Our general point is that any valid theory of statistical inference must at
least prescribe how to measure statistical evidence. This seems essential if we are go-
ing to speak of “the statistical evidence.” Many theories of inference stop somewhat
short of stating explicitly how to measure statistical evidence yet still make reference
to it. This leads to confusion and ambiguity, especially with respect to how one is
supposed to conduct a statistical analysis. While there can be many problems with
statistical analyses, particularly when these are carried out on observational data and
don’t account for inherent biases, the failure to use a tool appropriate to the task
of assessing evidence can be a contributing factor in any deficiencies that statistical
theory brings to this task.
