ABSTRACT
In the previous chapter we delineated reasons for the importance of stressing observables. In this chapter we shall be concerned with various ways that have been devised to accommodate statistical predictions that are not Bayesian. These will include the classical confidence approach, various methods that depend on the likelihood, on loss functions, and on sample reuse procedures. The material here is largely for those who are inclined toward predictive analysis but are not entirely comfortable with a Bayesian approach. However, they are advised to continue with subsequent chapters to become acquainted with the large variety of problems for which the Bayesian approach provides solutions.
