ABSTRACT
In Chapters 8 and 9, we dealt with the construction of and criteria for evaluating point estimators for some unknown parameter. Often it is desirable that a point estimate be accompanied by a measure of its precision. Instead of estimating the unknown parameter by some value, we would like to estimate an interval in which we believe the parameter lies. With this in mind, we define a two-sided confidence interval as follows. Suppose X 1,..., Xn is a sample from a distribution indexed by a real parameter θ in Ω. If T 1(X) and T 2(X) are two statistics that satisfy
