As a simple example, interval analysis is always specifies a symmetric bound around the mean, which is wasteful if there's interested in only one side. A statistical hypothesis, however, can be specified to provide a one-tail test, which moves the analysis to the side of the mean of interest. The amount of information remains the same in both cases, but hypothesis testing puts the information to more efficient use than does interval analysis.
This page last modified on 2 March 2005.