The interval hypothesis with applications to the problem of bioequivalence

A new procedure is proposed to test the interval hypothesis, which tests whether the mean of a population is within an interval (A,B) or not. Exact and asymptotic distributions of the test statistic will be studied for both known and unknown standard deviation $\sigma$. The power function is investigated for each case. The new method is generalized to testing whether the difference between two means is within a given interval or not. Two sampling designs are considered, the parallel and the crossover design. The new procedure is compared with methods currently used in testing for bioequivalence in pharmaceutical research. A comparison of the rejection regions of the different procedures is presented. A Monte Carlo study demonstrates the behavior of the new method compared to the other procedures.