Abstract
Generic drugs and drugs adopted for the use of alternative routes of administration can be approved by an only bioequivalence(BE) evaluation for human subject. In Japan, following Health, Labour and Welfare Ministry(1997), we evaluate BE based on confidence interval for the difference of the population means between test drug and reference drug. On the other hand, the above evaluation process includes various statistical problems. So it is an urgent need to investigate characteristic of BE systematically in practical clinical evaluation. In this paper, we focused on a standard BE-evaluation on the change-over design and presented a process of statistical inference assuming the power-normal distribution(Goto et al., 1979, 1983) as the underlying distribution of observations, where the power-normal distribution is defined as the distribution specified before the power-normal transformation.
