Finite sample corrections for average equivalence testing

Average (bio)equivalence tests check whether two treatments produce responses that fall within the same acceptable range. The standard two one-sided tests (TOST) method is easy to use but often becomes overly conservative, especially for highly variable drugs, leading to very low statistical power. We introduce the $\alpha$ -TOST, a finite-sample correction of TOST that adjusts its significance level to guarantee a true type-I error rate of $\alpha$ . This correction substantially increases power while preserving validity. The procedure is simple to compute, uniformly more powerful than TOST, and performs better than competing methods. A case study on econazole nitrate skin deposition illustrates its practical advantages.