Sample Size Calculation for Comparing Variabilities
In clinical research, in addition to comparing mean responses between treatment groups, it is also of interest to compare variabilities of responses between treatment groups. In practice, it is a concern that (1) larger variability may result in a worse safety profile and (2) larger variability may decrease the probability of reproducibility of the observed clinical results. Besides, efficacy and safety assessment are often done based on mean responses by assuming that the associated variabilities are the same, which may not be true in many cases. As a result, comparing variabilities of responses between treatment groups has received much attention lately. In order to achieve a desired statistical power when comparing variabilities of responses between treatment groups, commonly used procedures or formulas, which are derived based on comparing mean responses, cannot be applied directly and are necessarily modified.
