About statistical power and replicates in experimental design