Assumptions (Section 5.4)

Test scores are mutually independent
   *Assign subjects to treatments randomly, test them indivdually

Test scores are normally distributed
  * Check it out
          if not, use F-values with increased significance levels

Variances of the scores are equal
  * Check it out (e.g., Keppel et al., 119f,
                             Brown-Forsythe procedure test)
          if not, use F-values with increased significance levels