Full_Name: Katrina Chu Version: R 2.7.2 GUI 1.25 (5217) OS: Mac OS X Version 10.4.11 (Tiger) Submission from: (NULL) (131.202.107.54) I've been using Cochran's Test to check for outlying variance. I noticed that the degrees of freedom that are calculated and output are incorrect. The author has forgotten to take into account that df=n-1 and appears to have just left df=n. n=# samples per treatment and in the case where n is unequal across treatments, I believe that the average samples per treatment can be calculated, then minus 1.