Explanation of answers to Quiz #12
With equal sample sizes, and iid normal error terms, the F test
will either be exact, if the variances are all equal, or
anticonservative, if not all of the variances are equal --- it cannot be
conservative in this situation (but can be with an unbalanced design or
certain types of nonnormality).
(Note: 10 out of 14 students answered everything correctly.)