For the lettuce yield data, (page 494) it is thought that the yields from fertilizer A might have a larger variance than the yields from fertilizer B.
a. Compute the sample variances s2A and s2B of the yields assigned to A and B, respectively, and the quotient s2A/s2B.
b. Someone suggests using the F test in Section 6.11 for this problem. Is this a good idea? Why or why not?
c. Perform a randomization test of H0 : s2A ≤ s2B versus H1 : s2A > s2B , using the test statistic s2A/s2B, and a minimum of 1000 random outcomes.
(Proceed just as in the example in the text, but for each outcome compute s2A, s2B, and s2A/s2B rather than A̅, B̅, and B̅ − A̅. A fair amount of coding may be required, depending on the software used.)