For a random model and the data from exercise 5.12, estimate the
variance components using Type I and Type III SS's with Proc GLM.
You can use the random statement to obtain the expected mean
squares and the /E1 or /E3 option in the model statement to get
the Type I or Type III expected mean squares.
Obtain approximate standard errors for the estimated variance
components.
Create a module that given the vectors y, a, and
b will return estimates of the variance components and
standard errors for a random 2-way crossed classification: