Stata replied that there will be no effort to add a feature that helps
determine expected mean squares terms like in the SAS GLM random
statement. You have to figure it out yourself.
For the Keppel algorithm, see:
<www.psych.uleth.ca/3400/Keppel.pdf>
Fortunately, I just realized the GLM design window in Data Desk 6.2
automatically determines these expected mean squares.
