Many thanks for the detailed discussion, the references, and corrections
to my thinking. I will be experimenting with this over the next week.
I will probably post a separate question about the perturb.ado. The few
discussions in the StataList concerning perturb have not been
encouraging, and as Steve noted- John Hendrickx's website (the author)
has limited documentation and the the outbound links to other references
are broken.
