More fuel for the fire... As you can see, I'm lazy. BUT, I have the
newer version of tabi (Marcello has the older one) but an older Stata
executable than Tom. My combination gives the right answer.
It looks like the change in the algorithm occurred with the Oct. 6, 2004
executable. So, if you have an earlier version, I guess you will get the
right results, but it may take forever to get them depending on the
problem. As was noted in messages a few months ago, the improvement in
speed is dramatic.
