st: triprobit, mvprobit and suest

["Ravindran, Kiron" <kravindran03@exchange.uci.edu>]

How do I decide which method to adopt to estimate my model that looks like this:

y1 = v1 + cv1 + cv2 + cv3 +cv4
y2 = v1 + cv1 + cv2 + cv3 +cv4
y3 = v1 + cv1 + cv2 + cv3 +cv4  where all except cv3 and cv4 are binary.

Theoretically this could also be a a recursive model in the following manner

y1 = v1 + cv1 + cv2 + cv3 +cv4
y2 = v1 + cv1 + cv2 + cv3 +cv4 + y1
y3 = v1 + cv1 + cv2 + cv3 +cv4 + y1 + y2

I am interested in estimating Pr(y1=1, y2=1, y3=1 | v1=1) - Pr(y1=1,
y2=1, y3=1 | v1=0). I was using margeff for some related biprobit
models I ran but I don't know how to the same with three dependent
variables. .
(sample size is 43)

With the triprobit and mvprobit, the recursive model does not
converge. But suest seems to give out a result in no time, what is it
doing differently. (I looked up the help files, Greene, Lahiri and
Schmidt 78 and Knapp and Seaks 98 but got nothing definitive)

Any help appreciated
Kiron

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/