st: Regoprob vs Gologit Ordinal Level Panel Data

[bcg2104@columbia.edu]

Dear Statalist,

I have ordinal level panel data that violates the parallel lines
assumption. I read a thread on the archives that, in addition to regoprob
or gllamm, suggested just a simple gologit2 with clusters. I've run both
regoprob2 and gologit2 link(probit) with clusters and the results are very
different. Can anyone help me understand exactly what these two models do
differently? I assume that regoprob2 is the more conservative
option with panel data. Under what circumstances (if any) would it be
justifiable to use gologit2 with clusters instead? Thanks!

Cheers,
Brooke Greene
*
*   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/