panel ordered regression [was: Re: st: Is there anything in logit similar to the absorb option in areg?]

[Maarten buis <maartenbuis@yahoo.co.uk>]

--- On Wed, 26/5/10, Richard Williams wrote:
> It is always annoying/surprising when something that works
> with OLS does not generalize to other methods. Does that
> post help to explain why there is no such thing as an
> xtologit or xtoprobit command?  I keep on getting requests
> to write xtoglm and xtgologit2 commands, and the fact that 
> there is no xtologit/ xtoprobit command makes me wonder if
> it would be a bad idea.

My impression is that fixed effects estimators are harder to
generalize to non-linear models than random effects estimators,
so probably the easiest way to implement such a model is to 
implement a random effects version using the the tools 
disucssed in the special issue on maximum simulated likelihood 
of The Stata Journal 6(2):
<http://www.stata-journal.com/sj6-2.html>
 
> This also ties in with our discussion of logit versus
> probit the  other day.  It seems to make little difference
> which you use for basic models, but once you get into more
> advanced models there may be good reasons for choosing one
> link over the other.

I don't forsee any difference between the link function for 
the random effects, maximum simulated likelihood estimator.

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


      

*
*   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/