Re: st: gologit2 model

[Richard Williams <richardwilliams.ndu@gmail.com>]

At 01:10 PM 10/13/2010, Nilam Prasai wrote:
> Thanks Martin and Richard.
>
> Just to make my problem clear. I have all independent variable either
> categorical or dummy except one. Can it leads to negative probability
> as my model do not include any continuous variable. Furthermore, what
> would be the good sample size to run maximul likelihood estimates (how
> would I know if  a small sample size is causing problem to my model).

You are getting negative predicted probabilities, so apparently it is 
possible!  As for sample size, the rule seems to be that you should 
have at least 100 cases for ML analyses, and that is for relatively 
simple models.  I'd probably want several hundred cases for models 
like mlogit and gologit, and/or a model with very few independent variables.

> A revision to my problem:
> I am trying to estimate a generalized ordinal logistic
>  model by using gologit2 program in stata. I have approximately 40
>  observations for which predicted probability is negative. Should I be
>  concerned. My sample size is 115.

I've seen analyses with several thousand cases, and maybe 0 to 3 had 
negative predicted probabilities.  40 out of 115 is extremely 
high.  Are you sure you even need gologit2?  With a sample that 
small, even if assumptions are violated, the violations may not be 
statistically significant.  Again, see the FAQ.  You may need to 
combine categories, drop variables, or choose a different method altogether.


-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW:    http://www.nd.edu/~rwilliam

*
*   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/