Re: st: Predicted probabilities in a competing risks model in discrete time?

[Steven Samuels <sjhsamuels@earthlink.net>]

Katharina-

When you give vague references, you effectively ask that we spend  
time figuring out exactly what the references are. If the documents  
are long, you are also asking us to search them for the pages that  
pertain to your topic.  I find this inconsiderate. It is also self- 
defeating, because it discourages responses.  Please provide precise  
references, with page or section numbers, as requested in the  
Statalist FAQ, Section 3.4: http://www.stata.com/support/faqs/res/ 
statalist.html#advice.

-Steve

On Aug 4, 2008, at 11:51 PM, LachZak@web.de wrote:


> Dear Statalisters,
> I have estimated a competing risks model in discrete time,  
> following the approach by Allison (82)/Jenkins(95), i.e. estimating  
> a "multinomial logit" model applied to reorganised data. I would  
> like to interpret some of the regressors in more detail. Does it  
> make any sense to calculate predicted probabilities - for example  
> for a reference person and for a person who differs in just one  
> characteristic? Can I interpret these predicted probabilities in  
> the same way as in a "normal" multinomial logit model?
> I would be really glad for any help.....
> Thanks in advance!
> Katharina
> *
> *   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/
*
*   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/