st: mlogit and collapsed estimates

["Vahidnia, Farnaz" <FVahidnia@bloodsystems.org>]

Dear statalisters:

I am trying to run a logistic model to estimate OR for a group variable (y=0, 1, 2) associated with a number of independent variables (x1, x2, x3, ...). 
This is the first time I am using "mlogit". Using mlogit and y=0 as a base, I get a set of  separate coefficients for y=1 and y=2 associated with each x. 

However, I am also interested in estimating collapsed OR associated with some Xs, let's say x4 and x5, for y=1 and 2.
In other words I need to have coefficients for certain xs as if Y is binary (0, 1). I guess a "nested model" is what I need, but my data is single observation per subject, therefore "nlogit" won't work. 

I don't know if it is correct to add coefficients for y1x4 and y2x4. Even if it's correct, manually estimating 95% confidence intervals needs covariance matrix and I don't know a STATA command to get them directly. 

Any suggestions? 

Thanks,
Farnaz





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