Re: st: differences in the ratios of coeffs of 2 nonlinear models

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

--- On Tue, 18/1/11, Mirko wrote:
> I am wondering how to compute the statistical significance
> of the difference in the ratios of coefficients of 2 nonlinear
> models and if a Wald test is the right statistics in this case.
> 
> Please see the example below where I split the sample between
> female and male and estimate a -clogit-.
> 
> clogit choice X Y Z if female==1, group(choiceset)
> nlcom (ratio1: -(_b[X]/_b[Z])) (ratio2:
> -(_b[Y]/_b[Z])),post
> 
> clogit choice X Y Z if female==0, group(choiceset)
> nlcom (ratio3: -(_b[X]/_b[Z])) (ratio4:
> -(_b[Y]/_b[Z])),post
> 
> Now, I'd wish to test ratio1=ratio3 and ratio2=ratio4.
> Unfortunately, this cannot be done by using -test- because the
> results from the -nlcom- relative to the first model are "lost"
> after estimating the second -clogit-.

Why not add interactions between female and X, Y and Z. You
can't add the main effect of female, but you don't need to as
it is "present" in the form of your fixed effects. Than you 
have all the coeficients in one model.

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/