re: st: Multiple nonlinear equations regression with categorical variables

[Christopher Baum <kit.baum@bc.edu>]

<>
I have one dependent variable (continuous) that I would like to explain 
with five independent variables. Four of them are continuous and one is 
categorical. I am looking for a model that would allow me to fit 
different
 nonlinear (exponential, logarithmic) as well as linear equations to 
each of the continuous independent variables. I am also interested in 
testing some of
 the interaction terms in the same model, most notably between the 
categorical and the continuous independent variables. Is there a Stata 
command that 
would allow me to do this? Ideally, the model would also provide some 
indication of which main and interaction terms should be included for 
greatest parsimony.

You can readily fit such models with Stata's regress command. You need merely transform the independent variables as you wish and then estimate the linear regression. These are not 'nonlinear equations', that is, not nonlinear in the parameters, so standard regression will suffice.

Stata 11's support for factor variables (categorical variables) and interactions between them and continuous variables makes it much easier to estimate such interaction models, and test for the significance of interactions, than in earlier versions of Stata.

Kit



Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
                              An Introduction to Stata Programming  |   http://www.stata-press.com/books/isp.html
   An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html


*
*   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/