st: graphing predicted probabilities with quadratics

[David Torres <torresd@umich.edu>]

I'm working on trying to graph predicted probabilities from logit  
models that include quadratic terms.  Currently, I am using the  
predict command postestimation, then using the graph twoway command.   
Obviously, this method is ignoring the other variables in the model,  
which I'd like to hold at the mean or zero.

The x axis is continuous years, the y axis is the predicted  
probabilities, and I have six categories of a variable I'd like to  
compare.  Below is the code for the logit command:

xi: logit BAfail i.agecat male black hispanic otherrace biomahighgrade  
biopahighgrade intactfam hhsizeincep lnhhinc hhpovratincep ///
   overallgpa asvabpcnt evtkactsat parochial privtype northcent south  
west urban unknownurb cumpub4 cumpriv4 cumprof4 ///
   cumpub2 cumpriv2 cumprof2 cumpub4sq cumpriv4sq cumprof4sq  
cumpub2sq cumpriv2sq cumprof2sq if iptw<. [pw=sampwgtr1],  
cluster(pubid) nolog


As you can see, there are six "cumulative years in school" predictors  
followed by their squared terms; these are at the end of the command.

What is the best way to get to what I want?

Diego
*
*   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/