Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

st: confidence intervals for predicted probabilities for different x values after probit

 From Vincenzo Carrieri To statalist@hsphsun2.harvard.edu Subject st: confidence intervals for predicted probabilities for different x values after probit Date Thu, 3 Nov 2011 12:38:08 +0100

Dear Statalisters,
I want to calculate confidence intervals for predicted probabilites
after probit command. I know that the topic has been already addressed
(http://www.stata.com/support/faqs/stat/ciprob.html), but my case is
different, given that I am not using the command predict, as I want to
set some covariates not at their means. In particular I want to see
how probability of a given outcome changes when age varies from 15 to
79. (The model also contains a swuared age).
Here what I do to calculate the predicted probabilities:

qui forvalues i=15/79{
matrix`i'=mean
matrix`i'[1,1]=`i'
matrix`i'[1,2]=`i'*`i'*
matrix index`i' =`i'*b'
gen prob_`i'=normal(index`i'[1,1])}

Basically, I replace the matrix mean (which all the means of the
covariates) with all the values of age and agesquared (position 1 and
2 in the matrix `i'). I can calculate the estimated probabiltiies, but
I am not able to get the standard errors and then the Confidence
intervals. Is there a way to get them?
Any help is appreciated

Vincenzo
*
*   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/