Dear Statalist,
I am trying to manually calculate the average partial effects (APEs) and
their standard errors, and in order to verify them I compare the results
with results from -margeff-.
Using -predictnl- my manual results are exact when looking at the
coefficients but far away when looking at the Std. Errors.
The results from my std. errors are closer to the ones from the marginal
effects (computed both by -margeff- and by -mfx-).
Obviously I am doing something wrong and I would really appreciate if
somebody could give me advice:
A) On how to compute the std. Errors of the APEs manually.
B) On why the difference between the standard errors of the APEs and the
marginal effects is so large, giving completely different results
regarding the significance of the estimates.
Below is an example for the first category of -gologit2- estimation for
a dummy variable.
Thank you
Manos
*############# Start example ################
clear
use auto
gen y=1 if length>0 & length<170
replace y=2 if length>=170 & length<220
replace y=3 if length>=220
gen x1=1 if weight>0 & weight<3190
replace x1=0 if weight>=3190
rename trunk x2
gologit2 y x1 x2 , link(probit)
margeff
predictnl ape1 = normprob(- (xb(1) + (_b[1:x1]*(1-x1)))) - normprob(-
(xb(1) - (_b[1:x1]*x1))), se(se1)
sum ape1 se1
margeff, at(mean)
mfx, predict(p outcome(1))
*########### End example #####################
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/