Dear Listers,
A new version of margeff is available through my website which improves the estimation of average
partial effects. In the old distributions, reported average partial effects for continuous variables with small unit
of measurement were not reliable. Estimation of partial effects requires adding the unit of measurement to the linear prediction.
The problem was that the addition could have produced nothing for some observations because the linear prediction was not forced
to be a double variable. This is now fixed.
To install the update, please type
net from http://web.uni-corvinus.hu/bartus/stata
net install margeff , replace
or visit the above website.
Furthermore, the sixth edition of W. Greene's Econometric Analysis (pages 784-785) addresses the issue of how to
estimate correct standard errors of average partial effects. I replicated his empirical analysis reported on page 785.
I found that margeff standard errrors are appropriate by Greene's standards. I am surprised because
I was not aware of the problem he raises, still the margeff estimator, which seems to be a naive one,
works as if it implemented the appropriate estimator proposed by Greene.
Or do I miss something? Comments are welcome ...
Tamas
------------------------------------------------
Tamas Bartus, PhD
Associate Professor, Institute of Sociology and Social Policy
Corvinus University, Budapest
1093 Budapest, Közraktár utca 4-6.
Phone: +36-1-482-7301
Fax: +36-1-482-7348
Homepage: www.uni-corvinus.hu/bartus
*
* 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/