Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: how to get marginal effect when dependent variable has ln function?


From   Maarten buis <maartenbuis@yahoo.co.uk>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: how to get marginal effect when dependent variable has ln function?
Date   Tue, 23 Jun 2009 12:30:51 +0000 (GMT)

--- On Tue, 23/6/09, gjhxmu@sina.com wrote:
> Austin mentioned that "Each observation may have a
> different marginal effect of x1 on y". 
> 
> I doubt that and I calculate below:
> 
> So as long as x1 changes one unit, the change percent of y
> is constant, not vary with various levels of x1.

The marginal effect usually means the additive change in y 
for a unit change in x rather than the percentage change in
y for a unit change in x. So Austin is right for the usual 
definition of marginal effects. However that does not mean
that the percentage change in y for a unit change in x is 
a wrong way of representing the size of the effect, it is 
only not a marginal effect.

> However, why are the results of -reg- and -glm- different?

With -glm- you will also have to specify the -eform- option,
though this still will give slighly different results. This
due to the fact that the -regress- output will in this case
give the percentage change in the geometric mean of y for a
unit change in x, while -glm- will give the percentage 
change in the arithmatic mean of y for a unit change in x. 

This was discussed in the refference which I gave you 
yesterday and is freely downloadable.

-- Maarten

-----------------------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://home.fsw.vu.nl/m.buis/
-----------------------------------------


      

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index