Statalist


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

st: RE: RE: RE: marginal effects in probit


From   "Martin Weiss" <martin.weiss1@gmx.de>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: RE: RE: marginal effects in probit
Date   Fri, 31 Oct 2008 14:46:31 +0100

It is arguable that -dprobit- delivers added value with regard to
significance and CIs as it continues to display them for the underlying
-probit- while mfx changes them with regard to the values of the covariates
supplied by the author. 

******************
sysuse auto, clear
probit f rep we tr pr, nolog
dprobit f rep we tr pr, nolog
mfx
mat A=(2,2000,20,4500)
mat B=(4,2500,19,6000)
mat C=(3,2000,17,1500)
dprobit f rep we tr pr, nolog at(A)
dprobit f rep we tr pr, nolog at(B)
dprobit f rep we tr pr, nolog at(C)
mfx, at(A)
mfx, at(B)
mfx, at(C)
******************

The question is: Does the confusion caused by the command outweigh the
benefits? 


HTH
Martin


-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Sven Tengstam
Sent: Friday, October 31, 2008 11:21 AM
To: 'statalist@hsphsun2.harvard.edu'
Subject: st: RE: RE: marginal effects in probit



Thanks a lot! I read it and it is nice to see that more people than I have
been confused by this. But, I can't stop thinking that there must be a
reason that dprobit works like it does. If it was a mistake that mfx and
dprobit differ on this, the STATA people would have changed the command,
hadn't they? In the STATA manual it says that z correspond to the underlying
coefficient, but not why it is like this. I'm still puzzled, I'm afraid.

Best
Sven



-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Martin Weiss
Sent: den 31 oktober 2008 11:01
To: statalist@hsphsun2.harvard.edu
Subject: st: RE: marginal effects in probit

Line for the server

http://www.stata.com/statalist/archive/2008-10/msg00961.html



HTH
Martin

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Sven Tengstam
Sent: Friday, October 31, 2008 10:57 AM
To: 'statalist@hsphsun2.harvard.edu'
Subject: st: marginal effects in probit


Dear all,

I am new on this list and I have one question that I hope someone can help
me a little bit with. It is about the command "dprobit".

dprobit reports the Confidence Interval for the marginal effects, but "z and
P>|z| correspond to the test of the underlying coefficient being 0". Does
anyone know why Z is reported for the coefficient (instead of for the
marginal effect)? Could it be that when testing the null hypothesis for the
marginal effect, it is actually Z for the underlying coefficient that should
be used?

Thanks in advance!

Best,
Sven Tengstam, Sweden

*
*   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/

*
*   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/

*
*   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/

*
*   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