Feng:
1) The linear probability model was used in a time that maximum likelihood was too computationally burdensome. This is no longer an excuse for using the lpm, in a time when the average coffee maker has more computing power than a super computer of that long gone era.
2) The lpm should be a reasonable approximation of -probit- as long as the probabilities of success for given values of x1 and x2 remain between .2 and .8. Outside that range the linearity usually breaks down, and OLS is no longer is a good approximation of -probit- (or -logit-). Anyhow, why settle for an approximation when the real thing is easy to obtain?
HTH,
Maarten
-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands
visiting adress:
Buitenveldertselaan 3 (Metropolitan), room Z214
+31 20 5986715
http://home.fsw.vu.nl/m.buis/
-----------------------------------------
-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu]On Behalf Of Feng Liu
Sent: dinsdag 18 april 2006 16:13
To: statalist@hsphsun2.harvard.edu
Subject: st: compute elasticity: Probit VS LPM
I run a simple model: y = x1 + x2. Both y and x1 are dummy variables and
they are highly positively correlated. x2 is a continuous variable. I use
command mfx, eyex to estimate the elasticity of x2. I tried both Probit and
OLS. However, the elasticities from them are much different. The one from
Probit is -1.27 and the one from OLS is -0.27. I wonder what might be the
problem.
*
* 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/