Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

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

Re: st: Heckprob - Why no margins in selection equation?

From   Maarten buis <>
Subject   Re: st: Heckprob - Why no margins in selection equation?
Date   Tue, 30 Nov 2010 13:27:49 +0000 (GMT)

--- On Tue, 30/11/10, James McJoseph wrote:
> Could someone enlighten me plse. I use stata 11. I
> estimated -heckprob-. I requested -margins, dydx(_all)-.
> Stata gives me margins for the obejective equation
> only. For the selection equation, it gives "missing".
> Why?

A marginal effect is the first derivative of some 
outcome measure with respect to explanatory variables. 
In models like -heckprob- there are multiple outcome
measures posible: the marginal probability of success,
the probability of success given that someone is selected,
the probability of being selected, etc. So -margins- 
needs to know what outcome measure you mean. By default,
it looks at the predict option of your command, in this
case -heckprob-, and looks at the default prediction and
assumes that that is what you want. You can (and should) 
find out yourself what that default is by typing in Stata
-help heckprob postestimation-. Then you will find that 
the default prediction of -heckprob- is the marginal 
probability of succes, so it makes no sense to have that
marginal effect type for the selection equation and Stata
wisely keeps them missing. You can change your outcome
measure by specifying the -predict- option in -margins-.

Hope this helps,

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


*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index