Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

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

Re: st: marginal effects after clogit

From   Steve Samuels <>
Subject   Re: st: marginal effects after clogit
Date   Mon, 30 Aug 2010 12:54:22 -0400

Olivia Bacha-

I don't understand your question. The following code generates no
intercepts (_cons). I assume that "AVE" in your post is a typo for
"AME", or "average marginal effect".
webuse birthwt2, clear
clogit low lwt smoke ptd ht ui i.race, group(pairid)
margins, dydx(*) predict(xb)

Perhaps by "intercepts" you meant the estimated margins. In this
example, the estimated are identical to the original -clogit-
coefficients and so have the same interpretation. If I had specified
"predict(pu0)", the estimated margins would be first derivatives of
probabilities in which the intercepts in the logistic equation  are
set to zero. Such  AME's, like the "pu0" predictions themselves, can
be very misleading if the true intercepts are far away from zero.

If this does not answer your question, please clarify and include the
code and output which puzzle you.


Steven J. Samuels
18 Cantine's Island
Saugerties NY 12477
Voice: 845-246-0774
Fax:    206-202-4783

 Olivia Bacha <> wrote:

> How can I interpret AVE for intercepts generated by margins, dydx(*)
> predict(xb) after clogit?

*   For searches and help try:

© Copyright 1996–2016 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index