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: xtivreg2 or new Stata12 command sem?

From   Stas Kolenikov <>
Subject   Re: st: xtivreg2 or new Stata12 command sem?
Date   Mon, 23 Jan 2012 12:10:50 -0500

On Mon, Jan 23, 2012 at 11:07 AM, Dithmer, Jan
<> wrote:
> Thanks very much to Stas and Maarten!
> I will have a look at gllamm, which seems to be quite flexible.
> Maarten, the link you posted is very interesting, but does it also imply that I may use xtmixed, because in the example it does the same more easily?
> However, am I right with my intuition? I mean, I can estimate the direct effect of B on Y (equation 1), but with a single equation I miss the indirect effects, or am I missing here?
> Is there a way to use the sureg command, when I somehow account for the panel structure of the data?

-xtmixed- is not as flexible as -sem-. Some types of models, like
growth models with one response variable, can be estimated by both,
but the class of models to which both can be applied is quite limited.
What -gllamm- does is it relaxes some of the restrictions that
-xtmixed- imposes, essentially allowing interactions of variables and
coefficients that are fixed in -xtmixed-. You have a genuine
multivariate response if you start considering other endogenous
variables, as your interest appears to be, so I doubt you could tweak
-xtmixed- to your purposes, although I might be wrong on this.

Stas Kolenikov, also found at
Small print: I use this email account for mailing lists only.
*   For searches and help try:

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