Statalist


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

Re: st: gllamm


From   Stas Kolenikov <skolenik@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: gllamm
Date   Mon, 20 Jul 2009 11:01:11 -0500

On Mon, Jul 20, 2009 at 9:16 AM, Arulampalam,
Wiji<Wiji.Arulampalam@warwick.ac.uk> wrote:
> Could someone please tell me whether the following model could be
> estimated in gllamm in stata?

The short answer is, "Yes, you can".

You did not ask for the long answer about "How do I do that?", but
here's a sketch: you would need to have four -eq- options for each of
the latent constructs to describe the observed measures of your latent
variables, and -bmatrix- that links the three exogenous variables to
the endogenous one. You would also need to put constraints on the
covariance structure of the latent variables: the exogenous ones
should be allowed to correlate, but their correlation with the error
term in the latent regression should be zero. If you have no
experience working with -gllamm-, that looks terrifying, but once you
spend three-four days trying this and that, it will gradually start
making sense.

What you describe however looks like a reasonably standard SEM model
that can be estimated by pretty much any package if your data are
normal/continuous. If you don't have any budget constaints, get Mplus;
if you do, get R and figure out -sem- package there. I've had
convergence issues with it, but generally you can make it work.

-- 
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.
*
*   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