st: randon coefficients and level 2 variables in GLLAMM

[Paul Norris <P.Norris@ed.ac.uk>]

Dear all,

Apologises if the answer to this question should be obvious but I'm new
to using GLLAMM and having checked the documentation and the Statalist
archive I still not sure how to progress.

I'm running a 2 level ordinal logit model in GLLAMM.  Along the lines of
the following command:

gllamm y x1 x2, i(area) nrf (2) eqs(inter slope) link (ologit) adapt

where y is the (ordinal) dependent variable, x1 a covariate at level 1,
and x2 a covariate at level 2.  The coefficient associated with x1 is
allowed to vary across level 2 units (slope having been created using eq
slope:x1).

Now GLLAMM displays the direct impact of x2 on y (which I am interested
in) and the covariance between the random coefficients for the intercept
(inter) and x1 (slope).

In addition, I am interested in understanding if there is a relationship
between the variation in coefficients for x1 and the level 2 variable x2
i.e. do areas with a higher score on the level 2 variable have higher
coefficients associated with the variable x1. Is it possible to estimate
this relationship using GLLAMM? If so, is it done within the initial
model, or through storing and then modelling the random coefficients
after the model is estimated.

Kind regards,

Paul
--
Paul Norris
Lecturer in Social Policy
The University of Edinburgh
School of Social and Political Science
Chrystal Macmillan Building
15a George Square
Edinburgh
EH8 9LD

Tel: +44 (0) 131 6503922
Fax: +44 (0) 131 6503945
P.Norris@ed.ac.uk

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

*
*   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/