Re: st: gllamm syntax

[Debs Majumdar <debs_stata@yahoo.com>]

Thanks Mary. You cleared a lot of my doubts with the gllamm syntax.

-Joey




----- Original Message ----
From: Mary E. Mackesy-Amiti <mmamiti@uic.edu>
To: statalist@hsphsun2.harvard.edu
Sent: Wed, November 3, 2010 9:13:37 AM
Subject: Re: st: gllamm syntax

On 11/1/2010 5:14 PM, Shubhabrata Mukherjee wrote:
> Hi,
> 
>     I am trying to run gllamm for a 3-level nested model (patients nested 
>within
> doctors within clinics). I have 4 patient level covariates (gender, age, race
> and education)&  1 doctor level covariate (specialty). I am not sure about the
> correct syntax for running a random coefficient model. Can anyone take a look
> and point me to the right syntax. Here's what I used which is wrong.
> 
> ***********
> 
> gen byte one = 1
> eq cons : one
> eq gen  : new_gen
> eq race : new_race
> eq educ : edulvl
> eq age  : age1
> 
> gllamm zeta1 new_gen age1 edulvl new_race specialty, i(docid clinicid) nrf(1 
8)
> ///
> eqs(cons cons new_gen cons new_race cons edulvl cons age1)
> 
> *****
> I get the following error message:
> 
> equation new_gen not found
> r(111);
> 
> *****

1.  equation new_gen is not found because the equation name is "gen"

2.  As I understand it,

gllamm zeta1 new_gen age1 edulvl new_race specialty, i(docid clinicid) nrf(1 5) 
///
    eqs(cons cons gen race educ age)

would specify a 3-level model with random intercept at the doc level and random 
intercept & random slopes on gender, race, education and age at the clinic level 
(is this is what you want?)

> If I want to enter a random effect (intercept and slope) for doctor specialty,
> what will be the modification. Also is there a way to make it run faster using
> the `nip' command?
> 
> 

eq cons : one
eq spec : specialty
gllamm zeta1 new_gen age1 edulvl new_race specialty, i(docid clinicid) nrf(2 1) 
///
    eqs(cons spec cons)

would specify a model with fixed effects for patient variables, random intercept 
and random slope on specialty at the doc level, and random intercept at the 
clinic level.

3.  I believe nip() is used to increase the number of integration points for 
better precision

4.  equivalently,

xtmixed zeta1 new_gen age1 edulvl new_race specialty || clinicid: || docid: 
specialty




-- Mary Ellen Mackesy-Amiti, Ph.D.
Research Assistant Professor
Community Outreach Intervention Projects (COIP)
School of Public Health m/c 923
Division of Epidemiology and Biostatistics
University of Illinois at Chicago
ph. 312-355-4892
fax: 312-996-1450

*
*   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/



      
*
*   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/