Re: st: Re:multilevel model and gllamm

 From Stas Kolenikov To statalist@hsphsun2.harvard.edu Subject Re: st: Re:multilevel model and gllamm Date Wed, 22 Sep 2004 09:08:09 -0400

```I see now. I first understood that both varx and vary are the
measurement variables. Then your setup looks correct. There should
also be a way of making varz and varw regressors for the latent
variable, but whether you want to do that or not depends on your
interpretation on whether they affect the random coeffcients, or
directly the outcome.

Then also the unit correlations would indicate that you are making up
a model that is too complicated. All higher constants require lower
slopes means that all of your lines, i.e. the curves across all
individuals, are passing through the same point in varx / vary plane.
Does this seem plausible to you? Does it make sense in terms of your
problem? This is also where the model would become computationally
truciky and unstable, I suppose: the likelihood is going to be flat in
some directions, and the maximizer will stumble over it. Also, Stata's
-ml- does not like problems on the boundary when it cannot step
further than 1 for correlation coefficients. Did you receive any lack
of convergence diagnostic messages along with unit correlations?

Stas

On Wed, 22 Sep 2004 07:16:00 -0500, Jeffrey Simons <jsimons@usd.edu> wrote:
> >> Here is an example of my commands, varx  and vary are the repeated measures
> >> the remaining variables are level 2 predictors:
> >>
> >> Gen cons=1
> >> Eq cons: cons
> >> Eq slope: varx
> >>
> >> gllamm vary varx varz varw ,i(id) family(gamma) nrf(2) eqs(cons slope)
> >
> > I would tend to think that -gllamm- would take it that -vary- depends
> > on -varx varz varw-, as it only takes one dependent variable. You
> > would need to take your data to the long form by -reshape-, and then
> > code individual variables by dummies. Then your -s()- option would
> > give the name of those dummies so that you can have different
> > variances for different measures. See Section 4.1 of the manual for
> > similar treatment.
> >
> > This also explains your -1 correlations: you have -varx- both as an
> > explanatory variable for -vary-, and as a slope for random
> > coeffcients. That's kind of weird for -gllamm-.
>
> This is confusing to me. My data are in long form. My command seems to be in
> the same form as that given in the help file:
>
>   . eq idc: cons
>        . eq idt: time
>        . gllamm resp time, link(logit) fam(binom) denom(five) /*
>           */ i(id) nrf(2) eqs(idc idt) ip(g) nip(6) trace
>
> In this I thought the variable time (like my varx) was supposed to predict
> resp and it was expected to be a random slope. My varx doesn't include time
> per se but rather levels of another variable measured at successive
> measurement occasions. To be more specific, my response variable (vary) is
> number of drinks in the past 30 minutes, my varx is a time lagged level of
> negative affect measure. These are repeated measures (e.g., 50-100
> measurement occasions over a couple weeks). Then varz is gender and varw a
> trait measure and these would be level 2 predictors.
>
> So, what I wanted to examine is whether affect at t-1 is associated with
> with drinking rates at t1 and whether this association varied across
> individual, which I thought would be seen in the random slope and then
> examined by looking at interactions between the level1 and level 2
> predictors.
>
> My data are set up in long form with each row being a single measurement
> occasion.
>
> Thoughts?
>
> Jeffrey simons
>
> > ------------------------------
> >
> > Date: Tue, 21 Sep 2004 10:13:48 -0500
> > From: Fred Wolfe <fwolfe@arthritis-research.org>
> > Subject: Re: st: Multilevel analysis and GLLAMM
> >
> >>
> >> HLM (or M-plus) are more specific, and thus faster. With -gllamm-, you
> >> can use all Stata tricks for data management, testing, etc.
> >
> >
> > HLM Version 6 which is supposed to be released this month will import Stata
> > files.
>
> Thanks for the information.
>
> >
> >
> >
> > Fred Wolfe
> > National Data Bank for Rheumatic Diseases
> > Wichita, Kansas
> > Tel (316) 263-2125     Fax (316) 263-0761
> > fwolfe@arthritis-research.org
> >
> >
> > *
> > *   For searches and help try:
> > *   http://www.stata.com/support/faqs/res/findit.html
> > *   http://www.stata.com/support/statalist/faq
> > *   http://www.ats.ucla.edu/stat/stata/
> >
> > ------------------------------
>
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>

--
Stas Kolenikov
http://stas.kolenikov.name
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```