Statalist


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

Re: st: growth curve model with weights


From   Stas Kolenikov <skolenik@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: growth curve model with weights
Date   Wed, 17 Jun 2009 12:24:34 -0500

Look up chapter 3 of GLLAMM manual. Section 3.2 describes the general
framework for two-level random coefficient model, of which the growth
model is a special case. You will have two latent factors at student
level, one will be loading on the vector of 1's (the intercept), and
the other one on time (the growth rate; the wave number in the simple
case if they are equidistant in time... NJC, is there a proper long
English word for that?). Now, you would need to add the third level,
the school, on top of that, and I probably won't bother with random
coefficients there, just the random intercept. Alternatively, you
could move the whole random slopes structure to the school level.
Finally you can have both student level and school level variability
in both intercept and slope -- in -gllamm- terms, you will have four
random variables set up in two-by-two structure with levels (student
and school) and role (intercept and slope). Start with

gen ones = 1
* assume wave takes values 1 through 4
eq intrc: ones
eq slope: wave

gllamm , i(student school) nrf(2, 2) eqs(intrc slope intrc slope) trace noest

to get the model structure. Then you would need to constrain the
coefficients (which will likely be [f1]ones, [f2]wave, [f3]ones and
[f4]wave) to 1 with -constraint- command, and then add the constraints
to your -gllamm- call with -constraints()- option.

Now, the weights are sort of orthogonal to anything in the model
structure, so you can specify them at any point. Make sure that the
hierarchy of weights goes along with the hierarchy of the data. You
will probably have something like

gen wt1 = 1
gen wt2 = [student level weight]
gen wt3 = [school level weight]

gllamm ... , ... pweight(wt)

On 6/17/09, P C <calicep@yahoo.com> wrote:
> Thanks to Maarten and Stas. I already had these sources. But they do not have specific examples how I should handle growth curve model. I wish someone who has done that can share some experience. Thanks.
>
>  --- On Tue, 6/16/09, Stas Kolenikov <skolenik@gmail.com> wrote:
>
>  > From: Stas Kolenikov <skolenik@gmail.com>
>  > Subject: Re: st: growth curve model with weights
>  > To: statalist@hsphsun2.harvard.edu
>  > Date: Tuesday, June 16, 2009, 1:43 PM
>
> > See also: http://www.citeulike.org/user/ctacmo/article/850244
>  >
>  > On 6/16/09, Maarten buis <maartenbuis@yahoo.co.uk>
>  > wrote:
>  > >
>  > >  --- On Tue, 16/6/09, Alice wrote:
>  > >  > I have a question about estimating growth
>  > curve model with
>  > >  > survey methods in Stata.
>  > >  >
>  > >  > The data are logitudinal from Wave I to
>  > Wave IV.
>  > >  >
>  > >  > The sample is stratified random sample.
>  > Region is the
>  > >  > strata, school is the primary sampling unit
>  > with weights for
>  > >  > each individual within each wave.
>  > >  >
>  > >  > How should I estimate a growth curve model
>  > in Stata? Anyone
>  > >  > has examples for codings? Or suggestions
>  > for literature that
>  > >  > covers this topic?
>  > >
>  > >
>  > > You can use pweights in -gllamm-. You can read more
>  > about that
>  > >  here:
>  > >  http://www.gllamm.org/
>  > >
>  > >  -- Maarten
>  > >
>  > >  -----------------------------------------
>  > >  Maarten L. Buis
>  > >  Institut fuer Soziologie
>  > >  Universitaet Tuebingen
>  > >  Wilhelmstrasse 36
>  > >  72074 Tuebingen
>  > >  Germany
>  > >
>  > >  http://home.fsw.vu.nl/m.buis/
>  > >  -----------------------------------------
>  > >
>  > >
>  > >
>  > >
>  > >
>  > >  *
>  > >  *   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/
>  > >
>  >
>  >
>  > --
>  > 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/
>  >
>
>
>
>
>  *
>  *   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/
>


-- 
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