 |
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: GLLAMM and weights
Actually, Ariel, after reading Pfefferman et al.(1998), quoted in
Rabe-Hesketh and Skrondal (2006), I now think that rescaling the
weights is a good idea. One note: -xtgls- (if that is what you mean
by "gls") does not take probability weights.
-Steve
Pfeffermann, D., Skinner, C. J., Holmes, D. J., Goldstein, H., &
Rasbash, J. (1998). Weighting for unequal selection probabilities in
multilevel models. Journal of the Royal Statistical Society: Series
B: Statistical Methodology, 60(1), 23-40.
Rabe-Hesketh, S. & Skrondal, A. (2006). Multilevel modelling of
complex survey data. Journal of the Royal Statistical Society: Series
A (Statistics in Society), 169(4), 805-827.
On Dec 14, 2008, at 11:45 AM, Ariel Linden, DrPH wrote:
Thank you Steve for your comments. I use Stata 10.1 and have the most
updated version of GLLAMM installed.
Indeed, the weights represent the inverse probability of treatment, so
additional manipulation of the weights are not necessary.
Similarly, "treatment" and "treatment * time" variables must be in
the model
since those are the primary covariates under study. So an incremental
approach to model building will not get me to where I want to end up.
The authors use GEE in SAS. When I try using xtgee in Stata (with the
weights) I get an error stating that the weights must be the same
for each
id across waves.
Interestingly enough, I can get this to work using regress, glm,
and gls. I
may have to rely on gls for longitudinal modeling if I can't get
this to
work using gee or gllamm.
Thanks again
Ariel
*
* 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/