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

Re: st: Using xtgee with multiple weights per ID

From   Steven Samuels <>
Subject   Re: st: Using xtgee with multiple weights per ID
Date   Fri, 5 Dec 2008 09:29:22 -0500

-xtgee- is obviously the wrong tool. Try -gllamm-. Time points will be "level 1" units and you can specify -pweights- for them.


On Nov 26, 2008, at 3:02 PM, Ariel Linden, DrPH wrote:
I am trying to run a weighted xtgee using different weights per each
person/period. This model is based on the work of Hernan et al., (estimating the causal effect of zidovudine on CD4 count with a marginal structural
model for repeated measures, stats in med 2002). Conceptually each
person/period will have a weight based on the propensity of getting the treatment at that office visit, conditional on time varying covariates from
past visits.

However, when I run the xtgee using the weights, I get an error message stating "weight must be constant within ID". In addition, the model is tossing out the interaction variable for "change: treatment * time because of colinearity. This doesn't get tossed out when I use an unweighted model.
*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index