st: problems with prediction using gllapred and ci marg mu

["Joe Zickafoose" <josephzi@med.umich.edu>]

Hello,

We are attempting to use gllamm for a 2-level logistic analysis on weighted survey data and then to use gllapred and ci_marg_mu to estimate adjusted probabilities with confidence intervals.  States are the level 2 clusters, and we are interested in estimating the predicted probability by state controlling for level 1 (individual) characteristics.  We are doing this by appending data with "average" individual characteristics for each state with the outcome variable missing, and then trying to use gllapred for the original data set plus the appended data.  However, when we try to use gllapred we get an error message (below).  Oddly, ci_marg_mu is more than happy to give us confidence intervals for the appended data even though gllapred doesn't give a point estimate.
 
The commands we are using are below.  We appreciate any suggestions you can provide or suggestions for someone who might be able to help us.
 
Thank you.
 
 
 
xi: gllamm mh_comp `individual characteristics' if misscount1==0, i(state) link(logit) family(binom) nip(10) pweight(wt)
 
append "average state data for prediction"
 
gllapred mh_comp_method2 if misscount1==0 & inspubpriv!=., marg mu fsample
 
log-likelihood cannot be computed
for observations that should not contribute to posterior or ll,
   set response variable to missing


**********************************************************
Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues 


*
*   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/