st: Mixed models with analytic weights

[Nathaniel Schwartz <nschwart@umich.edu>]

Hello,

1) I have been running regressions in stata to look at the effect of a
program treatment on test scores. The treatment is at the district
level. However, my test scores are available only at the grade-school
level and I would like my results to be weighted at the student level.
Thus, my initial regressions look like this:

reg testscore program [aw=gradeenrollment]

2) Most of my regressions also include grade-district and grade-year
fixed effects and grade-school-level covariates. These regressions
look like this:

xi: reg testscore program covariates i.district*i.grade i.year*i.grade
[aw=gradeenrollment].

3) Now, I would like to run mixed model regressions to clearly
estimate district-level variance. I'd like to do this using either the
"xtmixed" or the "glamm" command. However, I have not been able to
figure out how to correctly set up analytic weights to make these
regressions work. xtmixed returns an error message saying that it does
not support weighting.

Do you have advice on how to set up my regressions so that I am
looking at schools nested within districts and weighting at the
student level?

Thank you very much,
Nate
*
*   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/