multilevel regression analysis via stata !?
I have a question regarding multilevel analyses using stata. My research question concerns a nested observational study design in which i collect questionnaire data from students (that cluster within classes, that cluster within schools: a 3 level design). Together with a set of confounders (SES-score, gender, age and school level) I wish to study both a couple of binary outcomes and a couple of continuous ones. Now I'm not sure what the proper commands in stata will have to be in order to execute this properly, so that I will get the B's/OR's for my regression analyses. Also, I'm not sure how I can take into account the problem of multiple testing in this setup, for example is it possible to manually adjust he a-level for the performed regressions? in that case I would simply do the Bonferroni a/k=a-adj by hand and be done with the multiple testing problem. Is there anyone that can help me with this?? Thank you so much!!
Vincent, (a health scientist who now wished he was a statistician....)
Re: multilevel regression analysis via stata !?
Your 3 nested levels shows that your data are good candidate for Mixed Model Design in which you can find variations between/within students, classes, schools so these variations are not attributed to main variables in the study.