I have am trying to analyze clustered data in STATA. I have experimented with 2 methods: xtmixed and svyset. However, I have a few questions.
1. I would expect that the results wouldn't be very different, and if they were, I would expect to have larger standard errors using the svy command and not xtmixed. But I get the opposite result - larger standard errors using xtmixed. Why is this the case?
2. Given the difference in standard errors, the p-values are different between the two analyses. Some variables that are significant in svy:regress are not significant with xtmixed. Which analysis is "more reliable"?
3. I am using xtmixed to analyze a 3-level model (child student district). However, the variance estimates at levels 2 and 3 are almost identical. I would assume this means that I don't need a multilevel model and would be safe in using svyset.
4. Finally, with both xtmixed and svy: regress, I get some standard errors larger than 1. This seems unusual.
Thank you.
*
* 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/
