HLM using Complex Survey Data PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Sung Tae Jang posted on Thursday, April 13, 2017 - 10:37 pm
Dear Dr. Muthen,

I am using a stratified cluster sample to conduct HLM. In the dataset, there are 1st level (student) weight, 2nd level (school) weight, 200 replicates for each weight, and PSE. If I want to calcuate correct standard error for estimates, how I can use or do I need to use the information of those weights and replicates when I use "TYPE-COMPLEX" command? Thank you.
 Tihomir Asparouhov posted on Friday, April 14, 2017 - 11:05 pm
Mplus doesn't support yet replicate weights for two-level, however I don't think you need that anyway. Usually replicate weights are given instead of PSU and strata information - you seem to have both.

Use type=complex twolevel; cluster= PSU school; starta= ;weight = ; bweight= ;

That should work.
 Sung Tae Jang posted on Monday, April 24, 2017 - 9:10 pm
Thank you so much, Dr. Asparouhov.
Could you also let me know what I need to use for bweight? Is it school-level weight?

In addition, do I need to use school ID or just type school in "cluster= PSU school"? I have PSU variable, but I am not sure what school means in this command.

Thank you again for your kind answer.
 Tihomir Asparouhov posted on Wednesday, April 26, 2017 - 9:06 am
Yes school level weight. The school will be the level that is modeled as level 2.
You can find more information here
http://statmodel.com/download/SurveyJSM1.pdf
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: