LGMM with nested data PreviousNext
Mplus Discussion > Growth Modeling of Longitudinal Data >
Message/Author
 Anthony Mancini posted on Thursday, January 22, 2009 - 8:50 pm
I am conducting an LGMM with a longitudinal dataset that includes both individuals (N = 50) and siblings (N = 225) within families (about 75).

My analysis is focused on the individual level, but I am obviously concerned about the non-independence of some of the data.

Is it required that I use a multi-level LGM model here?

Any help is greatly appreciated.
 Linda K. Muthen posted on Friday, January 23, 2009 - 4:52 pm
See the introduction to Chapter 9 of the Mplus User's Guide where two approaches for non-independence of observations are discussed.
 Anthony Mancini posted on Friday, January 23, 2009 - 8:10 pm
Okay, thanks very much!
 Eric Deemer posted on Saturday, June 06, 2015 - 11:22 am
Hello,
I'd like to fit a multilevel growth model with 3 observations nested within students. However, the students are also nested within classrooms so to account for this type of nonindependence would I mention this grouping variable in the CLUSTER command? Thank you.

Eric
 Linda K. Muthen posted on Saturday, June 06, 2015 - 12:07 pm
Yes.
 Eric Deemer posted on Saturday, June 06, 2015 - 12:16 pm
Okay, thanks Linda. Would I need to use TYPE=COMPLEX because there are 2 cluster variables or would TYPE=TWOLEVEL be sufficient?
 Linda K. Muthen posted on Saturday, June 06, 2015 - 4:45 pm
If your data are in wide format, you need just one cluster variable. I would recommend having the repeated measures be three variables in wide format. It allows more flexible modeling.
 Eric Deemer posted on Saturday, June 06, 2015 - 5:32 pm
Thanks, Linda.
 Eric Deemer posted on Saturday, June 06, 2015 - 8:23 pm
Hi Linda,
I followed your suggestion but I got an error message. Is it possible to fit a multilevel growth model using data in wide format?

cluster = class;
MODEL:
i s | ai1@0 ai2@1 ai3@2;

*** ERROR in VARIABLE command
The CLUSTER option can only be used for TYPE=COMPLEX, TYPE=TWOLEVEL, or
for the CLUSTER_MEAN function in the DEFINE command.

Eric
 Eric Deemer posted on Saturday, June 06, 2015 - 9:16 pm
Disregard my last question, I figured it out. I specified type=twolevel with no between-level variance and it worked with class as cluster variable.

Thanks,
Eric
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: