Message/Author |
|
|
Hello, I want to test a mediation model with 3 time points, different variables at each time point (not a growth model or simplex model). At each time point students are nested in a different teacher. There are several teachers (more than 50) so most students have a different pattern of nesting across years. I read somewhere (on this discussion board) that this might be called "crossed random effects" and that MPLUS currently does not handle this. Is that right? Or perhaps I am wrong that these are "crossed random effects" - Is there a way in MPLUS to remove the "teacher variance" at each occasion when the patterns of nesting change in this way? Do I need to remove it? Thanks! |
|
|
It sounds like crossed random effects, which Mplus currently doesn't handle. |
|
|
Thank you. Will future versions of MPLUS include this capability? Or do you otherwise have any suggestions how to deal with this? Thanks again. |
|
|
We plan to add this eventually. I am not sure when. There is no way to deal with this in Mplus. |
|
|
What if I took the teacher trajectories and created a single variable out of it? For example: A person who has TEACHER A - YEAR1 TEACHER B - YEAR2 TEACHER C - YEAR3 GETS A VALUE ON THE (NOMINAL) NEW VARIABLE OF "1" A person who has TEACHER C - YEAR1 TEACHER A - YEAR2 TEACHER D - YEAR3 GETS A VALUE ON THE (NOMINAL) NEW VARIABLE OF "2" etc. Would this type of categorization work? |
|
|
It sounds like you then treat teacher combinations as fixed effects, not random. |
|
Back to top |