|
|
 |
| Multi-mixed Factor: Classes of Within... |
 
|
| Message/Author |
|
|
|
|
Hi,
I am interested in modeling both within and between factor structure of daily diary data. I found this was very easy to do within the context of multilevel modeling in MPlus. However, I am interested in using mixture models to try to identify subgroups who show similar within-person factor structure.
I tried the following:
TITLE:
Multilevel Mixture Factor Model;
DATA:
FILE = "data.dat";
VARIABLE:
NAMES = ID P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 P16 P17 P18 P19 P20;
MISSING=.;
CLASSES = c (2);
CLUSTER = ID;
ANALYSIS:
TYPE = TWOLEVEL MIXTURE;
STARTS=100
MODEL:
%WITHIN%
%OVERALL%
fw BY P1-P20;
%c#1%
fw BY P1-P20;
%c#2%
fw BY P1-P20;
%BETWEEN%
%OVERALL%
fb BY P1-P20;
OUTPUT:
TECH1 TECH8;
However, it looks like this model allows classes to vary across measures of the within persons. Is there a way to constrain the classes to contain individuals while to still model within-person variation?
Thanks! |
|
|
|
|
| It sounds like you are interested in a "cb" latent class variable, that is, a variable put on the Between= list. See the UG for examples. |
|
|
| Back to top |
|
|
