|
|
 |
| Growth Mixture Model with Known Classes |
 
|
| Message/Author |
|
|
|
|
I'm trying to use growth mixture modeling with the known classes command. I can get my syntax to run when following Example 8.8 from the Mplus User Guide. However, it seems like I can't get around the Knownclass variable being completed crossed with the classes command. Specifically, my 'knownclass' is group (dual language learners or monolingual English speakers). My previous analyses indicate that there are four profile groups for the dual language learners and two profile groups for the monolingual English speakers. When using the following syntax I end up with 8 profile groups (or classes):
classes = cg (2) c (4);
KNOWNCLASS = cg (LANGHOME = 1 LANGHOME = 0);
Is there a work around that will allow me to specifically specify 4 profile groups for the dual language learners and 2 profile groups for the monolinguals? I've tried fixing the intercept and slope of the 7th and 8th classes to 0, but it hasn't helped.
Thank you in advance for your valuable time.
Best regards,
Matt |
|
|
|
|
Use the approach of
%cg#x.c#y%
for classes x and y where you apply parameter equalities across latent classes for the monolingual group so that it effectively has only 2 latent classes. |
|
|
|
|
I have three continuous variables for scaffolding. After looking at results, I am left with two classes being the best option. I want to see if use of a fourth variable(also continuous, called MMS) will predict class membership. Fro the manual and threads I have read here, I think KNOWNCLASS would be helpful but I most often see those groups identified as categorical. Is there a way to use continuous variables?
data:
file = scaffolding.dat
variable:
names = MMS S1 S2 S3;
usevariables = a1 a2 a3;
classes = g(2) c(2);
knownclass = g (MMS ??????);
analysis:
type = mixture;
model:
%overall%
c on g; |
|
|
|
|
| Use the R3STEP option with MMS. |
|
|
| Back to top |
|
|
