Peng Liu posted on Friday, June 06, 2014 - 12:02 pm
I'm trying to identify latent classes in longitudinal data. The outcome variable is continuous. I'm wondering if I don't specify growth parameters (e.g. linear or quadratic) in the model statement, or just run the following code without giving a model statement, how should I interpret the mplus output. Is mplus fitting a mixture of multivariate normal with 'unstructured' covariance in this case?
DATA: FILE IS "C:\....\mydata.dat"; FORMAT IS FREE; VARIABLE: NAMES ARE ID T1-T5; MISSING ARE .; USEVAR are T1-T5; CLASSES = c (3); ANALYSIS: TYPE = MIXTURE; STARTS = 100 10; STITERATIONS = 10; OUTPUT: TECH11;