Latent State-Trait Analysis PreviousNext
Mplus Discussion > Latent Variable Mixture Modeling >
Message/Author
 Vanessa Jaensch posted on Thursday, April 17, 2014 - 6:33 am
Hello, I am using latent state-trait approach and longitudinal design with 3 measurement points.
The idea is to test whether the trait component will increase the predictive ability beyond that provided by the score of the construct measured at T3.
Model 1 contains regression: SWL_T3 ON VI_T3
(VI_T3 is VI measured at T3) – result: VI_T3 is a sig predictor

Model 2 contains measurement invariance restriction (invariant factor loadings, intercepts, items’ unique variances), LST-modelling and the regression: SWL_T3 ON VI
(VI = trait component of VI measured at T1, T2, and T3) – result: VI is a sig predictor

Model 3: Model 1 + Model 2 - two regressions: SWL_T3 ON VI and SWL_T3 ON VI_T3;
If I use MLR I get the following error:
MAXIMUM LOG-LIKELIHOOD VALUE FOR THE UNRESTRICTED (H1) MODEL IS -8523.947
THE STANDARD ERRORS FOR H1 ESTIMATED SAMPLE STATISTICS COULD NOT BE COMPUTED. THIS MAY BE DUE TO LOW COVARIANCE COVERAGE.
THE ROBUST CHI-SQUARE COULD NOT BE COMPUTED.
THE STANDARD ERRORS FOR THE STANDARDIZED COEFFICIENT COULD NOT BE COMPUTED.

If I use ML I get the following error:
MAXIMUM LOG-LIKELIHOOD VALUE FOR THE UNRESTRICTED (H1) MODEL IS -8523.947
THE STANDARD ERRORS FOR THE STANDARDIZED COEFFICIENT COULD NOT BE COMPUTED.

I conducted the same analysis using samples with and without missing data. That makes no difference.

Thanks,
Vanessa
 Bengt O. Muthen posted on Friday, April 18, 2014 - 2:38 am
It sounds like you have low covariance coverage, that is, some pairs of your variables are observed in less than 10% of the sample. You should check why that is so low.
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: