

Dual change score MLM not converging 

Message/Author 

Fadel Matta posted on Tuesday, November 27, 2018  6:09 pm



I’m modeling a construct using dual change score modeling (a class of latent change score modeling), but I get the following, perhaps somewhat common, warning: WARNING: THE LATENT VARIABLE COVARIANCE MATRIX (PSI) IS NOT POSITIVE DEFINITE. THIS COULD INDICATE A NEGATIVE VARIANCE/RESIDUAL VARIANCE FOR A LATENT VARIABLE, A CORRELATION GREATER OR EQUAL TO ONE BETWEEN TWO LATENT VARIABLES, OR A LINEAR DEPENDENCY AMONG MORE THAN TWO LATENT VARIABLES. CHECK THE TECH4 OUTPUT FOR MORE INFORMATION. The variance for my intercept estimate (LCOH21) is indeed negative. If I constrain this estimate to be zero, the model converges but the fit statistics are notably bad (e.g., CFI = 0). The identical model for almost all of the other constructs in the dataset converge without any issues. Is it just that this model for this particular construct is inappropriate for my data, or is there something else I could do to overcome this nonconvergence/terrible fit? If it would help, I can email my syntax, data, and license. Thanks! 


I am not familiar with this model. Perhaps you can check previous answers here by Kevin Grimm. Or SEMNET. 

Back to top 

