I would like to ask for your opinion on a planned data evaluation.
I have the following "setting: n=100 (75 for all 4 measurement time points) 4 measurement time points (1-t4) 2 latent factors (X, Y) per measurement time point X (measured by x1, x2 and x3) Y (measured by y1, y2 and y3)
Data come from a 1-5 likert scale (x1, x2, x3) and a 1-6 likert scale (y1, y2, y3).
I would like to collect and plot the mutual influences of the two latent variables (X and Y) over time (t1-t4).
I have already tried all steps for the RI-CLPM "multiple indicators" on (https://jeroendmulder.github.io/RI-CLPM/lavaan.html#Ext_3:_multiple_indicator) but consistently get very poor model fits, no significant AR and CL paths and always error messages such as.
- The variance-covariance matrix of the estimated parameters (vcov) does not appear to be positive definite!
- some estimated ov variances are negative
- covariance matrix of latent variables is not positive definite
What do you think? Do I have to move away from the RI-CLPM approach in general and consider another evaluation method?
I am new to SEMs and I am stuck at the moment.
Do you have any solutions? What could be the problem? The missing normal distribution, the small sample size? The different Likert scales? Other?
I would be very grateful for your estimations and tips! Thanks a lot in advance!
