Common Latent Factor

[ecelik]

I'm trying to solve three Common Latent Factor (CLF) models that were previously estimated in Mplus using Jamovi.

Here are the models:

# Model A – CLF with no constraints
CLF =~ cr1 + cr2 + cr3 + cr4 + f1 + f2 + f3 + f4 + pp1 + pp2 + pp3 + pp4
# Measurement Model
CR =~ cr1 + cr2 + cr3 + cr4
FB =~ f1 + f2 + f3 + f4
PP =~ pp1 + pp2 + pp3 + pp4
# Orthogonal factors
CLF ~~ 0*CR
CLF ~~ 0*FB
CLF ~~ 0*PP

# Model B – CLF with constraints all loadings fixed to 0
CLF =~ 0*cr1 + 0*cr2 + 0*cr3 + 0*cr4 + 0*f1 + 0*f2 + 0*f3 + 0*f4 + 0*pp1 + 0*pp2 + 0*pp3 + 0*pp4
# Measurement Model
CR =~ cr1 + cr2 + cr3 + cr4
FB =~ f1 + f2 + f3 + f4
PP =~ pp1 + pp2 + pp3 + pp4
# Orthogonal factors
CLF ~~ 0*CR
CLF ~~ 0*FB
CLF ~~ 0*PP

# Model C – CLF with constraints all loadings fixed to 1
CLF =~ 1*cr1 + 1*cr2 + 1*cr3 + 1*cr4 + 1*f1 + 1*f2 + 1*f3 + 1*f4 + 1*pp1 + 1*pp2 + 1*pp3 + 1*pp4
# Measurement Model
CR =~ cr1 + cr2 + cr3 + cr4
FB =~ f1 + f2 + f3 + f4
PP =~ pp1 + pp2 + pp3 + pp4
# Orthogonal factors
CLF ~~ 0*CR
CLF ~~ 0*FB
CLF ~~ 0*PP

In Mplus, I get consistent test statistics and parameter estimates for these models.

In Jamovi, Model B and Model C produce very different test statistics and parameter estimates compared to Mplus.

Are the constraints for Model B and Model B correct in Jamovi?

What are your opinions and suggestions on making the Jamovi solution compatible with Mplus for CLF models?

I appreciate any help you can provide.

[mcfanda@gmail.com]

Hi,
comparisons with other software can be done by setting exactly the same options in each software. Semlj, when set up exactly, gives the same results as lavaan R. It is difficult to know what setup (defaults and options) you use in Mplus.

For the constraints in Semlj, just check the output to be sure that the coefficients in the tables correspond to the constraints you set up. If they do, they are fine.