I'm running the same data for SEM but getting two different results

[Charmain Farrar]

I have given my data set to my professor, and when he runs the SEM, he gets constraints with values, whereas when I run the same data,

I get constraints that are 0. What am I missing?

[Joseph Coveney]

I have given my data set to my professor, and when he runs the SEM, he gets constraints with values, whereas when I run the same data, I get constraints that are 0. What am I missing?

Do you mean constants that are zero? I'm not sure which is yours and which your professor's, but on the left-hand side, the constants are all free to be estimated; on the right-hand side, the constants are all constrained to be zero, and so that's why they are all zero.

If you mean constraints, well the left-hand side doesn't impose any other than for one factor loading per latent variable for identification. It's equivalent to (names of latent factors abbreviated)

import excel "Copy of CSWS - ERBTS_May 25, 2024.xlsx", sheet(Sheet0) firstrow clear

rename *, lower

sem ///
    (ab ad ae aq ar <- I) ///
    (az ba bc bd be <- A) ///
    (bf bg bi bk bl <- M), ///
        method(mlmv)

And the one on the right-hand side is equivalent to

sem ///
    (ab ad ae aq ar <- I, noconstant) ///
    (az ba bc bd be <- A, noconstant) ///
    (bf bg bi bk bl <- M, noconstant), ///
        method(mlmv) covstructure(_LEx, diagonal)

By the way, it's not clear why the left-hand side specifies covstructure(_lexogenous, diagonal), but then goes right ahead and specifies all of the latent factors' covariances anyway.