Using SEM for regression to take advantage of MLMV, all models appear to be saturated?

[Jo Cotton]

I am using SEM to run some multiple regressions to take advantage of MLMV/FIML. All of my models have this line at the bottom:
LR test of model vs. saturated: chi2(0) = 0.00, Prob > chi2 = .

I noticed in online examples (slides done by Chuck Huber; http://www.cair.org/wp-content/uploa...EMWorkshop.pdf) and in the user guide, that when just a regression was run under SEM this same message indicating a saturated model also appeared. Is this meaningful, or does it just appear because I am using SEM to do a simple regression (only one dependent variable, no measurement component)?

[Richard Williams]

A simple regression would be a fully saturated model, so the message is correct.

Incidentally, one of my suggestions to StataCorp is that mlmv be added as an option to commands like regress so you don't need to use sem.

[Jo Cotton]

Thanks Richard - and good suggestion!