Lu Gram posted on Saturday, April 02, 2016 - 2:49 am
If the latent variable and all observed variables are continuous is there any difference in the underlying mathematical model between the models:
y ON factor; i1 i2 i3 i4 i5 BY factor;
and the model:
y i1 i2 i3 i4 i5 BY factor;
In either case it reduces to an overall multivariate normal model no?
The reason I am asking is because in Stata, there is no difference between BY and ON as you only specify the probability distribution of a link (normal, binomial, etc.), not whether it is a structural or a measurement model link. But maybe Mplus performs some kind of rotation?