Latent interaction needs estimation o... PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 Anonymous posted on Sunday, June 19, 2005 - 1:59 pm
hi Prof. Muthen

I have a question about the latent interaction model estimated by Mplus.

The literature on latent interaction always mentioned the necessarsity of estimation of latent means for the latent interaction term.

Consider Mplus example 5.13:
E(f3) =E(intercept+gamma1*f1+gamma2*f2+gamma3*(f1xf2)+error)
(assume E(f1)=E(f2)=0 for identification),
still,
E(f3)= intercept+gamma3*E(f1xf2)
= intercept+gamma3*cov(f1,f2)
if we want E(f3)=0, then
intercept has to be equal to -1*gamma3*cov(f1,f2).
Is this what Mplus doing?

Also, cov(f1xf2,f1) and cov(f1xf2,f2) --- are
they fixed to be zero all the time internally
(as they should be under normality for f1 and f2)??

Thanks a lot for your help.
 BMuthen posted on Monday, June 20, 2005 - 8:17 am
The means of the factors don't have to be restricted to zero when as in Mplus you estimate measurement intercepts (nu's). The covariances that you mention are not parameters that are estimated.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: