 Estimated means and variance for LGC ...    Message/Author  James Martin posted on Sunday, November 21, 2010 - 7:40 pm
Hello,

I have a question about calculating the values for the factor variance provided by Tech4. If I had a simple model such as:

i s | y1@0 y2@1 y3@2 y4@3;
i s ON x;

Using the Mplus short course notes (from 5/17/10; slide 96), I can see the equations for calculating the estimated estimated means using covariates, but not the estimated conditional variance. So, the estimated mean for the intercept factor would would just be:

E(eta_0) = intercept + gamma*E(x)

After searching the message boards, looking through the technical appendices, and the short course notes, I can't seem to find the (scalar) formula for the variance of eta_0 once covariates are added. I realize this can be done rather straightforwardly using the TECH4 option, but was wondering if it could be hand calculated from different pieces of the output Mplus provides.

Thanks!  Linda K. Muthen posted on Monday, November 22, 2010 - 8:23 am
V(eta_0) = gamma squared*var(x) + residual variance (eta_0)

where the variance of x is obtained from sample statistics.    Topics | Tree View | Search | Help/Instructions | Program Credits Administration