Standardized Estimates
Posted on

At a recent workshop, someone asked how to obtain standardized estimates from OpenMx. Using the path-style input this is straightforward, and can be done with a couple of mxEval() statements and an omxRAMtoML() call:
# Now standardize solution
mxEval((solve(vec2diag(sqrt(diag(S)))))%&%S,threeLatentMultipleReg1Out)
mlthreeLatentMultipleReg1Out<-omxRAMtoML(threeLatentMultipleReg1Out)
mxEval(solve(vec2diag(sqrt(diag(solve(I-A)%&%S))))%*%A%*%vec2diag(sqrt(diag(solve(I-A)%&%S))),mlthreeLatentMultipleReg1Out,compute=T)
To get the standard errors of these standardized estimates is another matter, any volunteers?