Standardized Parameter Estimate Confidence Intervals

Posted on
No user picture. Hal Joined: 09/10/2021
I'm using the OneACE5c Sex-Limitation model script which outputs both the standardized and unstandardized parameter estimates. However, it only includes the unstandardized CIs (I believe using the "mxRun( ModelName, intervals = T" command). Is there a way to get the standardized CIs as well?
Replied on Sat, 05/21/2022 - 01:49
No user picture. Hermine Joined: 07/31/2009


# Create Algebra for Variance Components
rowVC <- rep('VC',nv)
colVCm <- rep(c('Am','Cm','Em','SAm','SCm','SEm'),each=nv)
colVCf <- rep(c('Af','Cf','Ef','SAf','SCf','SEf'),each=nv)
estVCm <- mxAlgebra( expression=cbind(Am,Cm,Em,Am/Vm,Cm/Vm,Em/Vm), name="VCm", dimnames=list(rowVC,colVCm))
estVCf <- mxAlgebra( expression=cbind(Af,Cf,Ef,Af/Vf,Cf/Vf,Ef/Vf), name="VCf", dimnames=list(rowVC,colVCf))

# Create Confidence Interval Objects
ciACE <- mxCI( c("VCf[1,1:3]","VCm[1,1:3]") )

The VCm and VCf matrices hold both the unstandardized and standardized variance components. To get confidence intervals around both, we simply extend the subscripts for the columns from VCf[1,1:3] to VCf[1,1:6] and similarly for VCm.