Estimation of RAM-model

2 replies [Last post]
sbremer's picture
Offline
Joined: 11/08/2011

Sorry for the beginner's question...
But please help me understand: When I specify a path like

mxPath(
from=latents, # names of latent variables
arrows=2,
connect='single',
free=T,
values=NA,
labels=('lab1','lab2','lab3')
)

and run a model (RAM) with this and other path for the indicators with mxRun(), are 'lab1', 'lab2', 'lab3' the estimates of variances of the latent variables or are they estimates of variances of the latent disturbance?

Thanks in advance!!

Ryne's picture
Offline
Joined: 07/31/2009
It depends on what else is in

It depends on what else is in the model. This path statement creates two-headed arrows. If the latent variable is exogenous (meaning that no other variables predict it/have one-headed arrows leading to this variable), then these terms model the total variance of the latent variable. If the latent variable is endogenous (meaning that one or more other variables predict this latent variable via one-headed arrows/regressions), then this term represents the disturbance or latent residual variance.

sbremer's picture
Offline
Joined: 11/08/2011
Thanks!

Thanks!