why covariance lines between dependent variables + why orthogonal option affects dependent variables

[fahd alfarsi]

Hello everyone,

I am new in R and I have two questions. I hope someone can help me, please. First question: why there are covariance lines shown between dependent variables when I use SemPaths. Second question: when I am setting orthogonal =TRUE, I expected only independent variables to uncorrelate, but this also affects the dependent variables and became uncorrelated. 

This is the model I am using: 

model1<-'

# the measuremnt model

MAV =~ MAV1+MAV2+MAV3+MAV4+MAV5+MAV6;

MAP =~ MAP1+MAP2+MAP3+MAP4;

PAP =~ PAP1+PAP2+PAP3+PAP4;

PAV =~ PAV1+PAV2+PAV3+PAV4;

OSE =~ OSE1+OSE2+OSE3+OSE4+OSE5+OSE6+OSE7+OSE8;

# the structural model

MAV+MAP+PAP ~ PAV + OSE;

'

m1<-sem(model=model1,data=UK) 

summary(m1,fit.measures=TRUE,standardized=TRUE)

semPaths(m1,title=FALSE, what="std", style="ram", layout="tree")

Many thanks,

Fahd

[Terrence Jorgensen]

why there are covariance lines shown between dependent variables when I use SemPaths.

The sem() function sets the option auto.cov.y = TRUE, so the endogenous variables will covary unless you set that option to FALSE or fix it to zero in your syntax.  See the ?lavOptions help page for a description of the option.

when I am setting orthogonal =TRUE, I expected only independent variables to uncorrelate, but this also affects the dependent variables and became uncorrelated. 

Yves recently updated that description in the help page (available in the development version).  orthogonal = TRUE actually sets all factor covariances to zero.  It is still a useful shorcut if you want most covariances fixed to zero, because you can still free particular covariances in the syntax, e.g., PAV ~~ NA*OSE

Terrence D. Jorgensen

Assistant Professor, Methods and Statistics

Research Institute for Child Development and Education, the University of Amsterdam

http://www.uva.nl/profile/t.d.jorgensen