'tau','alpha','delta'

76 views
Skip to first unread message

asas asas

unread,
Jan 21, 2019, 2:05:34 PM1/21/19
to lavaan
Dear lavaan users,
I built a lisrel model with meanstructure and the inspect method gives the model matrices.Seems 'nu' is the intercept for latent variables,'beta' is the coefficients for structural model and 'lambda' is the coefficients for the measurement model.
I still want to know the meaning of the extra matrices,'tau','alpha'.'delta'.When will these matrices show up in the model?I guess 'tau' shows up if I add '~1' in the measurement model?But what is  the 'alpha', and 'delta'?I find them in the code.
Best,
Yilun

asas asas

unread,
Jan 21, 2019, 2:07:41 PM1/21/19
to lavaan

These matrices are defined in 'lav_representation_lisrel.R' in the unzipped package source file.lr.png



在 2019年1月21日星期一 UTC-5下午2:05:34，asas asas写道：

Terrence Jorgensen

unread,
Jan 25, 2019, 9:45:34 AM1/25/19
to lavaan
Seems 'nu' is the intercept for latent variables,

No, nu is the intercept for the observed variables, alpha is the intercept for the latent variables.  lavaan borrows the labels of Mplus, not the original LISREL name for observed-variable intercepts (tau, which Mplus uses to represent thresholds... unnecessarily confusing, isn't it?).
 
'beta' is the coefficients for structural model

Yes
 
and 'lambda' is the coefficients for the measurement model.

Specifically the factor loadings.
 
I still want to know the meaning of the extra matrices,
 
'tau'

Thresholds
 
'alpha'

latent intercepts
 
'delta'

scaling factors:


When will these matrices show up in the model?

tau and delta appear when you have categorical endogenous variables.  


alpha and nu appear whenever you have a mean structure in your model, which is not the default in single-group models unless you set meanstructure=TRUE.


Terrence D. Jorgensen
Assistant Professor, Methods and Statistics
Research Institute for Child Development and Education, the University of Amsterdam

asas asas

unread,
Jan 25, 2019, 2:57:17 PM1/25/19
to lavaan
VERY MUCH THANKS.


在 2019年1月25日星期五 UTC-5上午9:45:34，Terrence Jorgensen写道：
Reply all
Reply to author
Forward
0 new messages