R help - Mar 2009 - Descriptive stats for factors in SEM

I feel really dumb for having to ask this, but here I go anyway...

I perform structural equation modeling on a survey, using about 25
variables that create a total of 5 latent variables (factors).

Applying sem (using the sem package) was a piece of cake, even for an
(SEM) layman, thanks for THE excellent work here. I have all the
variable/path coefficients, A- and P-matrix etc. But now I am stuck with
a seemingly simple task.

How do I compute the values of my latent variables, using the data I have?

I need a table for my factors, showing mean, st.dev, and correlations,
as seems common in journals. But I don't seem able to calculate the
values that my factors take. How would I go best about to get
descriptive statistics and a correlation table for my factors?

I searched the archives and read as many articles by John Fox as I
could, but this seems way to basic to deserve any mention :-).

Thanks for any hint,
Sebastian Spaeth

Dear Sebastian,

What you're looking for are factor-score coefficients, which would allow you
to estimate the values of the factors from the observed variables. Then,
given the original dataset from which the input-covariance matrix to sem()
was computed, you could find the factor scores. The sem packages doesn't
provide factor-score coefficients, but this is a reasonable request, and
I'll add it to my to-do list. I can't promise when I'll get this
done.

I'm not sure why you want the means, standard deviations, and correlations
of the latent variables. In most models, the means will be 0. You could
figure out the covariance matrix of the latent variables; sem()doesn't
provide that either. 

Regards,
 John

------------------------------
John Fox, Professor
Department of Sociology
McMaster University
Hamilton, Ontario, Canada
web: socserv.mcmaster.ca/jfox

> -----Original Message-----
> From: r-help-bounces at r-project.org [mailto:r-help-bounces at
r-project.org]
On
> Behalf Of Sebastian Spaeth
> Sent: March-04-09 5:20 AM
> To: r-help at r-project.org
> Subject: [R] Descriptive stats for factors in SEM
> 
> I feel really dumb for having to ask this, but here I go anyway...
> 
> I perform structural equation modeling on a survey, using about 25
> variables that create a total of 5 latent variables (factors).
> 
> Applying sem (using the sem package) was a piece of cake, even for an
> (SEM) layman, thanks for THE excellent work here. I have all the
> variable/path coefficients, A- and P-matrix etc. But now I am stuck with
> a seemingly simple task.
> 
> How do I compute the values of my latent variables, using the data I have?
> 
> I need a table for my factors, showing mean, st.dev, and correlations,
> as seems common in journals. But I don't seem able to calculate the
> values that my factors take. How would I go best about to get
> descriptive statistics and a correlation table for my factors?
> 
> I searched the archives and read as many articles by John Fox as I
> could, but this seems way to basic to deserve any mention :-).
> 
> Thanks for any hint,
> Sebastian Spaeth
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.