[R] parameter constraints in sem

[John Fox]

Dear James,

I've been away and reading email at intervals, and also not following
the r-help list during my trip. I've taken the liberty of copying this
response to the r-help list along with your original posting. My
apologies for the delay.

The short answer to your question is "no," I'm afraid. sem()
distinguishes between fixed and free parameters by the fact that the
latter have names and the former don't. As well, it permits equality
constraints between parameters (by giving two or more the same name)
but doesn't accommodate more complex constraints.

I'll take your question as a suggestion for incorporating more general
parameter constraints and, when I have a chance, think about how that
might be implemented within the current general structure of the
program. I can't promise that I'll be able to get to this soon,
however.

If you're interested in tackling the problem yourself, I'd be happy to
incorporate your work.

Regards,
 John

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

Original message:

[R] parameter constraints with sem
James Fowler jhfowler at ucdavis.edu
Thu Mar 16 11:37:54 CET 2006

    * Previous message: [R] Did I use "step" function correctly? (Is
R's step() functionreliable?)
    * Next message: [R] Bioconductor package on linux machine
    * Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

I would like to constrain three of the parameters I am estimating using
sem (strucutral equation model package) but do not know how.
Specifically, I would like to use the following constraint:

beta1 + beta2 + beta3 = 1

can I assign variable names as the values for fixed parameters?  If so,
I
could define the third parameter to be equal to 1 - beta1 - beta2.

thanks,
james