add control variable in SEM

[rosie]

Hi all, I want to use control variable in my SEM model.

my basic SEM model is below:

"model1 <-

addiction =~ s + sm + so

depression =~ emotionalproblem

brain =~ rpcv + lfv

depression ~ addiction

brain ~ depression

"

and I want to add control variable: age, sex, handedness, eTIV

to be more specific, 

regression in 'depression ~ addiction' (control variable - age, sex, handedness)

and 'brain ~ depression' (control variable - age, sex, handedness, eTIV)

As far as I know, the concept of control variable in SEM is not clear. But some people say there are ways to act as a control variable. It would be really helpful if you could share what you know.

sorry about my english.

regard, Rosie

[Keith Markus]

Rosie,

Forgive me if I am missing the import of your question but I believe you can simply add the control variables as additional predictors/causes of the endogenous variables.  There is no statistical distinction between predictors/causes of interest to you and control variables in SEM.  It is not like ANCOVA where one class of variables has to be categorical but the other can be continuous.  The same rules apply to both types of variables.  The effect coefficients estimate the expected change in the DV accompanying a unit change in the IV holding constant all the other variables in the equation.  Under the right circumstances, this can also estimate the effect of the IV on the DV holding constant the other variables in the equation.

Keith

------------------------
Keith A. Markus
John Jay College of Criminal Justice, CUNY
http://jjcweb.jjay.cuny.edu/kmarkus
Frontiers of Test Validity Theory: Measurement, Causation and Meaning.
http://www.routledge.com/books/details/9781841692203/