The Choices
    Standardization can occur at many places within causal modeling.  It can    
occur in the data when scores themselves are standardized.  This is not 
usually advisable because it introduces too much rounding error.
Standardization can also occur in the input to a SEM program.  Even if
one wants to analyze a covariance matrix, it is advisable, in terms of
computation accuracy, to input a correlation matrix with the standard 
deviations, instead of a covariance matrix.

    Most estimation procedures presume that a covariance matrix has been
inputted.  Even if the correlation matrix is analyzed, latent endogenous 
variable are not standardized (though see below).  Nonstandardized solutions 
can be transformed into a standardized solution.

    Technically, standard errors refer to input using a covariance matrix.
If a correlation matrix is inputted, the solution and standard errors refer
to variables that have transformed using the sample standard deviations.
It is not incorrect, as some have asserted, to use the significance tests if 
the correlation matrix is analyzed.

    There are new estimation methods based on the assumption that the 
correlation matrix has been entered. See the Browne program RAMONA.  This 
procedure does allow the standardization of latent endogenous variables.

    For more discussion on this issue, consult Ed Rigdon's page.
    
When to Standardize
    units of measurement not very interpretable
    desire to compare coefficients with different units of measurement
    more experience with betas than b coefficients

When Not to Standardize
    units of measurement are meaningful
    paths are usually set equal (so in multiple groups analysis one should 
        analyze the covariance matrix)

Go to the causal modeling page.