Professor Swank, I'm not sure I understand what you'r saying here. Are the mean structures you're referring to the sufficient statistics? And I'm not sure what you're referring to regarding what I'm "doing".
Are you referring to the fact many users of SEM, since they only need the sample means and covariance matrix to execute model fitting, only care about the resulting estimates for the covariance parameters and the effect coefficients? If so, I'm curious why they wouldn't also want to test the model on the actual data. Hence, my initial question about using fit statistics to compare the observed outcomes versus the predicted. Thanks for any input you might have. [EMAIL PROTECTED] (Paul R Swank) wrote in message news:<025101c21c77$694aa220$520e6a81@PEDUCT225>... > That seems to me to be what you're doing in SEM when you use mean > structures. > > Paul R. Swank, Ph.D. > Professor, Developmental Pediatrics > Medical School > UT Health Science Center at Houston > > > >>> Original Snipped <<< ########################################################################### . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
