Thanks J.J .This puts things in perspective. I was not visiting google groups for some time and saw ur response today. Sorry for that. > > I have a few questions also, perhaps you want to think about them > before doing any analysis. whatever entity you are measuring using > the semantic differential technique, is there any reason to believe > the variable is distributed normally in the population. No. They are as I expected them to be and the skewness is in the expected direction.The point is ---can I or rather, should I run a structural equation mode3l using Lisrel if they are non-normal?
> regarding (2), the answer, it seems to me, is not the software to use, > but rather what are you doing? what is the criterion measure? are > there multiple predictors, one of which is binary? is it truly binary > or a continuous variable that has been dichotomized? i want to know this in order to evaluate the measurement model. the binary variable is a dependent variable. the point is I need to claculate the correlation between constructs to demonstrate discriminant validity and nomological validity. On the face of it, discriminant validity is pretty apparant. . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
