Anonymous posted on Tuesday, January 11, 2005 - 4:31 am
I performed an analysis with a dichotomous dependent variable (like in example cat5) and latent independent variables. My dependent variable is an observed variable. However, I do want to include the measurement error in that variable in the model. I assume that the error variance of that variable is about 20 percent of the total variance of that variable. I try to put this in the model by treating the dependent variable as a latent variable with one indicator and to fix the residual variance of that indicator to one fifth of its total variance. However, this does not seem to work. Is there a way to specify the size of the measurement error of a dichotomous dependent variable with one indicator?