Reply
Thu 25 Jun, 2015 01:47 am
In LISREL, if the modification report after running the model ask for letting the error covariances of two variables to correlate, is it valid to let the error covariances of two variables belonging to two different latent variables to correlate?