Dear all,
I have tested for the measurement invariance between pre- and post-test scores by running sem. Scalar invariance is supported. I have also generated the pre- and post-test latent factor score by using the "predict" command. Afterwards, I would like to run a regression to see how the pre-test latent factor score and some other demographic variables relate to the post-test latent factor score. However, I have read from some posts that it is incorrect to run regression by using the latent factor scores. Instead, an sem incorporating all variables to be tested should be run. May I know what stata command I should write for this sem?
Thank you!
The stata commands that I have run are:
I have tested for the measurement invariance between pre- and post-test scores by running sem. Scalar invariance is supported. I have also generated the pre- and post-test latent factor score by using the "predict" command. Afterwards, I would like to run a regression to see how the pre-test latent factor score and some other demographic variables relate to the post-test latent factor score. However, I have read from some posts that it is incorrect to run regression by using the latent factor scores. Instead, an sem incorporating all variables to be tested should be run. May I know what stata command I should write for this sem?
Thank you!
The stata commands that I have run are:
PHP Code:
sem (Ilo -> ilo1 ilo2 ilo3 ilo4 ilo5 ilo6) (ilo1 <- Ilo _cons@0), cov(e.ilo2*e.ilo3) cov(e.ilo1*e.ilo2) cov(e.ilo1*e.ilo3) group(pre_post) mean(Ilo) ginvariant(mcoef mcons)
predict Ilo if ilo1!=., latent (Ilo)
