0
$\begingroup$

I am using structural equation modeling (SEM). My model is a simple mediation model with latent variables (each latent variable has 3 indicators).

I want to run a Monte Carlo simulation to estimate if my fixed sample size (n = 2000) is large enough.

In the article, "Sample Size Requirements for Structural Equation Models: An Evaluation of Power, Bias, and Solution Propriety" the authors give syntax on conducting a Monte Carlo simulation in Mplus. They set the factor mean to 0. Why? Is it just convention to do so?

$\endgroup$

1 Answer 1

1
$\begingroup$

You've got to set them to something, or the model isn't identified.

If you've got to set a parameter to some value, you can choose anything, but if you want an easy life:

  • If you're going to have to add it to something, make it zero.
  • If you're going to multiply it by something, make it one.

Sometimes you need to add factor means, so you make them zero. (And loadings are typically multiplied, so you make them one).

But feel free to make them -17.43 if you want to. It won't matter.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.