Normality of exogenous variables in SEM PreviousNext
Mplus Discussion > Structural Equation Modeling >
 Lesa Hoffman posted on Friday, September 01, 2006 - 10:19 am

I am recalling a conversation I had a few years ago with Linda at a workshop in Alexandria regarding the assumption of normality within SEM. My recollection is that she told me that this assumption only applies to endogenous variables in the model, and thus that normality in exogeneous variables is not a problem for the model (assuming there is no missing data on the predictor, and thus that it would not be treated as an outcome for that reason). Is my recollection correct? Do you have suggestions for references that refer to this idea explicitly?

Thank you!
Lesa Hoffman
 Linda K. Muthen posted on Friday, September 01, 2006 - 10:30 am
When a model is estimated conditioned on the observed exogenous variables, no distributional assumptions are made about these variables. I think a good reference would be a regression textbook.
 Lesa Hoffman posted on Friday, September 01, 2006 - 1:44 pm
Hi Linda,

Thank you for your quick response! Just so I'm absolutely clear, is it true that the lack of distributional assumptions made about exogneous variables holds for latent OR observed variables? I thought latent variables are assumed to be normally distributed - or is that only latent outcome variables?

 Linda K. Muthen posted on Monday, September 04, 2006 - 3:15 pm
The model is estimated conditioned on observed exogenous variables not latent exogenous variables. The latent exogenous variables are part of the model and as such have distributional assumptions made about them.
 Lesa Hoffman posted on Monday, September 04, 2006 - 5:39 pm
Got it, thanks!
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message