Observed measures as a dependent vari... PreviousNext
Mplus Discussion > Structural Equation Modeling >
 karen dunn posted on Wednesday, March 14, 2012 - 2:33 am
Hi, I have been asked to look through some analysis for a collegue. One of the models I have been given uses an observed variable (obs10) as a dependent variable. The model specification is shown below.

latent1 BY obs1 obs2 obs3;
latent2 BY obs4 obs5 obs6 obs7;
latent3 BY obs8 obs9;
obs10 ON latent1 latent2 latent3;

I have heard that it is ok to use observed variables in this way, but am struggling to see how obs10 plays a different role in the model to that of an indicator variable that loads on all three latent variables. Is there a statistical difference in the ON command as opposed to the BY command? Are you able to recommend any papers/resources that give information about using measured variables as dependent variables in this way?

Many thanks,
 Linda K. Muthen posted on Wednesday, March 14, 2012 - 2:47 pm
There is no statistical difference between regressing obs10 on the three latent variables or having obs10 be a factor indicator for each factor. The difference comes in the interpretation.
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