Command: factors to explain observabl... PreviousNext
Mplus Discussion > Structural Equation Modeling >
 Aymeric Punel posted on Thursday, February 16, 2017 - 2:47 pm

Sorry if this question has already been posted or is not put in the right section of the forum.

I'm building a model from data I got from a survey. I design my factors with the items that significant load on them. These items are likert scale statement.

Then, I'd like to see if the factors that I've built can explain one particular behavior of my respondent. In the survey, I have asked a question to know whether people have a special characteristic. In my Mplus model, I'd like to find the command that links my factors to this characteristic, where the factors explain this characteristics.
 Linda K. Muthen posted on Thursday, February 16, 2017 - 5:12 pm
It sounds like you want to use factors as covariates. You would say

factor ON covariates;
 Aymeric Punel posted on Friday, February 24, 2017 - 12:43 pm

Thank you for your answer and sorry for the delay.

I used this command (factor ON covariates) but this would mean that it's the covariate that affect the factor.
In my case, I also want to know the command for "the contrary" where it's the factor that can predict my dependent variable.

For example:
I have a factor which is "Willingness to Play Money". It's unobserved, and I define it by a set of items. Let's call it WPM

I say that age affects this factor.
So I'll write WPM on AGE (This is an example I got from a marketing class I took, professor used this way of coding when a variable explain the factor. Maybe this is wrong)

But also, I have another variable where I asked people (from a survey) whether they were gambler or not. And I want to predict whether they are gambler or not with my factor.

It would be something like GAMBLER on WPM or GABMLER by VPM

Let me know if I'm not clear.

 Linda K. Muthen posted on Friday, February 24, 2017 - 4:07 pm
If you want a factor to be a covariate, you would use an ON statement, for example,

y ON factor;

This is technically the same as

factor BY y;

because the factor model regresses the factor indicator on the factor. BY is used as a vehicle for naming the factor.
 Aymeric Punel posted on Wednesday, March 01, 2017 - 12:53 pm
Hello Linda,

Thank you for your answer.

Just to be sure, how would you code this model (link to the .png) : ?

and then this one :

The difference between both is the direction of the arrow

Thanks a lot for your quick answers !
 Linda K. Muthen posted on Wednesday, March 01, 2017 - 2:55 pm
We are reluctant to go to links. Please send pdf's of the two diagrams along with your license number to
 Aymeric Punel posted on Thursday, March 02, 2017 - 5:49 am

I just sent an email.

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