Covariates of Distal Outcomes PreviousNext
Mplus Discussion > Latent Variable Mixture Modeling >
Message/Author
 Tamiko Younge posted on Tuesday, August 07, 2012 - 7:54 am
Hello,

I am trying to use latent class variable (3 classes) to predict distal outcomes (continuous variables). However, the outcomes covary with age and gender. If I'd like to control for these covariates in my prediction of the distal outcome, do I include a regression of the outcome on the cofactors in the model (i.e., outcome on age gender) or do I include covariates in the AUXILIARY option along with the outcomes of interest? Thank you so much for your help.

Best, Tamiko
 Jon Heron posted on Wednesday, August 08, 2012 - 4:32 am
Hi Tamiko

this sounds very similar to a question I asked in January:-

http://www.statmodel.com/discussion/messages/13/8632.html?1325836286

I think if you replace "binary" with "continuous" and "thresholds" with "means" then that's probably your answer. Reckon that agrees with the first option you mention.
 Linda K. Muthen posted on Wednesday, August 08, 2012 - 9:32 am
I would include them in the model.
 Tamiko Younge posted on Wednesday, August 08, 2012 - 10:00 am
Thanks for all your help!

Just to confirm, by "include them in the model" do you mean include the covariates of the distal outcome variable as covariates of the latent class variable (i.e., include them in the usevar command and include c ON age gender)?

I tried to include a regression of age and gender on the distal outcome as suggested in the earlier post (i.e., AUXILIARY=distal (e) in the VARIABLE command and in the MODEL command distal on age gender), but Mplus said "Unknown variable(s) in an ON statement: distal."

Let me know if you have any further suggestions. THANKS AGAIN!

Best, Tamiko
 Linda K. Muthen posted on Wednesday, August 08, 2012 - 11:51 am
Yes, I mean c ON age gender.

If age and gender are on the AUXILIARY list, they cannot be used in the MODEL command.
 matt thullen posted on Thursday, December 19, 2013 - 10:35 am
Is there anyway to get descriptive stats on how covariates vary across classes if they have been used in the MODEL command?
 Bengt O. Muthen posted on Thursday, December 19, 2013 - 1:26 pm
Use the RESIDUAL option of the OUTPUT command.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: