Groupmean centering in type=complex PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Matt Easterbrook posted on Thursday, April 12, 2012 - 7:45 am
Hi,

I am conducting a complex sample analysis and want my predictor variables to be group-mean centred. I've been advised by someone who know mplus and the methodoloy very well that when doing this type of analysis MPLUS automatically adjusts the predictor variables to be group-mean centred - is this true, or should I be using centering is groupmean (x1 x2)?

Thanks
 Linda K. Muthen posted on Thursday, April 12, 2012 - 2:16 pm
Mplus does not center observed variables as the default. You must center observed variables using the CENTERING option. Perhaps your advisor was thinking of latent variable interactions in Mplus where the latent variables do not need to be centered.
 Matt Easterbrook posted on Thursday, April 12, 2012 - 4:24 pm
Thanks. So does MPLUS group centre latent variables by default if they are included in an interaction?

I am using interactions with latent variables in the structural part of my two level SEM, along with observed variables too. Does MPLUS centre the latent variables but not the observed ones? Would I have to group mean centre the observed ones so that they are all expressed the same way?

Thanks for you help.
 Linda K. Muthen posted on Thursday, April 12, 2012 - 4:55 pm
No, it does not. These latent variable do not need to be centered. Their means are zero.

You need to center the observed variables if you want them centered.
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: