

Latent groupmean centering 

Message/Author 


Dear Dr. Muthen, I am confused by the descriptions about how Mplus (version 8.2) centers variables that are not mentioned on neither "WITHIN" nor "BETWEEN". In the User Book (Version 8, pp. 274275) the following explanations are provided: "When a covariate is not mentioned on the WITHIN statement, it is decomposed into two uncorrelated latent variable... This decomposition can be viewed as an implicit, latent groupmean centering of the latent withinlevel covariate" After reading this description, I am very confused. What is the difference between this "latent groupmean centering" and groupmean centering Mplus does by using "DEFINE: center x y m(groupmean)"? When writing the Method part in my paper, is it correct to equal this "latent groupmean centering" with normal groupmean centering, and simply say, all these variables are groupmean centered? Thank you! 


No, the two approaches are different and usually the centering using the latent variable decomposition is better. This is described in the 2 papers: Lüdtke, O., Marsh, H.W., Robitzsch, A., Trautwein, U., Asparouhov, T., & Muthén, B. (2008). The multilevel latent covariate model: A new, more reliable approach to grouplevel effects in contextual studies. Psychological Methods, 13, 203229. Asparouhov, T., & Muthén, B. (2019). Latent variable centering of predictors and mediators in multilevel and timeseries models. Structural Equation Modeling: A Multidisciplinary Journal, 26, 119142. DOI: 10.1080/10705511.2018.1511375 

song hye sun posted on Tuesday, September 08, 2020  2:01 am



Dear. Muthen Hello, I have a question while studying DSEM. In example 9.31 of userguide 8, the covariate X is person mean centering, but can't the covariate be latent person mean centering? I wonder why you did cluster mean centering(person mean centering). If I can do latent person man entering, how should I set the code? Thank you always. 


Latent centering wasn't available with random slope when UG example 9.31 was written but today I would do it that way instead. 


Thank you!! if X (group mean) is not separately set in DEFINE, is latent person mean centering? 


Yes, if you don't put the X on the Within or Between list. Unless the output states that latent variable decomposition cannot be made. 

Back to top 

