I am doing a mixed model using the different subjects as a random effect. For avoiding any between-subject comparison I am mean-centering the different explanatory variables. Is it correct to also mean-center the dependent variable? I do not find any reason not to do it, but I have not find any paper doing it.
Thank you
Best Answer
There is no reason to center the dependent variable. All this will achieve is to change the estimate for the global intercept (fixed effect). All the other estimates will remain unchanged. If you do center it, then you will need to add the value of the mean to get predictions on the original scale.
For example: