I have a 2 level linear mixed effects model with Time nested within University. Time is treated as continuous and I am modeling this within SPSS. The outcome variable is the number of undergraduate degrees awarded in a STEM discipline.

At level 1, I have a number of continuous covariates that change from year to year (e.g., full-time enrollment, percent of students that receive federal aid, etc.).

I would like to grand-mean center all my covariates (both Levels 1 and 2), but how is the mean calculated for Level 1 time varying covariates? Do I simply use the mean across all time points, ignoring the nesting?


It depends on the model you are trying to fit, something only you can determine as a function of your objectives. Judith Singer has an article about fitting individual growth models in SAS which is very helpful in this regard. It doesn't matter that it's SAS focused. Her recommendation is that for certain types of models, centering on the grand mean is a best practice, Using SAS PROC MIXED to Fit Multilevel Models, Hierarchical Models, and Individual Growth Models


| cite | improve this answer | |
  • $\begingroup$ I appreciate your answer, but I don't need information on whether I should grand mean center, I would like information on how i grand mean center a continuous time-varying covariate. $\endgroup$ – Michelle H Mar 10 '16 at 0:07
  • $\begingroup$ If you read Singer's article, her recommendation is to mean center across all time points. $\endgroup$ – Mike Hunter Mar 10 '16 at 2:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.