Hi,
I am running an analysis in proc Glimmix using the LaPlace method, since I need to compare models.
My design is unbalanced and I am wondering if this will affect the degrees of freedom estimation, since the LaPlace method is, more or less, restricted to the containment or betwithin method. The SAS documentation indicates that the betwithin method is appropriate for moderately unbalanced designs.
I would really appreciate some support with this issue.
Best regards,
B