Skip to content

gamm4 model formulation clarification

1 message · Reinhold Kliegl

#
On the difference between the two specifications: Perhaps you have
treatment contrasts for design and soa. Then the fixed effect
coefficients will not estimate the main effect and the interaction.
You would need to specify  "contrasts(design) <- contr.sum(2)/2, etc."
for them if you specify the contrast matrix as a I suggested (which
returns estimates of the difference between levels, not as a
difference from the grand mean as the default).

Reinhold Kliegl
On Sun, Aug 1, 2010 at 2:42 AM, Mike Lawrence <Mike.Lawrence at dal.ca> wrote: