Hi Listers,
I am wondering about how 'K' (number of parameters in AICc equation) is
calculated for mixed effects models. I am using aictab in the AICcmodavg
library, to produce an AICc table for a set of models. In the table, K
seems to equal number of fixed effects + number of random effects + 2. Is
that the correct calculation? In Vaida and Blanchard (2005), they suggest
that for the 'population' focus (ie when you are interested in the fixed
effects, not the random effects, which is the case for my study), K is the
number of fixed parameters counting mean parameters and variance
components. I'm not sure how you calculate the number of variance
components, so I'm not sure if this gives the same value for K as is
supplied in aictab? So I'm wondering if the K given in aictab is really
correct for a 'population focus', and if not, how should K be calculated?
Thanks for your help
Megan G.
[[alternative HTML version deleted]]