Have you considered the 'lme' function in library(nlme)? The 'lme'
function is new, much more flexible, and as far as I know at least as
good and usually better than 'aov'. I would try the following:
fit - lme(response~CO2*NITROGEN, random=~1|ROOM)
Checking
How do I set up my model equation in aov to analyze a split-plot design?
I have two factors (CO2 and NITROGEN), each with two levels (high and
ambient). CO2 is my whole-plot factor with three replicates for each level
(i.e., 6 rooms total).
Is this syntax below correct?
summary(aov(response