I have a randomised complete block deisgn for an animal trial where the animals were blocked according to weight, litter etc. Each block contained a complete set of treatments. There are 24 replicates (blocks) in total, but these were carried out across two batches, so 12 reps in each batch. The experiemntal design is a 2 x 2 factorial, with two levels of dietary supplements.
I have been analysing it using a 2 way ANOVA with rep as a fixed factor and looking at interactions, however when i do this i get a significant interaction between rep and treatment...so I presume the differences amoung treatments changes between blocks. What can I do now? Is this important? Does this violate the assumption of additivity?
Also, do I need to include batch as another fixed factor? If I do this, the program cant compute a significance level for batch?