compare combination of factors with actual combined treatment Assume I have two different treatments "Nitrogen" and "Phosphorus" which were tested each as a single treatment and as a combination. How can I determine wheter the combination is more or less than the two single treatments added together? So in fact I want to check whether there are interactions between the treatments.
Currently, my data looks like this:
Treatment1  Treatment2 Result
Nitrogen    NA    1
NA    Phosphorus    2 
Nitrogen    Phosphorus    3

Thanks alot, I can't seem to wrap my head around this.
 A: I suggest first reorganizing your data so you either have 1 treatment variable with levels of Phosphorus, Nitrogen, and Both, (Which would be useful to model), or three treatment variables, each of which represents one of the three treatments. 
One of the things you might try to do is to create an interaction plot. This can be done in R with the command 

interaction.plot(Treatment1, Treatment2, result)

Leaving the interaction (Nitrogen and Phosphorus) off this plot. The idea being that if the two lines on the graph intersect, then there is some interaction between your variables. If they do not (and do not appear likely to), then your variables do not interact. I'm not entirely sure how helpful this will be with your data, but I think it's worth a shot . 
You could also model the result based on your three terms, and take a look at the β value to assigned to your interaction term (in this case treatment 3). Something like: 

lm(Result ~ Treatment1, Treatment2, Treatment3, data)

Or, if you have just one treatment variable with multiple levels:

lm(Result ~ I(factor(treatment)), data)

The idea being that if the interaction term is 1) insignificant and 2) close to zero, then there's not much interaction going on. 
