I performed generalized linear mixed models to test the effects of a treatment on two different groups (I performed a model for each group). Now I would like to test which of the groups is most affected by the treatment, that is to compare the magnitude of the response. I know how to test for difference in means, but now how to test for a difference in a magnitude. Could someone give me a clue?
Are you using group as a predictor in your model? If so, the coefficient tells you the difference in magnitude. You could bootstrap CIs for the coefficient to get your answer. A little bit more information on what/how you are doing the analysis would be helpful.