I want to know if difference-in-difference approach can be used for two treatment groups. Can I use it it between them or should I first identify a control group and then compute the diff-in-diff between the two sets i.e. treatment one and control and second between treatment two and control?
or is there be any other approach to address this problem??