I am analyzing an experiment that has the following design:
Factor1 Factor2
Level1 Level1
Level2
Level2 Level1.1
Level1.2
Level2
In other words, Factor2 has different number of levels across Factor1. I'd like to use logistic regression, but how should I set it up? Is it legitimate to run two regressions ignoring one of the levels in each (e.g. Level1.1 or Level1.2)?