I'm quite confused about how R calculate likelihood/AIC in glm.nb after adding the offset. In this example, the regression coefficients are the same but the AICs are different. I don't really understand how this works.
This is the example from: Where does the offset go in Poisson/negative binomial regression?
