In previous chapters, we have seen how to model a binomial or Poisson response.
Multinomial response models can often be recast as Poisson responses and the stan-
dard linear model with a normal (Gaussian) response is already familiar. Although
these models each have their distinctive characteristics, we observe some common
features in all of them that we can abstract to form the generalized linear model
(GLM). By developing a theory and constructing general methods for GLMs, we are
able to tackle a wider range of data with different types of response variables. GLMs
were introduced by Nelder and Wedderburn (1972) while McCullagh and Nelder
(1989) provides a book-length treatment.