The binomial, Gaussian and Poisson GLMs are by far the most commonly used, but
there are a number of less popular GLMs which are useful for particular types of data.
The gamma and inverse Gaussian are intended for continuous, skewed responses.
In some cases, we are interested in modeling both the mean and the dispersion of
the response and so we present dual GLMs for this purpose. The quasi-GLM is a
model that is useful for nonstandard responses where we are unwilling to specify the
distribution but can state the link and variance functions.