ABSTRACT
There are many variations on the logistic regression model beyond the applications to unmatched and matched binary outcome data considered in Chapters 12 to 16. These include (1) variations in the basic linear form assumed for risk variables in the model, (2) more complex outcome data, and (3) more structured forms of data sampling. We briefly discuss each of these in turn. In Sections 17.3 to 17.5, we turn in a different direction to consider a detailed look at how risk changes dynamically over the risk period via a regression model for the hazard function.
