ABSTRACT
Now that we have spent some effort developing an understanding of various simple versions of the logistic regression model, we are eager to apply the model to examples. We want to know, for example, answers to the following: Does logistic regression really describe how body weight influences the risk of CHD? What happens when we add behavior type into the model? How about the level of coffee drinking and incidence of pancreatic cancer? Before going into further issues, such as confounding and interaction, in the context of logistic regression, we first consider estimation of the parameters of a specific population model using randomly sampled data. Initially we direct our attention to population-based and cohort designs. In Section 13.3 we extend estimation techniques to allow for case-control data.
