ABSTRACT
In many longitudinal studies carried out in medicine, it will be clear that the assumption of normality for the response variable is simply not justified. Two examples are shown in Tables 14.1 and 14.2. The one in Table 14.1 results from a clinical trial comparing two treatments for a respiratory illness (Davis 1991). In each of two centres, eligible patients were randomly assigned to active treatment or placebo. During treatment, the respiratory status (categorised as 0 = poor, 1 = good) was determined at each of four monthly visits. A total of 111 patients were entered into the trial, 54 into the active group and 57 into the placebo group. The sex and age of each participant were also recorded and a baseline respiratory status. Here the response variable is binary, making the models described in the previous chapter inappropriate for these data. (These data were used previously in Chapter 12.) The observations for the first five patients in the data set are shown in Table 14.1.
