ABSTRACT
In modelling survival data, we attempt to explain observed variation in times to events such as death by dierences in the values of certain explanatory variables that have been recorded for each individual in a study. However, even after allowing for the eects of such variables, there will still be variation in their observed survival times. Some individuals may have a greater hazard of death than others, and are therefore likely to die sooner. They may be described as being more frail, and variation in survival times can be explained by variability in this frailty eect across individuals.
