ABSTRACT
This chapter presents the analysis of incomplete data, caused by censoring in event-time survival data. Cox's proportional hazards model is widely used for the analysis of survival data. This model is interesting due to its semi-parametric nature, where the baseline hazards assume non-parametric but treatment effects are modeled parametrically. Either a log-normal or gamma distribution can be used as the frailty distribution; therefore, normal and log-gamma distribution can be adopted for the log frailties. Traditionally two legitimate and identifiable quantities to analyze competing risks data have been the cause-specific hazard function and the cumulative incidence function. The dataset has the variable "smkonset", coded by i when the event occurs in the ith time interval and the variable "event", coded 1 if he/she smoked, 0 otherwise. Competing risks data usually arise when an occurrence of a competing event prevents the occurrence of the event of interest.
