ABSTRACT
The nature and strength of dependencies among the failure times themselves may be interest in some settings. Failure time methods have application in many subject matter and research areas, including biomedical, behavioral, physical, and engineering sciences, and various industrial settings. The usual presence of right censoring has implications for the types of statistical models that can be reliably applied, even if the covariates are time-independent. In an important subclass of applications there is a single failure time axis with individuals experiencing a failure continuing to be followed for second and subsequent failures. An additional goal of multivariate failure time data analysis in some contexts may be a summary of the effects of a treatment or an exposure across a range of health-related outcomes or across a set of failure-types more generally. There is a long history of modeling failure time data using survivor and hazard functions. The chapter also presents an overview of the key concepts discussed in this book.
