• DOI:10.1198/tech.2003.s768
• Corpus ID: 10069035

The Statistical Analysis of Failure Time Data

@article{George2003TheSA,
  title={The Statistical Analysis of Failure Time Data},
  author={Laurence L. George},
  journal={Technometrics},
  year={2003},
  volume={45},
  pages={265 - 266}
}

• L. George
• Published 1 August 2003
• Mathematics
• Technometrics

Chapter 5, “Inference Procedures for Log-Location-Scale Distributions,” is concerned with likelihood-based inference under various censoring schemes for the important case where the logarithm of the lifetime is modeled with a location-scale distribution. The Weibull, extreme value, lognormal, and loglogistic distributions are discussed. In addition, two distributions with an additional parameter, the generalized log-Burr and the generalized log-gamma, are discussed. Chapter 6, “Parametric Regression Models,” presents models where one or more distribution parameters are modeled as a linear function of some regression parameters. The most popular of such models, the accelerated failure time model, where the location parameter in a location-scale distribution is modeled as a linear function of regression parameters, is discussed in detail. Extensions of this model in which the distribution is allowed to be the generalized log-Burr or the generalized log-gamma and the scale parameter is modeled with regression parameters are discussed. Also discussed in detail are graphical modelchecking techniques for assessing regression model Ž t. Chapter 7, “Semiparametric Multiplicative Hazards Regression Models,” focuses on the proportional hazards model. Chapter 8, “Rank-Type and Other Semiparametric Procedures for Log-Location-Scale Models,” presents the accelerated failure time analog to the models of Chapter 7. The regression model has a location-scale form, but no speciŽ c distribution is assumed. Chapter 9, “Multiple Modes of Failure,” gives a careful treatment of the analysis of data when failures can occur due to multiple failure modes. Chapter 10, “Goodness-of-Fit Tests,” begins with a general discussion of methods for testing goodness of Ž t. Some speciŽ c tests of Ž t for the exponential, Weibull, extreme value, normal, and lognormal are then presented for the case of type 2 censoring or complete data. Tests of Ž t in regression models are brie y discussed. Simulation methods for overcoming restrictions on the testing procedures are discussed in each case. Finally, Chapter 11, “Beyond Univariate Survival Analysis,” introduces multivariate lifetime data, discussing clustered lifetimes, event history data, and data where an internal process is related to the lifetime of interest. For those working with reliability data, the books by Meeker and Escobar (1998) or Nelson (1982) may be more useful as primary references because of their emphasis on engineering methodology and applications. However, this book complements the other references well, and merits a place on the bookshelf of anyone concerned with the analysis of lifetime data from any Ž eld.