Statistical and Computational Methods for
Survival and Reliability Data
Organizer: Luis A. Escobar
(luis@stat.lsu.edu)
Louisiana State University

Description:
Censored and truncated data arise frequently in survival and product reliability studies. An efficient use of the data require non-standard, non-parametric, semi-parametric, and parametric methods.

Most of the useful proposed methodology is likelihood based (see for example, Meeker and Escobar, 1999; Lawless, 1982; Fleming and Harrington, 1991). Computational intensive methods like Bootstrap and likelihood confidence intervals seem to offer some advantages in the assessment of uncertainty of point estimates in this case. For planning survival and reliability studies, simulation can be extremely useful to calibrate and complement traditional asymptotic methods.

Format:
In this session, the speakers will discuss current and future research issues that relate to the software implementation of statistical methods for survival and reliability data.

Participants:
Bradley Jones (presentation, A Case Study in Competing Risk Reliability Analysis Using JMP Software)
Bradley Jones is a Principal Systems Developer and Senior Statistician at SAS Institute Inc., where he is a member of the JMP development team. He worked for five years as a practicing statistician at Rockwell International. For more than 10 years, he has worked in computational statistics building data analysis statistical tools for scientists and engineers, like the MATLAB Statistics Toolbox. He was a founding partner in Catalyst, Inc., and designed Catalyst DOE, an innovative graphical approach to design and analysis of experiments. His areas of current interest are design of experiments, graphical user interface design, nonlinear models, and statistical simulation. In his talk, Brad will discuss current survival and reliability methodology as implemented in the JMP SAS software: methods for censored and truncated data, multiple failure modes, methods for recurrence data. He will also discuss the use of the JMP scripting language to enhance the current GUI capabilities.

William Q. Meeker (presentation, Reliability Data Analysis Using S-Plus)
Bill Meeker is a Distinguished Professor of Liberal Arts and Sciences at Iowa State University. He is an expert on Applied Statistics and recognized as an expert on Reliability, Accelerated Testing, Industrial Statistics, and Statistical Computing. In his talk, Bill will discuss computational and statistical issues related to SLIDA, a collection of S-PLUS functions for Life data analysis that has been designed to extend and enhance the S-PLUS capabilities. Bill will also briefly describe related ongoing and future work in reliability.

Terry M. Therneau (presentation, Random Effects Survival Models for Familial Data)
Terry Therneau is the Head of the Section of Biostatistics at Mayo Clinic. His research interests include Clinical Trials, Survival Analysis, Bootstrap, CART, and other computationally intensive methods; but particularly statistical aspects of the clinical research problems that arise in the section's day-to-day consulting practice. Terry will discuss the use of random effects survival models for correlated familial data, both computational issues and examples of their use in some current analyses.