Sankhya: The Indian Journal of Statistics

2000, Volume 62, Series B, Pt. 3, pp. 417--432

BAYESIAN ANALYSIS OF BIVARIATE COMPETING RISKS MODELS

By

J. BÉRUBÉ, Becton Dickinson, Franklin Lakes

and

C.F.J. WU, University of Michigan, Ann Arbor

SUMMARY. Robust parameter design works by identifying factor settings to reduce variation in products or processes. One of its key elements is the use of the Signal-to-Noise (SN) ratio for parameter design optimization, which has stirred some controversies in the past. In this article, modeling and analysis of the SN ratio and related measures are considered along with their validity under various models. Results show that the performance of the SN ratio is very much model-dependent and its validity deteriorates as the true model deviates from the assumed model. Use of the log sample variance is less model-dependent.

AMS (1991) subject classification. Primary 62K25; secondary 62P30.

Key words and phrases. Signal-to-noise ratio, quality improvement, variation reduction.

Full paper (PDF)

This article in Mathematical Reviews