ABSTRACT
Gupta and Das (Citation2000) studied the resin data for improving the resistivity of urea formaldehyde through the setting of process parameters. They noticed that variances of the responses are non-constant, affected by some factors. In quality-improvement engineering applications, achieving high precision by minimizing variance is as important as getting the mean at the target. To identify factors affecting variance they used the analysis of variance method for signal-to-noise ratio. However, their method could be statistically inefficient to miss important factors as insignificant. We propose to use joint modeling for the mean and dispersion, which gives completely different analysis for the resin data.
ACKNOWLEDGMENTS
The authors thank the referee whose comments greatly improve this article. This work was partially supported by the second stage Brain Korea 21 Project.