Abstract
Two factors are related when one factor's desirable range depends on the level of another factor. For example, in a chemical reaction, the desirable reaction-time range shifts downward as temperature increases. This is reflected in the experimental design by choosing levels of one factor that depend on the level of the other factor. Taguchi refers to this technique as sliding levels and justifies it by rationales of bad region avoidance and interaction elimination. We explore these rationales and their implications. In particular, we point out problems with Taguchi's analysis strategy which is based on the latter rationale. We propose an alternative analysis method and demonstrate it by reanalyzing three documented experiments; the reanalysis provides new insight about their respective processes. Finally, using the bad region avoidance rationale, we discuss alternative experimental designs.
Additional information
Notes on contributors
M. Hamada
M. Hamada is a Research Associate Professor in the Department of Statistics and Actuarial Science, and The Institute for Improvement in Quality and Productivity. He is a Member of ASQC.
C. F. J. Wu
C. F. J. Wu is a Professor in the Departments of Statistics, and of Industrial and Operations Engineering. He is a Senior Member of ASQC.