Testing in Industrial Experiments with Ordered Categorical Data

Abstract

The resurgence in the use of experimental design methods in industrial applications in the past few years is primarily due to the pioneering work of Professor G. Taguchi. Traditional experimental design techniques focus on identifying factors that affect the level of a production or manufacturing process. We call this the location effects of the factors. Taguchi was the first to recognize that statistically planned experiments could and should be used in the product development stage to detect factors that affect the variability of the output. This will be termed the dispersion effects of the factors. By setting the factors with important dispersion effects at their “optimal” levels, the output can be made robust to changes in operating and environmental conditions in the production line. Thus the identification of dispersion effects is also crucial in improving the quality of a process.