A0964
Title: Cause-and-effect diagram-based supersaturated designs
Authors: Chang-Yun Lin - National Chung Hsing University (Taiwan) [presenting]
Abstract: Supersaturated designs (SSDs) are often used for screening experiments, and cause-and-effect diagrams (CEDs) are useful quality tools that help engineers decide which factors to use in experiments. Based on engineers prior knowledge, certain factors (referred to as the primary factors) in the cause-and-effect diagram are considered more likely to be active than the others (referred to as the potential factors). Due to the unequal importance of the factors, the traditional E(s2) criterion is unsuitable for selecting supersaturated designs. A CED-based approach is proposed to select supersaturated designs that have smaller variances in the estimates of the primary factor effects and less severe aliasing between the effects of the primary and potential factors. Simulation studies show that the proposed method selects supersaturated designs that outperform other supersaturated designs in terms of power and type I error when engineers have accurate prior knowledge.
