A Systematic Approach to Planning for a Designed Industrial

100 votes to spread among the interactions the

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 00 votes to spread among the interactions. The remaining items are found on the Master Guide Sheet. 8. RESTRICTIONS, PREFERENCES FOR THE DESIGN, ANALYSIS, AND PRESENTATION Box, Hunter, and others have repeatedly exhorted, “Attention to detail can determine the success or failure of the experiment.” Item 8 in Figure 2 is part of heeding that advice. Theoretical optimal experimental design and pracfical experimental design are often worlds apart, and restrictions often make the difference. Since a single unknown restriction can render worthless an otherwise well-considered, laboriously developed design. the statistician should encourage experimenters to be quick to put these limitations and pitfalls on the table. In particular, there appears to be a lack of awareness in the applied statistics community of the prevalence of experiments with unidentified split-plot structure. Because it is unidentified (different experimental units used for different parts of the experiment), the analysis is often done incorrectly-using the wrong error terms to test statistical significance. The discussion of the issues surrounding the choice of experimental unit and analysis strategies goes beyond the scope of this article but should take place on the experimentation team. Items 10 and 11 of Figure 2 are intended for the following three circumstances: First. when expcrimcnters are statistically sophisticated and have a good idea of appropriate designs or analysis techniques; PLANNING FOR A DESIGNED INDUSTRIAL 11 EXPERIMENT Interactions (Tutorial) Taylor series approximation: f(x,y) = ao + alx + bly + CIIXY + a2x2 Plot I la A*B R 1 NO Interaction Low Plot High A Table lb - A.8 Interaction LOW A :Eje :I (tutorial) THE NEXT STAGE By the time the experimentation team has come to a consensus concerning the information collected in items l-10 of the guide sheet, the statistician (or surrogate) will have had the opportunity to step beyond the generic confines of the guide sheet and discuss more problem-specific issues that will affect the experimental design. such as multilev...
View Full Document

Ask a homework question - tutors are online