Design of Experiments with Minitab

Citation

Narasimhan, K. (2005), "Design of Experiments with Minitab", The TQM Magazine, Vol. 17 No. 2, pp. 208-209. https://doi.org/10.1108/09544780510583263

Paul Mathews has nearly 20 years experience 12 years experience in industry, five years as a an academic and six years in quality engineering and applied statistical training and consultancy for a wide range of organizations. He is a Certified American Society for Quality Certified Quality engineer and a Certified Six Sigma Black Belt. He describes himself as a “physicist by education, engineer by experience and a statistician out of necessity”.

This book is a practical introduction to design of experiments (DOE) using Minitab software and includes a CD‐ROM that contains Excel files with the data from many examples provided in the book. It comprises 11 chapters, nine appendices and a short bibliography.

Chapter 1 deals with basics of presenting data and briefly explains how to use Minitab software. Chapters 2 and 3 introduce respectively descriptive and inferential statistics that are important to methods of designing experiments and analyzing the results. Measures of location, variation and the basics of the normal distribution are covered in Chapter 2. Chapter 3 focuses on hypothesis tests and confidence intervals.

The basic language and concepts required to understand all aspects of DOE are presented in Chapter 4. The concepts introduced here include variables and responses, types of experiments, models, and designs. An 11‐step procedure for planning, executing, analyzing, and reporting an experiment is described in detail.

The following two chapters deal respectively with experiments involving single or multiple variables using analysis of variance (ANOVA). First, the theory is introduced and then the design considerations are explained. A number of tables and graphs are used to support the explanations. Advanced ANOVA topics dealing with mixed (fixed and random) variables, nested variables are covered in the following chapter, which primarily covers qualitative experimental variables.

The theme of Chapter 8 is linear regression. The rationale underpinning linear regression is explained first. This is followed by a detailed explanation of the regression coefficients, the satisfying conditions required, confidence limits for the regression line and correlation coefficient, and goodness of fit tests. Multiple regression and general linear models are also briefly covered.

The next two chapters focus respectively on two‐level full factorial and fractional factorial experiment designs. In Chapter 9, first, 2¹ factorial experiments are explained with the aid of an example. This is followed by explanations of 2² and 2³ factorial designs before extending the procedures to analyses of 2^k designs. The application of Minitab for creating and analyzing 2^k designs is included. Chapter 10 explains how to cut back on the size of experiments as k gets very large, using fractional factorial designs, and how to interpret the results obtained by using the Minitab software. The final chapter deals with designs for quadratic models or response‐surface designs.

Two valuable features that standout are the numerous worked examples that support the chapters and detailed instructions and examples given for calculating sample sizes for most common DOE problems. The topic of sample size is especially useful considering the cost implications of using wrong sample sizes. The accompanying CD‐ROM contains descriptions of simple experiments, involving magic dice, paper helicopters, catapults, etc., that could be conducted at home or in a DOE class. The CD‐ROM also contains Minitab macros for analyzing various designs covered in the earlier chapters. It is a really valuable book for those that are interested in designing experiments and analyzing them using Minitab.