Experimental Design

by Berger, Paul D.; Maurer, Robert E., Ph.D.

• ISBN13: 9780534358228
• ISBN10: 0534358225
• Format: Hardcover
• Copyright: 2001-08-01
• Publisher: Cengage Learning

Summary

Based on decades of teaching, consulting, and industrial experience in the field of design and analysis of experiments, the authors provide an intuitive understanding of the principles of experimental design and analysis. The emphasis is on the application of experimental design concepts in such traditional management and industrial engineering areas such as marketing, operations, management information systems, organizational behavior, and others. The authors also apply this material to such non-profit areas as education, health care, and government. Using popular analytical tools such as SPSS, JMP, and Microsoft Excel, Berger and Maurer emphasize the modern application of experimental design to real problems.

Author Biography

Dr. Paul D. Berger, Professor of Marketing and Quantitative Methods at the School of Management, Boston University Dr. Robert E. Maurer has more than 35 years of industrial experience at Bell Telephone Laboratories. Dr. Maurer has more than 20 years of experience teaching in the areas of statistical communication theory at the graduate school of engineering at Northeastern University and statistical analysis at the school of management at Boston University

Table of Contents

Introduction to Experimental Design	1	(18)
What Is Experimentation?	1	(1)
The Growth in Experimental Design	2	(1)
The Six Steps of Experimental Design	3	(6)
Plan the Experiment	3	(2)
Design the Experiment	5	(2)
Perform the Experiment	7	(1)
Analyze the Data from the Experiment	7	(1)
Confirm the Results of the Experiment	8	(1)
Evaluate the Conclusions of the Experiment	8	(1)
Experimental-Design Applications in Management	9	(8)
Corporate Environmental Behavior	9	(2)
Supermarket Decision Variables	11	(1)
Financial Services Menu	12	(1)
The Qualities of a Superior Motel	13	(2)
Time and Ease of Seatbelt Use: A Public Sector Example	15	(1)
Emergency Assistance Service for Travelers	15	(2)
Perspective	17	(2)
PART ONE Primary Focus on Factors Under Study	19	(226)
One-Factor Designs and the Analysis of Variance	21	(38)
One-Factor Designs	22	(9)
The Statistical Model	25	(1)
Estimation of the Parameters of the Model	26	(2)
Sums of Squares	28	(3)
Analysis of (the) Variance (ANOVA)	31	(4)
Forming the F Statistic: Logic and Derivation	35	(13)
The Key Fifth Column of the ANOVA Table	35	(13)
A Comment	48	(11)
Exercises	49	(10)
Some Further Issues in One-Factor Designs and ANOVA	59	(28)
Introduction	59	(1)
Basic Assumptions of ANOVA	59	(4)
Kruskal-Wallis Test	63	(3)
Review of Hypothesis Testing	66	(10)
p Value	70	(1)
Type I and Type II Errors	71	(2)
Back to ANOVA	73	(3)
Power
Power Considerations in Determination of Required Sample Size	76	(6)
Confidence Intervals	82	(5)
Exercises	84	(3)
Multiple-Comparison Testing	87	(42)
Logic of Multiple-Comparison Testing	88	(2)
Type I Errors in Multiple-Comparison Testing	90	(3)
Pairwise Comparisons	93	(15)
Fisher's Least Significant Difference Test	93	(6)
Tukey's Honestly Significant Difference Test	99	(4)
Newman-Keuls Test with Example	103	(2)
Two Other Tests Comparing All Pairs of Column Means	105	(1)
Dunnett's Test	106	(2)
Post Hoc Exploratory Comparisons---The Scheffe Test	108	(21)
Carrying Out the Test	109	(2)
Discussion of Scheffe Test	111	(14)
Exercises	125	(4)
Orthogonality, Orthogonal Decomposition, and Their Role in Modern Experimental Design	129	(26)
Introduction	130	(1)
Forming an Orthogonal Matrix	131	(24)
Exercises	151	(4)
Two-Factor Cross-Classification Designs	155	(46)
Introduction to Studying Two Factors	156	(1)
Designs with Replication	157	(20)
The Model	159	(1)
Parameter Estimates	159	(1)
Interaction	160	(4)
Back to the Statistical Model: Sums of Squares	164	(13)
Fixed Levels versus Random Levels	177	(5)
Two Factors with No Replication and No Interaction	182	(5)
Blocking	187	(1)
Friedman Nonparametric Test	188	(13)
Perspective on Friedman Test	191	(1)
Exercises	192	(9)
Nested, or Hierarchical, Designs	201	(20)
Introduction to Nested Designs	202	(2)
The Model	204	(9)
Discussion	213	(8)
Exercises	214	(7)
Designs with Three or More Factors: Latin-Square and Related Designs	221	(24)
Introduction to Multifactor Designs	222	(1)
Latin-Square Designs	223	(11)
The Latin-Square Model and ANOVA	227	(7)
Graeco-Latin-Square Designs	234	(3)
Other Designs with Three or More Factors	237	(8)
Exercises	240	(5)
PART TWO Primary Focus on the Number of Levels of a Factor	245	(166)
Two-Level Factorial Designs	247	(42)
Introduction	248	(1)
Two-Factor Experiments	249	(3)
Remarks on Effects and Interactions	252	(1)
Symbolism, Notation, and Language	252	(1)
Table of Signs	253	(4)
Modern Notation and Yates' Order	257	(1)
Three Factors, Each at Two Levels	258	(6)
Estimating Effects in Three-Factor, Two-Level Designs	259	(5)
Number and Kinds of Effects	264	(2)
Yates' Forward Algorithm	266	(2)
A Note on Replicated 2k Experiments	268	(3)
Main Effects in the Face of Large Interactions	271	(2)
Levels of Factors	273	(1)
Factorial Designs versus Designs Varying Factors One at a Time	274	(3)
Factors Not Studied	277	(1)
Errors of Estimates in 2k Designs	278	(2)
Comment on Testing the Effects in 2k Designs	280	(9)
Exercises	281	(8)
Confounding/Blocking in 2k Designs	289	(22)
Introduction	290	(1)
Simple Confounding	290	(5)
Partial Confounding	295	(3)
Multiple Confounding	298	(1)
Mod-2 Multiplication	299	(2)
Determining the Blocks	301	(3)
Number of Blocks and Confounded Effects	304	(2)
Comment on Calculating Effects	306	(5)
Appendix: Detailed Example of Error Reduction through Confounding	307	(1)
Exercises	308	(3)
Two-Level Fractional-Factorial Designs	311	(47)
Introduction	313	(1)
2k--p Designs	314	(10)
Yates' Algorithm Revisited	324	(3)
Quarter-Replicate Designs	327	(4)
Orthogonality Revisited	331	(14)
Power and Minimum Detectable Effects in 2k--p Designs	345	(13)
Appendix: Selection of a Workable Set of Dead Letters	353	(1)
Exercises	354	(4)
Designs with Factors at Three Levels	358	(20)
Introduction	359	(1)
Design with One Factor at Three Levels	359	(2)
Design with Two Factors, Each at Three Levels	361	(10)
Nonlinearity Recognition and Robustness	371	(1)
Three Levels versus Two Levels	372	(2)
Unequally Spaced Levels	374	(1)
Comment	375	(3)
Exercises	375	(3)
Introduction to Taguchi Methods	378	(33)
Introduction	379	(1)
Taguchi's Quality Philosophy and Loss Function	379	(3)
Control of the Variability of Performance	382	(2)
Taguchi Methods: Designing Fractional-Factorial Designs	384	(10)
Experiments without Interactions	386	(2)
Experiments with Interactions	388	(6)
Taguchi's L16	394	(1)
Experiments Involving Nonlinearities or Factors with Three Levels	394	(6)
Further Analyses	400	(6)
Perspective on Taguchi Methods	406	(5)
Exercises	407	(4)
PART THREE Response-Surface Methods, Other Topics, and the Literature of Experimental Design	411	(56)
Introduction to Response-Surface Methodology	413	(41)
Introduction	413	(3)
The Underlying Philosophy of RSM	416	(2)
Method of Steepest Ascent	418	(9)
Testing the Plane: Center Points	424	(3)
Method of Local Exploration	427	(5)
Central-Composite Designs	427	(2)
Box-Behnken Designs	429	(1)
Comparison of Central-Composite and Box-Behnken Designs	430	(1)
Issues in the Method of Local Experimentation	431	(1)
Perspective on RSM	432	(19)
Concluding Remark	451	(3)
Exercises	451	(3)
Literature on Experimental Design, and Discussion of Some Topics Not Covered in the Text	454	(13)
Introduction	454	(1)
Literature Discussion	455	(4)
Some Classics	455	(1)
Recommendations for Specific Topics	456	(3)
Discussion of Some Topics Not Covered in the Text	459	(4)
Outliers	459	(1)
Missing Data	460	(1)
Power and Sample Size	460	(1)
Time-Series and Failure-Time Experiments	460	(1)
Plackett-Burman Designs	461	(1)
Repeated-Measures Designs	461	(1)
Crossover Designs	462	(1)
Mixture Designs	462	(1)
Bibliography	463	(1)
References	463	(4)
Statistical Tables	467	(10)
Index	477