Making Hard Decisions - dandelon.com · Strenlar 148 Job Offers 150 SS Kuniang, Part II 152 References i52 Epilogue 154 Chapter 5 Sensitivity Analysis 155 Eagle Airlines 155 Sensitivity

MakingHard Decisions

An Introduction to Decision Analysis

2nd Edition

Robert T. ClemenFuqua School of Business

Duke University

Duxbury PressAn Imprint of Brooks/Cole Publishing Company

I (T) P® An International Thomson Publishing Company

Pacific Grove • Albany • Belmont • Bonn • Boston • Cincinnati • DetroitJohannesburg • London • Madrid • Melbourne • Mexico City

New York • Paris • Singapore • Tokyo • Toronto • Washington

Contents

Preface xix

Chapter 1 Introduction to Decision Analysis 1Gypsy Moths and the ODA 1

Why Are Decisions Hard? 2

Why Study Decision Analysis? 3

Subjective Judgments and Decision Making 5

The Decision-Analysis Process 5

Requisite Decision Models 8

Where Is Decision Analysis Used? 8

Where Are We Going from Here? 9

Summary 10

Questions and Problems 10

Case Studies: Dr. Joycelyn Elders and the War on Drugs 11

Lloyd Bentsen for Vice President? 12

Du Pont and Chlorofluorocarbons 13

References 14Epilogue 15

CONTENTS

Section 1 Modeling Decisions 17

Chapter 2 Elements of Decision Problems 19Values and Objectives 19

Making Money: A Special Objective 20

Values and the Current Decision Context 21

Boeing's Supercomputer 22

Decisions to Make 23

Sequential Decisions 24

Uncertain Events 25

Consequences 27

The Time Value of Money: A Special Kind of Trade-Off 28

Larkin Oil 31

Summary 34

Questions and Problems 34

Case Studies: The Value of Patience 36

Early Bird, Inc. 37

References 39Epilogue 39

Chapter 3 Structuring Decisions 41Structuring Values 42

Hiring a Summer Intern 42

Fundamental and Means Objectives 44

Getting the Decision Context Right 49

Structuring Decisions: Influence Diagrams 50

Influence Diagrams and the Fundamental Objectives Hierarchy 52

Using Arcs to Represent Relationships 53

Some Basic Influence Diagrams 55

The Basic Risky Decision 55 Imperfect Information 56

Sequential Decisions 59 Intermediate Calculations 61

Constructing an Influence Diagram (Optional) 63

Toxic Chemicals and the EPA 63

Some Common Mistakes 65 Multiple Representations and

Requisite Models 67

CONTENTS

Structuring Decisions: Decision Trees 67

Decision Trees and the Objectives Hierarchy 69

Some Basic Decision Trees 70

The Basic Risky Decision 70 Imperfect Information 72

Sequential Decisions 72

Decision Trees and Influence Diagrams Compared 74

Decision Details: Defining Elements of the Decision 74

More Decision Details: Cash Flows and Probabilities 76

Defining Measurement Scales for Fundamental Objectives 77

Using Computers for Structuring Decisions 82

Logical Decisions 83 DPL (Decision Programming Language) 85

DATA 88

Summary 89

Exercises 90

Questions and Problems 92

Case Studies: Cold Fusion 96

Prescribed Fire 97

The SS Kuniang 97

References 98Epilogue 100

Chapter 4 Making Choices 101

Texaco Versus Pennzoil 101

Decision Trees and Expected Monetary Value 105

Solving Influence Diagrams: Overview 109

Solving Influence Diagrams: The Details (Optional) 112

Solving Influence Diagrams: An Algorithm (Optional) 117

Risk Profiles 118

Cumulative Risk Profiles 122

Dominance: An Alternative to EMV 123

Making Decisions with Multiple Objectives 127

The Summer Job 128

Analysis: One Objective at a Time 130

Subjective Ratings for Constructed Attribute Scales 130

• Assessing Trade-Off Weights 132

Analysis: Expected Values and Risk Profiles for Two Objectives 133

CONTENTS

Computer Programs for Decision Analysis 136

DPL 136 DATA 137 Electronic Spreadsheets 140

Summary 140

Exercises 141

Questions and Problems 142

Case Studies: GPC's New Product Decision 145

Southern Electronics, Part I 146

Southern Electronics, Part II 147

Strenlar 148

Job Offers 150

SS Kuniang, Part II 152

References i52Epilogue 154

Chapter 5 Sensitivity Analysis 155Eagle Airlines 155

Sensitivity Analysis: A Modeling Approach 156

Problem Identification and Structure 157

One-Way Sensitivity Analysis 160

Tornado Diagrams 161

Dominance Considerations 162

Two-Way Sensitivity Analysis 164

Sensitivity to Probabilities 165

Two-Way Sensitivity Analysis for Three Alternatives

(Optional) 169

Investing in the Stock Market 169

Sensitivity Analysis in Action 172

Heart Disease in Infants 173

Sensitivity Analysis by Computer 174

Sensitivity Analysis with DPL 174 Sensitivity Analysis

with DATA 176

Sensitivity Analysis: A Built-in Irony 178

Summary 178

Exercises 178

Questions and Problems 179

CONTENTS

Case Studies: The Stars and Stripes 181

Dumond International, Part I 182

Strenlar, Part II 183

Facilities Investment and Expansion 184

Job Offers, Part II 184

References 186Epilogue 186

Chapter 6 Creativity and Decision Making 187What Is Creativity? 188

Theories of Creativity 189

Chains of Thought 189

Phases of the Creative Process 190

. Blocks to Creativity 192

Framing and Perceptual Blocks 192

The Monk and the Mountain 192

Making Cigars 193

Value-Based Blocks 195 Cultural and Environmental Blocks 197

Ping-Pong Ball in a Pipe 197

Organizational Issues 199

Value-Focused Thinking for Creating Alternatives 200

Fundamental Objectives 200 Means Objectives 200

Transportation of Nuclear Waste 201

The Decision Context 202

Other Creativity Techniques 203

Fluent and Flexible Thinking 203 Idea Checklists 203

Brainstorming 206 Metaphorical Thinking 206

Other Techniques 208

Creating Decision Opportunities 209

Summary 209

Questions and Problems 210

Case Studies: Modular Olympics 211

Burning Grass-Seed Fields 212

References 212Epilogue 214

CONTENTS

Section 2 Modeling Uncertainty 217

Chapter 7 Probability Basics 219A Little Probability Theory 220

Venn Diagrams 220

More Probability Formulas 221

Uncertain Quantities 226

Discrete Probability Distributions 227 Expected Value 229

Variance and Standard Deviation 230 Covariance and Correlation for

Measuring Dependence (Optional) 232 Continuous Probability

Distributions 236 Stochastic Dominance Revisited 237 Stochastic

Dominance and Multiple Attributes (Optional) 238 Probabilities Density

Functions 239 Expected Value, Variance, and Standard Deviation: The

Continuous Case 240 Covariance and Correlation: The Continuous

Case (Optional) 241

Oil Wildcatting 242

John Hinckley's Trial 248

Decision-Analysis Software and Bayes' Theorem 250

Summary 250

Exercises 251

Questions and Problems 255

Case Studies: Decision Analysis Monthly 257

Screening for Colorectal Cancer 258

AIDS 259

Discrimination and the Death Penalty 262

References 264Epilogue 264

Chapter 8 Subjective Probability 265Uncertainty and Public Policy 265

Probability: A Subjective Interpretation 267

Accounting for Contingent Losses 268

Assessing Discrete Probabilities 269

Assessing Continuous Probabilities 274

Pitfalls: Heuristics and Biases 281

CONTENTS

TomW. 281

Representativeness 282 Availability 283 Anchoring and Adjusting 284

Motivational Bias 284 Heuristics and Biases: Implications 284

Decomposition and Probability Assessment 285

Experts and Probability Assessment: Pulling It All Together 291

Climate Change at Yucca Mountain, Nevada 294

Coherence and the Dutch Book (Optional) 296

Summary 298

Exercises 299

Questions and Problems 300

Case Studies: Assessing Cancer Risk—From Mouse to Man 305

Breast Implants 307

The Space Shuttle Challenger 308

References 310Epilogue 313

Chapter 9 Theoretical Probability Models 314Theoretical Models Applied 315

The Binomial Distribution 316

The Poisson Distribution 320

The Exponential Distribution 323

The Normal Distribution 325

The Beta Distribution 331

Probability Distributions and Decision-Analysis Software 335

Summary 335

Exercises 335

Questions and Problems 337

Case Studies: Overbooking 345

Earthquake Prediction 346

Municipal Solid Waste 350

References 353Epilogue 354

Chapter 10 Using Data 355Using Data to Construct Probability Distributions 355

Histograms 356 Empirical CDFs 357

xiv CONTENTS

Halfway Houses 357

Using Data to Fit Theoretical Probability Models 361

Software for Fitting Distributions: BestFit 362

Using Data to Model Relationships 363

The Regression Approach 365

Estimation: The Basics 368 Estimation: More than One Conditioning

Variable 375 Regression Analysis and Modeling: Some Do's and Don't 's 378

Regression Analysis: Some Bells and Whistles 383 Regression Modeling:

Decision Analysis Versus Statistical Inference 386 An Admonition: Use with

Care 387

Natural Conjugate Distributions (Optional) 387

Uncertainty About Parameters and Bayesian Updating 388 Binomial

Distributions: Natural Conjugate Priors for p 390 Normal Distribution:

Natural Conjugate Priors for I* 392 Predictive Distributions 394

Predictive Distributions: The Normal Case 395 Predictive Distributions: The

Binomial Case 396

A Bayesian Approach to Regression Analysis (Optional) 397

Summary 397

Exercises 398

Questions and Problems 398

Case Studies: Taco Shells 405

Forecasting Sales 406

Overbooking, Part II 408References 408

Chapter 11 Monte Carlo Simulation 410Fashions 411

Using Uniform Random Numbers as Building Blocks 414

General Uniform Distributions 415

Exponential Distributions 416

Discrete Distributions 417

Other Distributions 417

Computer Software for Simulation 417

Crystal Ball 418 @ RISK 420

Simulation and Sensitivity Analysis 422

Distributions on Parameters (Optional) 423

CONTENTS

Simulation, Decision Trees, and Influence Diagrams 424

Summary 426

Exercises 426

Questions and Problems 426

Case Studies: Choosing a Manufacturing Process 428

Organic Farming 430

Overbooking, Part III 433References 433

Chapter 12 Value of Information 435Investing in the Stock Market 435

Value of Information: Some Basic Ideas 436

Probability and Perfect Information 436 The Expected Value

of Information 438

Expected Value of Perfect Information 439

Expected Value of Imperfect Information 441

Value of Information in Complex Problems 447

Value of Information, Sensitivity Analysis, and Structuring 448

Seeding Hurricanes 448

Value of Information and Nonmonetary Objectives 449

Value of Information and Experts 450

Calculating EVPI: DATA and DPL 451

Summary 451

Exercises 452

Questions and Problems 453

Case Studies: Texaco-Pennzoil Revisited 455

Medical Tests 456

Dumond International, Part II 456References 457

1 Section 3 Modeling Preferences 459

Chapter 13 Risk Attitudes 461E. H. Harriman Fights for the Northern Pacific Railroad 462

Risk 463

xvi CONTENTS

Risk Attitudes 465

Investing in the Stock Market, Revisited 467

Expected Utility, Certainty Equivalents, and Risk Premiums 469

Keeping Terms Straight 473

Utility Function Assessment 473

Assessment Using Certainty Equivalents 474 Assessment Using

Probabilities 475 Gambles, Lotteries, and Investments 411

Risk Tolerance and the Exponential Utility Function 477

Risk Tolerance and Sensitivity Analysis: Eagle Airlines, Revisited 480

Decreasing and Constant Risk Aversion (Optional) 482

Decreasing Risk Aversion 483 An Investment Example 484 Constant Risk

Aversion 485

Some Caveats 488

Summary 488

Exercises 489

Questions and Problems 490

Case Studies: Interplants, Inc. 497

Texaco-Pennzoil One More Time 499

Strenlar, Part III 501

References 501Epilogue 502

Chapter 14 Utility Axioms, Paradoxes, and Implications 503

Preparing for an Influenza Outbreak 503

Axioms for Expected Utility 504

Paradoxes 510

Implications 514

Implications for Utility Assessment 514 Managerial and Policy

Implications 516

A Final Perspective 518

Summary 519

Exercises 520

Questions and Problems 521

Case Studies: The Life Insurance Game 524

Nuclear Power Paranoia 525

The Manager's Perspective 526

CONTENTS xvii

References 527Epilogue 529

Chapter 15 Conflicting Objectives I: Fundamental Objectives and theAdditive Utility Function 530Objectives and Attributes 532

Trading Off Conflicting Objectives: The Basics 534

Choosing an Automobile: An Example 534

The Additive Utility Function 536

Choosing an Automobile: Proportional Scores 538 Assessing Weights:

Pricing Out the Objectives 539 Indifference Curves 540

Assessing Individual Utility Functions 542

Proportional Scores 542 Ratios 544 Standard Utility-Function

Assessment 546

Assessing Weights 546

Pricing Out 547 Swing Weighting 547 Lottery Weights 550

Keeping Concepts Straight: Certainty Versus Uncertainty 552

An Example: Library Choices 553

The Eugene Public Library 553

Using Software for Multiple-Objective Decisions 560

Summary 560

Exercises 561

Questions and Problems 562

Case Studies: The Satanic Verses 569

Dilemmas in Medicine 570

A Matter of Ethics 572

FDA and the Testing of Experimental Drugs 573References 574Epilogue 575

Chapter 16 Conflicting Objectives II: Multiattribute Utility Models withInteractions 576Multiattribute Utility Functions: Direct Assessment 577

Independence Conditions 579

Preferential Independence 579 Utility Independence 580

Determining Whether Independence Exists 580

xviii CONTENTS

Using Independence 582

Additive Independence 583

Substitutes and Complements 586

Assessing A Two-Attribute Utility Function 586

The Blood Bank 587

Three or More Attributes (Optional) 591

When Independence Fails 592

Multiattribute Utility in Action: BC Hydro 593

Strategic Decisions at BC Hydro 593

Summary 598

Exercises 599

Questions and Problems 600

Case Studies: A Mining-Investment Decision 603

References 605Epilogue 606

Chapter 17 Conclusion and Further Reading 607A Decision-Analysis Reading List 608

Appendixes 613A Binomial Distribution: Individual Probabilities 614

B Binomial Distribution: Cumulative Probabilities 622

C Poisson Distribution: Individual Probabilities 630

D Poisson Distribution: Cumulative Probabilities 635

E Normal Distribution: Cumulative Probabilities 640

F Beta Distribution: Cumulative Probabilities 644

Answers to Selected Exercises 653Author Index 654Subject Index 657