Reviewed by Ken Deal
Marketing Research, Winter 2005
Most readers will say, "It's about time!" Conjoint analysis has been around since Duncan Luce and John Tukey’s work in the 1960s and yet there hasn’t been a good, simple, readable, and informative book introducing the topic until now.
The 11 chapters in Bryan Orme’s Getting Started with Conjoint Analysis are worthwhile reading for anyone interested in this topic. It’s difficult to highlight special areas because nothing should be ignored. Even those who’ve conducted several conjoint analysis studies will find that something Orme restates clears up misunderstandings; the writing is always direct, understandable, and instructive.
Chapter 1 motivates managers to consider conjoint analysis as a viable marketing research methodology, and helps those new to conjoint analysis develop their speech patterns for talking about it. Then Chapter 2 lays out a simple conjoint analysis exercise, and bolsters confidence by having readers complete worksheets with calculations for part-worth utility scores and importances, and graph those values in charts.
Chapter 3 concisely explains conjoint analysis for managers, and makes sure that they don’t misinterpret fundamental parts of it (e.g., simulating shares of preference and not market share). Those who’ve had to convince clients that conjoint analysis doesn’t produce market shares directly—after a previous supplier mislabeled preference shares as market shares—will dog-ear section 3.4 for clients the next time it happens. Chapter 4 gives an excellent historical presentation of conjoint analysis that not only is interesting, but also helps readers put the methodology in perspective in later chapters.
Skipping through this book greatly reduces its value. For example, in his discussion of adaptive conjoint analysis, Orme slips in: “Dropping the importance questions from ACA [adaptive conjoint analysis] surveys may result in better market share predictions and greater discrimination among attributes (as long as hierarchical Bayes is used to estimate the part-worths).” Those new to conjoint analysis might end up spending a bit of time to discover that elsewhere.
Sample-size issues are not dealt with in many conjoint analysis articles, and Chapter 7 gives a valuable primer in that topic. Plus, Orme provides excellent answers to one of those questions that suppliers often hear: “How large a sample size do I need?” The nine questions in response are worth learning. He also offers helpful hints that would be hard to find in other sources—for example, Sawtooth Software founder Rich Johnson’s rule of thumb on minimum sample size for aggregate level, full-profile, choice-based conjoint studies.
Chapter 8 presents another confidence-building exercise in which the readers analyze a small, traditional conjoint analysis problem using multiple regression in Excel. Then Chapter 9 focuses on the interpretation of conjoint analysis findings and includes an important section on price elasticity, price sensitivity, and willingness to pay (Orme cites Johnson’s contributions.) The potentially misleading nature of converting differences between attribute levels to monetary values is the focus of a detailed discussion, followed by the recommended procedure of basing strategic decisions on the results of simulations. Orme introduces topics such as hierarchical Bayesian analysis and latent class analysis, and gradually educates readers in their use. He also provides references for further investigation.
Although the whole book is greatly rewarding to readers, the glossary is pure gold. This 52-page compendium of conjoint analysis terminology will be Orme’s legacy in this field. Undoubtedly, many professors will pass it along to their students and many consultants will furnish it to their clients. He not only explains conjoint analysis terms in the glossary, but also illustrates many with examples and references. Orme supplies tips for better practice, which beginners can use to quickly leverage their strengths. My main suggestion for improving this book is for Orme to reference the glossary each time a term is introduced in the body of the book.
Bryan Orme certainly will be widely cited for this effort. And what better recommendation can any marketing research book receive than a foreword by Paul Green, emeritus professor of marketing? Well done!
Reprinted with permission from Marketing Research, published by the American Marketing Association, Ken Deal/Editor Chuck Chakrapani, Winter 2005, vol. 17 No. 4.