the delphi method: techniques and applications

Book Re vie w 441 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLK Linstone, Harold A., and Murray Turoff, zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPO The Delphi M ethod: Techniques and Applications, Addison-Wesley. Advanced Book Program, Reading, MA, 1975, xx, 620 pp, $29.50 (hardbound), $16.50 (paperbound). The first thing to clarify about this book is what it is not. It is not an elementary introduction to Delphi. It is not for the neophyte, who is looking for help in conducting his first Delphi. He could read through the entire book and glean only some dim idea of the mechanics of Delphi. The book presumes a great deal of prior knowledge about Delphi on the part of the reader. For the practitioner, who has already obtained a working acquaintance with Delphi, what does the book offer? In short, a gold-mine of up-to-date information on the theory and application of Delphi. The editors specifically decided against reprinting a collection of papers. They included only four which had been printed earlier elsewhere. Instead, they obtained a total of 22 original papers for the book, not counting seven summary papers they wrote themselves to introduce the various sections of the book. It is impossible, in a short review, to cover in detail the wealth of information in this book. A few samples will have to suffice to indicate the type of material it contains. The first major section of the book deals with the philosophy of Delphi. The opening paper, by Mitroff and Turoff, discusses the five basic Inquiring Systems of Western philosophy (Leibnizian, Lockean, Kantian, Hegelian, and Singerian), and places Delphi in the context of these Inquiring Systems. The authors point out that the original use of Delphi involved an almost purely Lockean Inquirer, in which there was no formal model at all but only the data input from the Delphi respondents. Later Delphis represented a more Kantian type of Inquirer, in which the objective is to develop a structure as well as to collect data. Hegelian, or conflictual, Inquirers have been almost absent from the use of Delphi, with a few exceptions noted by the authors, in which there was a deliberate attempt to elicit contradictory views. Singerian Inquirers have been “notably absent” from Delphi uses, hence the authors can give no examples. The authors also point out that Delphi is not a Leibnizian Inquirer. Moreover, they point out that most of the attacks on Delphi, especially those by social scientists, are launched from a Leibnizian viewpoint, which criticizes Delphi for not being Leibnizian, and refuses to recognize that there are any legitimate Inquiring Systems other than the Leibnizian. The next section, which is by far the largest in the book (153 pages), deals with applications. Here we find descriptions of Delphi used to identify the pros and cons of policies, to develop resources management programs, to tap expertise in the field of drug abuse, and to make forecasts in areas of technology such as aeronautics, communications, plastics and steel. One interesting application involved the use of Delphi to evaluate the potential contributions of military laboratories to a set of aeronautical weapon systems. The same Delphi questionnaire was used in three different laboratories, each involved in a different area of technology, and in a group responsible for planning aeronautical system programs. Among the interesting findings were that laboratories did not tend to be biased in favor of systems which depended upon their own areas of technology; that the systems planners tended to take a more future-oriented view than did the laboratories; and that the systems planners tended to take a different view of the relative importance of certain systems than did the laboratories as a whole, indicating a possible communications gap between planners and laboratories. The author points out that this last point might not have been recognized had the answers from all four groups of experts been merged in one consolidated Delphi. 442 BOOK REVIEW The next section deals with the evaluation of Delphi. Papers cover such topics as the theory of group estimation, the effect of feedback between rounds, and comparisons with Delphi and face-to-face groups. One interesting paper deals with a correlation between Delphi respondents’ propensity to change opinions between rounds and the respondents’ scores on a test of Dogmatism. The original hypothesis was that the high-dogmatism respondents would be those unwilling to change their opinions. The surprising result was that the high-dogmatism respondents showed a slightly stronger tendency to change opinions than did low-dogmatism respondents. Cross impact analysis is the topic of the next section of the book. Although cross impact analysis is not limited to use with Delphi, it was originally devised to overcome the problem that Delphi respondents faced in considering the interactions among several forecasts. The three papers in this section are all reprints, and present Dalkey’s and Turoff’s contrasting views on the mathematical theory behind cross impact, as well as Kane’s computer language KSIM, which simplifies the computerization of cross impact analysis. The next section of the book is the second longest (102 pages), and deals with specialized techniques. In a sense this section is a grab-bag of techniques which are related to, or involve, Delphi. Two especially interesting ones involve the use of Cluster Analysis to group large sets of items into smaller sets to make the task of the Delphi respondents easier, and the use of multidimensional scaling to identify similarities among respondents. The use of computers with Delphi is taken up in the next section, which contains several papers on various aspects of computerized on-line conferencing. The final paper in this section is the fourth reprint in the book, Turoff’s “Meeting of the Council on Cybernetic Stability: A Scenario,” which portrays a world in which a computerized Leibnizian Inquirer has succeeded in abolishing all other philosophical approaches to knowledge. The final section of the book contains a single paper on pitfalls of Delphi. The paper presents a checklist, to help the user of Delphi avoid many of the mistakes which have marred the design and execution of Delphi exercises in the past. In addition to the wide range of papers about Delphi, some of which are mentioned above, the book also contains an excellent bibliography. There are 489 references to items about Delphi, and an additional 181 references to related work, including documents on communication in small groups, on subjective probability, and on scaling of subjective data. In particular, the bibliography identifies 48 relevant papers from the RAND Corporation, 80 from the Institute for the Future and The Futures Group, and 45 foreign language articles. Not only does the book provide wide-ranging coverage of Delphi, it avoids one of the common problems of collections of papers. The editors have taken care to see that the articles fit together, that overlap is minimized, and that references from one article to another are made when these are needed. This is, of course, a great deal easier when most zyxwvuts of the articles are written especially for a book, as these were. Nevertheless, the reader can be grateful for the fact that the book does form a coherent whole despite the varied nature of the sources of the papers. Finally, some mention must be made of the sense of fitness of the editors, who have chosen to use as endpapers in a book on Delphi, a photograph of the Tholos at the Sanctuary of Athena, in the Temple of Apollo at Delphi (in hardbound edition). This is an appropriate final touch to a book which will serve as an authoritative source of information on Delphi for years to come. JOSEPH P. MARTIN0 Associate Editor