Browse Books

Reviewer: John J. Hirschfelder

Few aspects of computer software have changed so radically in the last 20 years as the role of the user interface. Whereas in batch processing days, often less than 10 percent of a programs code was devoted to this interface (if there was one at all), now interface-related code can represent 90 percent or more of the program. Thus, the recent proliferation of works on interface design is no surprise. The six books reviewed here extend across the spectrum, from the basic handbook (Powell) to the highly theoretical treatise (Eberts). Some of these books are oriented to specific software environments. Comparative data are provided in Table 1. The range of topics treated varies considerably from book to book, due to the different intended audiences. A comparison is provided in Table 2. The topics treated are similarly varied. On the practical side, the reader may find guidance in such areas as screen layout; selection of menu items; justification and alignment; keying procedure; and use of color. At the theoretical end, one may read about models of human mental processes; effects of social conditions on computer usage; methodology of experiments in human-computer interaction; and recent research developments. The books under review include two handbooks, two textbooks, and two theoretical works. (Table 1 shows these desig nations for each book). The theoretically oriented works may be used as texts at the graduate level. Hix and Hartson The authors present the development of user interfaces from two aspects: the characteristics of the product and the process of developing it. Groups of exercises are provided at the end of many sections. References are provided at the end of each chapter. The emphasis is on the process. Topics treated include systems analysis (requirements definition and conceptual design), representations of interface designs, and usability specifications. A formal language called User Action Notation is introduced for defining interaction requirements. Other topics are rapid prototyping, formative evaluation (that is, iterative evaluation during the development process), usability testing, quantitative methods, and interface development tools. The emphasis on the user interface development process, rather than on the interface itself, places this book in the systems analysis and project management arena; the treatment of user interface design itself is little more than a catalog of common features. Powell Powell has written the closest of these six to a pure handbook: the information and guidelines are well organized and concisely presented, with a minimum of background or theory. The type is large and the white space is generous, making this book significantly shorter than the others reviewed here. It is a quick, easy read, but it covers nearly all the issues that the average developer of in-house applications and small commercial packages will encounter. The author states that he included guidelines that he thought trivial, because even these are too often forgotten or ignored. The bibliography lists only 12 references, but each is accompanied by a brief review. This books weaknesses are its orientation to character-array screens under MS-DOS and its reliance on the FoxPro database system for examples. A second edition, oriented to graphical user interface systems, would be a good contribution to the literature. Nevertheless, if you are embarking on a project, need a refresher course on user interface design, and have time to read only one book, read this one. Shneiderman This comprehensive treatment mixes theory with practice in about equal portions. All major topics are covered well. Evaluations of interface features are supported by descriptions of relevant experiments, the results, and references to the literature. The treatments of direct object manipulation, interaction devices, response time, and display rate are particularly comprehensive. Each chapter ends with two summaries entitled “Practitioners Summary” and “Researchers Agenda.” The reader seeking practical guidance quickly will find this book frustrating; the guidance is there, but is interspersed with theory and with results of research projects. This book should be read in a leisurely fashion—between projects, rather than at the start of a project. Table 1: Descriptive Data Hix and Hartson Powell Shneiderman Cox and Walker Galitz Eberts Number of pages 381 390 593 362 532 649 Number of chapters 12 13 14 9 15 25 Number of appendices 0 1 1 0 0 0 Type Textbook Handbook Theoretical treatise Textbook Handbook Theoretical treatise Audience 4th-year graduate student Practitioner Researcher advanced practitioner Undergraduate Practitioner Researcher Exercises Yes No No Yes No Yes Reference software Non- specific MS-DOS, FoxPro Non- specific Non- specific Non- specific X Windows Color plates No No Yes No No No Cox and Walker The authors have written a textbook targeted at an introductory undergraduate course in an information systems curriculum. Basic topics are introduced and illustrated by two running examples, one a vehicle scheduling application, the other an airline route selection program. The role of usability testing in product development is emphasized. Cox and Walker provide no systematic presentation of interface design features, and little guidance on the application of features. This text should be easy to teach, but it should be supplemented by a basic handbook, such as Powell. Galitz Galitzs book is the successor to his Handbook of screen format design , whose most recent edition was published in 1989 [1]. Unlike the other books in this review, Galitzs is narrowly focused on screen layout, and this topic is treated exhaustively. Good and poor design practices for every feature of any screen are presented. Unlike Powells handbook, the rationale for the recommended practices is presented with references to the research literature. Examples are numerous and well chosen. The first three chapters, “Introduction,” “The System User,” and “System Considerations,” contain general principles. The remainder of the book is best used as a reference, which is facilitated by an exceptionally detailed table of contents. This book is the most comprehensive handbook I have seen. It is an essential reference for developers of serious commercial software products. Eberts Eberts focuses on the underlying theory of human interactions with computers and with evaluation methods. The introductory section, consisting of three chapters, reviews the common forms of information presentation. Subsequent sections, of several chapters each, treat four approaches to analyzing human-computer interaction: empirical, cognitive, predictive modeling, and anthropomorphic. The section on empirical methods includes extensive guidance on the design of experiments regarding human-computer interaction, and lists pitfalls to avoid. The section on the cognitive approach treats mental models, neural networks, and human spatial reasoning. The section on predictive modeling emphasizes the goals, operations, methods, and selection (GOMS) model of human-computer interaction, and introduces Natural GOMS Language. The section on the anthropomorphic approach treats such topics as alternative input devices and affordances of objects. This work is highly theoretical, and is directed at researchers. It is of marginal value for practitioners. Comparison These six books are best considered in three groups: the textbooks (Hix and Hartson, and Cox and Walker); the handbooks (Powell and Galitz); and the research treatises (Shneiderman and Eberts). Each group addresses a different market. Of the textbooks, the Hix and Hartson book is at a significantly higher level, and contains more comprehensive practical guidance than Cox and Walker. Hix and Hartson is suitable for an advanced course in a computer science program, while Cox and Walker is suitable for an introductory course in an information systems curriculum. Of the handbooks, Eberts is much more comprehensive, but Powells book, being concise and easy to read, will prove more useful to the average developer. Table 2: Coverage of Topics Hix and Hartson Powell Shneiderman Cox and Walker Galitz Eberts Command languages Good Good Very good Poor Very good Good Menus Good Good Very good Poor Very good Good Data entry Good Good Very good Poor Very good Good Error messages Good Good Good Poor Good Very good Graphs Poor Good Adequate Poor Excellent Very good Color Poor Poor Adequate None Excellent Adequate Design methods Very good Adequate Adequate Good None None Product development Very good Adequate Adequate Good None None Testing Excellent Adequate Good Excellent None Excellent Social issues Good Poor Excellent Poor Adequate Good Psycho-motor issues None None Excellent None Adequate Very good Cognitive theory Poor None Excellent None Good Very good User documentation Poor Good Good Good None None GUI systems None Poor Adequate None None Good Development tools Good Poor None None None Adequate Standards and guidelines Good Adequate Good Good None None Of the treatises, Shneidermans is the more general and also the more practical. It presents a balanced view of the entire field at the research level while translating current knowledge to practical guidance. Eberts book is a very thorough treatment of narrower scope.