Browse Books

Reviewer: Mark Monmonier

Edward Tufte's well-illustrated book-length essay on the design of statistical graphics warrants the attention of computing professionals for two reasons. First, the book contains many logical, straightforward rules about visual focus, graphic contrast, scaling, and other important design concepts that could serve as a knowledge base for an intelligent system for producing statistical charts. Second, these rules and Tufte's convincing supporting examples of good and bad charts and maps set a standard of excellence that developers of software for data graphics might readily adopt. Although this book has been available since 1983, software vendors seem inclined to follow the cumbersome, rigid graphics successfully peddled to eager but noncritical buyers of SAS/GRAPH and Microsoft Chart. Despite this corporate indifference, Tufte's growing following among graphic artists, statisticians, geographers, journalists, business analysts, and social scientists has created a significant submarket eager for such improvements as scales that represent range and overall designs that attract the eye to relationships in the data. In demonstrating the aesthetic and practical advantages of good design at a time of growing interest in computer-supported visualization, Visual display might well become this century's most influential treatise on data graphics. The book has many merits, both physical and intellectual. Before microcomputers and laser printers made self-publishing practicable, Tufte decided to bypass conventional dissemination channels, establish Graphics Press, and hire his own designer, typographer, artists, and printer. The result is a well-organized, superbly produced book that is a pleasure to look at and to read. Tufte, who writes well, has mastered the art of integrating words and pictures. His text is lively and mildly glib, and his approach is both practical and scholarly. He gives the reader a sense of the origin of statistical graphics in the seventeenth and eighteenth centuries, and presents several classic examples of visually efficient, innovative graphics, including Charles Joseph Minard's dramatic map and time-series chart portraying the attrition of Napoleon's army in its unsuccessful Russian campaign during the winter of 1812-1813. Minard's graphic is used to illustrate Tufte's data-ink concept—reinforcing the most appropriate visual focus by maximizing the proportion of “a graphic's ink devoted to the non-redundant display of data-information.” The book's chapter titles and two main divisions support Tufte's description of it as “a celebration of data graphics.” Part 1, addressing graphical practice, consists of three chapters showing how well-conceived graphics can convincingly communicate verbally complex relationships. Chapter 1, “Graphical Excellence,” is an unabashed celebration, whereas chapter 2, “Graphical Integrity,” acknowledges the power of graphics to mislead and distort. Chapter 3, “Sources of Graphical Integrity and Sophistication,” finds explanations for mediocre graphics in quantitatively ignorant illustrators and the widely held belief that readers cannot digest any but the most simplistic graphics. Part 2, addressing the theory of data graphics, presents guidelines, caveats, exemplars, and examples of abominations in its six chapters. This section is more prescriptive than part 1. Chapter 4, “Data-Ink and Graphical Redesign,” uses before-and-after examples to demonstrate the data-ink concept. Chapter 5, “Chartjunk: Vibrations, Grids, and Ducks,” condemns a variety of distracting embellishments, and chapter 6, “Data-Ink Maximization and Graphical Design,” applies the data-ink concept to common statistical charts showing frequency distributions, time series, and bivariate relationships. Chapter 7, “Malfunctioning Graphical Elements,” examines comparatively complex graphs portraying three or more variables, and chapter 8, “Data Density and Small Multiples,” demonstrates that rich graphics need not consume much space. The final chapter, “Aesthetics and Techniques in Data Graphical Design,” presents for the sake of completeness a few further principles concerned with size, shape, balance, and layout. Although this book is not intended as a classroom text, it is useful as a required or supplemental text because it identifies pressing problems and offers effective solutions. Its organization is logical and refreshingly unorthodox, and it addresses all relevant topics well except color, which it treats with but a handful of examples, most of which illustrate the meritorious use of contrasting hues. The selective references reflect a careful examination of a large literature. The index would be considered skimpy were this book designed largely as a text or reference, yet the work's strength as a perceptive, intriguing, timely synthesis far outweighs this minor annoyance.