Volume 4 Number 1 Copyright 1998
Visual Explanations: Images and Quantities, Evidence and
Narrative
Edward R. Tufte
Cheshire, CT: Graphics Press, 1997
ISBN: 0-9613-9212-6, 158 pages, $45
Key Words: Images, Information, Evidence
Reviewed by: DATA QUALITY Editorial Staff
Visual Obfuscation
There are many ways to develop knowledge about the world. We rely on our senses to pick up stimuli from the environment. We try to transform this data into information. The information is either someone else's interpretation or what we've discovered ourselves, through induction, deduction, or both. And we can compare our interpretation of the data we have gathered with the interpretations of others.
Apart from the pure pleasure of learning something new, we like to have data interpreted for us to help us make rational decisions. We look to statistics to make sense of our data. Are we on the right track? Is the difference among data as large as we would like to believe? Would a well-designed experiment change our belief? As statisticians and quality professionals, we hope so!
On the other hand, in the world of art, creatitivity is judged by how original an artist can be when interpreting data. Artists present perceptions of stimuli we may not even consider to be data and force us to think about what we see, hear, and feel. We don't have to agree with the artist's interpretations, but would disagreeing detract from the uniqueness of the artist's interpretation?
In Visual Explanations: Images and Quantities, Evidence and Narrative, Edward Tufte tries to entice the reader to view different kinds of data in media that are not the usual means of expression for the data. According to the author, "Visual Explanations is about pictures of verbs, the representation of mechanism and motion, of process and dynamics, of causes and effects, of explanation and narrative. Since such displays are often used to reach conclusions and make decisions, there is a special concern with the integrity of the content and the design."
This heavily illustrated book is organized into seven chapters. The first chapter, "Images and Quantities," discusses how art reproductions tend to "dequantify" what they present by not providing a sense of scale with appropriate labeling. Chapter 2, "Visual and Statistical Thinking: Displays of Evidence for Making Decisions, " looks at evidence from two historical events in which data presentations were used to determine courses of action: the London cholera epidemic of 1854 and the 1986 decision to launch the Challenger space shuttle. "Explaining Magic: Pictorial Instructions and Disinformation Design" (Chapter 3) depicts how printing cannot satisfactorily capture the deceptive practices of magic on paper. The author lists his strategies for good presentations to audiences. Chapter 4, "The Smallest Effective Distance," explores the advantages of a minimalist approach to information design, asserting that "less is more."
In Chapter 5, "Parallelism: Repetition and Change, Comparison and Surprise," the author tries to show how multiple parallel images can provide a context with which to assess variation in data (i.e., before and after pictures). "Multiples in Space and Time" (Chapter 6) continues the previous chapter's premise that parallelism is useful for evaluating multiple images. The author uses many illustrations in an attempt to demonstrate how sequential images in a frame can illuminate a concept better than one image can. Chapter 7, "Visual Confections: Juxtapositions from the Ocean of the Streams of Story," prolifically illustrates the use of concocted images throughout history. Tufte opines that confections offer images that can be used to offer new insights into ideas that may be too advanced, obscure, or insipid to be easily understood.
Individual pages and sub-chapters of Visual Explanations are graphically excellent and exhibit data and information in ways that are useful today. For example, Tufte's graphical summary of patient status and time series plots of laboratory data organized in q-q plot format (page 111) on one page is a significant advance over the way medical records are usually displayed. Such summaries could make clinical and laboratory data more accessible for patient diagnosis and evaluation. Similarly, the 1985 public safety warning (page 144) from The Washington Post about wading or swimming in the Potomac River above Washington, ("Why is the Potomac River So Dangerous?") is as appropriate today as it was 13 years ago.
Tufte's narrative and illustrations about John Snow's investigation of London's 1854 cholera epidemic are the best we have read about an important public health event in history. Most public health textbooks provide only a cursory account of Snow's epidemiological methodology. By most accounts, observational metholology today hasn't improved much beyond how Snow conducted such studies over 144 years ago.
From a data quality and information quality perspective, Visual Explanations is mediocre at best. Tufte brings together data from a variety of disciplines, then presents them in a haphazard way. The type size is often too small to read easily, especially by readers approaching middle age who haven't yet acquired bifocals. So are many of the graphs. Pictures and photographs are displayed without a sense of good page layout and design. Many of the reproduced graphs and images are very small. Many of the larger graphs and pictorial reproductions look "muddy." It appears that Tufte rummaged about various museums and libraries until he found prints, charts, graphs, and photographs that appealed to him. He used these visual materials to illustrate his book and to "prove" each chapter's hypothesis. Data quality and information quality are not addressed as topics anywhere in Tufte's book. Moreover, Tufte assumes that data and information from any time and any place are accurate - a serious error.
Regarding Tufte's comprehensive coverage of the Challenger accident data, is his analysis another example of "Monday morning quarterbacking?" Where was Tufte before the Challenger accident occurred? And why is there no mention of the important contributions that American Statistical Association members like Robert Hogg (ASA President in 1988), Richard Scheaffer, and others made by spreading basic quantitative literacy techniques that would have prevented the Challenger disaster if such techniques had been widely adopted when first proposed?
Although we enjoyed Tufte's book, it was hard to decide for what audiences the book was intended. The various topics that Tufte superficially covers can be explored in more depth with textbooks on those subjects. As previously mentioned, the book is heavily and - in the reviewers' opinions - poorly illustrated. The classic art and high-tech computer graphics are seldom self-explanatory, as most scientific images and graphs are required to be. Therefore, the book seems to be too much an "art" book for physical science researchers and academics, yet too technical for those lured by its promise to be an art-oriented approach to visual images.
Go to:
Comments: dqemail@aol.com (10/23/98)