The Visual Display of Quantitative Information, 2nd edition

Good graphic design, he argues, reveals the greatest number of ideas in the shortest time with the least ink in the smallest space. Interestingly, some of the best examples of this come from the pre-computer era, when graphics had to be drawn by hand (and therefore more thought had to go into their design, rather than the author just calling up the Bar Graph template on the desktop.) For example, that picture you can see on the front cover of the book is actually a train timetable that packs a whole list of arrivals and departures at many different stations into a single little picture. A better example (and the "best statistical graphic ever drawn") shows Napoleon's route through Europe. It shows a) the map b) where he went c) how many people were in his army at each point and d) the temperature on the way back that killed off his army. At a glance you can see the factors that led to his army losing. AND it was drawn by hand in 1885 and is little more than a line drawing!

He also gives examples of really bad design, (including "the worst graphic ever to make it to print"), and shows what makes it so bad. His examples prove that information-less, counter-intuitive graphics can still look dazzlingly pretty, even though they're useless. In some examples, he shows how small changes can make the difference between an awful graphic and a really good one. My favourite example of this is how he drew the inter-quartile ranges on the x and y axes of a scatterplot, thus adding more information to the graphic without cluttering it up.

In summary, there's a lot more to good graphic design than being an Adobe guru. Reading this book made me feel like a more discerning viewer of graphics!


1st edition compared to 2nd   March 1, 2002
S. M Marson (Lumberton, NC)
151 out of 158 found this review helpful

Years ago, I purchased the first edition of VISUAL DISPLAY OF QUANTITATIVE INFORMATION. The second edition provides high-resolution color reproductions of the several graphics found in the first edition. In addition, corrections were made. However, to most readers/users, I doubt that the changes would be worthy of purchasing the second edition if one already owns the first edition.

Edward R. Tufte is a noteworthy scholar and the presentation of the material presented in this book is awe-inspiring. Tufte has also compiled two other books that can be best described as quite remarkable. These additional books are entitled, ENVISIONING INFORMATION and VISUAL EXPLANATIONS. All three of these volumes are not merely supplemental textbooks; they are works of art.

My intent was to use VISUAL DISPLAY OF QUANTITATIVE INFORMATION as part of teaching my statistics course. Students, but mostly faculty, are overly impressed with inferential statistics. Graphics play an important role in the understanding and interpretation of statistical findings. Tufte makes this point unambiguously clear in his books.

Two features of VISUAL DISPLAY OF QUANTITATIVE INFORMATION are particularly salient in teaching a statistics course. First, the concept of normal distribution is wonderfully illustrated on page 140. Here the reader is reinforced with the notion that in the normal course of human events, cultural/social/behavioral/ psychological phenomena usually fall into the shape of a normal distribution. The constant appearance of this distribution borders on miraculous. Just as importantly, it is the basis for accurate predications in all areas of science. Tufte's illustration (page 140) speaks to this issue much more clearly than a one-hour lecture on the importance of the normal distribution. Which goes to show -- once again -- "a picture is worth a thousand words." Sadly, the illustration on page 140 is small and in black and white. I wish the second edition included a larger reproduction of this photo. A color presentation would have been helpful.

Second, Tufte continues his unrelenting pattern to reinforce the importance and impact of illustrations in understanding complex concepts. In particular, page 176 demonstrates the impact of Napoleon's march to Moscow. The illustration is both profound and eerie. The reader is left with a feeling of death and pain for the foot soldiers...


Very Short on Substance; Mostly Has Only One Major Point   April 20, 2004
Mark A. Weiss (Germantown, MD United States)
54 out of 72 found this review helpful

Other reviewers have mentioned a few negatives. To me, these mostly boil down to short-on-substance problems. The author is a bit pompous -- which wouldn't matter that much if he had a lot to say. Alas, he does not. The author's major point is: eliminate "chart junk" (e.g. 3-D effect bars, etc). He is manically obsessive-compulsive about this point so that he takes it to extremes -- get this: computing "data ink" to "junk ink" ratios he even eliminates the axis line (to increase the ratio). Giving the book a second chance now over a year later, I found that the "eliminate chartjunk" is not the sole point the author makes -- but 80% of the book is about that one point. During my "2nd chance" read of the book I did find a couple of substantive ideas: the white-grid run-thru of bars on bar charts, and the discussion of aspect ratios. (Plus the tics-at-data to give marginal disbn's of X and Y.) However, I still maintain, if you really want to learn new techniques and real-value PRINCIPLES get William Cleveland's two books "Elements of Graphing Data" and "Visualizing Data." Cleveland's book are enjoyable to read and filled with eminently useful ideas. I've used principles from Clevelend's book to great effect. I'd been graphing for decades, but with Cleveland's book I made a GIGUNDOUS jump in the quality of my graphical communication. Skip the low-on-substance, one-note Tufte and go for the full-of-substance, emminently useful Cleveland.


The essential guide to avoiding graphical lies   March 20, 1997
49 out of 52 found this review helpful

This book, and the two companion volumes ("Envisioning Information" and "Visual Explanations") are must-haves for anyone who is in the business or producing or interpreting statistical information.

Tufte starts with a simple proposition: graphs and graphics that represent statistical data should tell the truth. It's amazing how often designers of such graphics miss this basic point. Tufte clearly and entertainingly elucidates the most common "graphical lies" and how to avoid them.

Read this book and you'll never look at a newspaper or presentation graphics the same way again -- you'll be left wondering if the author *intended* to lie about what the data were saying, or if he/she just didn't know any better.

Another reviewer claimed that this book talks about how to make graphics accurate, not beautiful. He's right in some sense, but who cares? There are a million books on how to make "pretty" graphical displays, but precious few on how to make useful ones. These books are they.


Sets the stage for all information architects   May 25, 2000
Thomas Schultz (Aarhus, Denmark)
47 out of 50 found this review helpful

This book will teach you some basics on how to most effectively present quantitative information using various sorts of graphs and charts. Afterwards you will know how and why you should get rid of chart junk (gridlines, tick marks, ornaments, etc.) or alternatively using some of the examples on bad design presented, you will see how to manipulate your audience using the "Lie Factor". Actually the advice given in this book could easily fit within just one piece of paper, but then: This book is simply beautiful. It is state of the art for printed books, you almost feel a passion for it. Mr. Tufte takes his own medicine: No words in this book are superfluous. Illustrations and examples are carefully selected and reprinted with the utmost care. It takes no more than some hours to read the book, but afterwards you can use more than just a few hours to study the examples of timeless graphic displays. The only reason why this book is short of five stars is the following: Mr. Tufte uses quite some space providing statistics about charts found in different publications (chart junk percentages, lie factor. Personally I find this information fairly irrelevant and would have preferred more examples of chart remakes. However this book is definately still a MUST have!