This book deals with the theory and practice in the design of data graphics and makes the point that the most effective way to describe, explore, and summarize a set of numbers is to look at pictures of those numbers, through the use of statistical graphics, charts, and tables. It includes 250 illustrations of the best (and a few of the worst) statistical graphics, with detailed analysis of how to display data for precise, effective, quick analysis. Also offered is information on the design of the high-resolution displays, small multiples, editing and improving graphics, and the data-ink ratio. Time-series, relational graphics, data maps, multivariate designs, as well as detection of graphical deception: design variation vs. data variation, and sources of deception are discussed. Information on aesthetics and data graphical displays is included. The 2nd edition provides high-resolution color reproductions of the many graphics of William Playfair (1750-1800), adds color to other images where appropriate, and includes all the changes and corrections during the 17 printings of the 1st edition.
Tufte reviews how information can be presented (i.e. a minimal amount via a sentence; a moderate amount via a table; a huge amount via a graphic) and then turns his attention to graphics -- from their beginnings in cartography to how to achieve graphic excellence today.
He urges a multi-disciplinary approach, cautioning that, “Allowing artist-illustrators to control the design and content of statistical graphics is almost like allowing typographers to control the content, style, and editing of prose.” He touches on psychology and cognition. He rails against using graphic design to deceive, and enlightens readers by pulling numerous examples of misrepresentation from prominent media. He devotes a large part of the book to improving the effectiveness of graphs by urging the elimination of “chart junk” (e.g. moiré-effect cross-hatching) and numerous other sources of “non-data ink.” In fact, a chapter wherein he strips away seemingly necessary text, frames, hatch marks, etc. (leaving little more than an ether vapor but in the process simplifying and clarifying the meaning) is revelatory.
So many books I've read recently have referenced Tufte, and I'm glad to have finally read him directly. Highly recommended.
Two lessons emerge from Tufte's masterpiece:
1. Eliminate from a graph anything that doesn't add information
2. Maximize the amount of information per square inch
Don't let the antiquated looks of the book deceive you: keep reading till the end and you'll find that Tufte's teachings are as valid today as they were 30 years ago.
I do wish, however, that this second edition incorporated more advice specific to computer-generated graphics. I'm particularly disappointed because this edition was published as recently as 2001, many years after the personal computer became THE tool for graph making. Without such applied advice, and given the old look of the examples and the large format of the volume, I'm afraid this book will be regarded by many as a geeky coffee-table piece.
Rarely do I find a book that I would call beautiful, but this meets the criteria, both as a physically appealing book, apropos to the purpose of the book, and an informationally dense, and well presented one. A favorite quote of mine, from Zen and the Art of Motorcycle Maintenance, where the protagonist says; "I remember... remarking about the analytic craftsmanship displayed." This was my reaction to Tufte's book.
The book manages to decompose graphical presentation of data into categories other than the x- and y-axes, and instead talks about multifunctional elements and data density. The book reimagines the nature of numerical information using a graphical design perspective, with a healthy dose of common sense as to how graphs are used, and a veritable treasure trove of examples of both good and bad design.
This book, along with "How Buildings Learn," by Stewart Brand, is a rare example of a narrow focus with an incredibly broad appeal. This book is not for the narrow specialist in constructing the sometimes obscurely complex graphics displayed, but rather for anyone who is interested in the data presented to them, and certainly anyone who produces this data in any form.
Of the Tufte book's I've read (two and a half out of four), this is the one to read if you don't manage to get to the others. It lacks the complexity and beautiful Japanese examples of his other works; but its focus is razor sharp and its examples are, again and again, things that we encounter in our everyday lives. The design principles are similarly brilliant: applicable to all, and will make you think the next time you cobble together a table or graph on the computer.
Perhaps his later work is more useful. I'll have to check it out and see.
Please comment in my profile if you know of software tools that support directly or indirectly the principles advocated by Tufte.
Immaculately designed and packed with fantastic illustrations of good and bad approaches to visualisation, this book is a pleasure to read and absorb. I found that it worked well both when reading just a couple of pages at a time and when immersing myself in it for a longer period of time.
This is one of those books that I know I will be revisiting for reference in the future.
Read/ scanned in just a few days, because a fair bit was over my head. The parts I did understand, however, were terrific and valuable, both for readers and creators of graphs and tables.
Tufte has a sense of humor, too. When showing how graphs can be presented to be misleading (ie, making a bar graph meant to represent gains vs losses start at a large negative number, and disguising the numbers by having a bold 'background' illustration), he quantifies the apparent discrepancy as a lie factor."
I would *love* to take a course under Tufte, or even under any decent professor who uses this book as the primary text.
If your library has this, I highly recommend you at least browse it."