The Grammar of Graphics

This book is closely related with the upcoming new book "ggplot2" by Hadley Wickham. "ggplot2" is an R package for producing elegant graphics based on the Grammar of Graphics.

This is an unusual book. It should be of interest to those who have to display data in graphic form, but it is not a typical book of that kind. A subset of those who enjoy Tufte (The Visual Display of Quantitative Information, 2nd edition) or Cleveland (The Elements of Graphing Data, Visualizing Data) will also enjoy this. This book does not attempt to coach you on how to make good graphics, but if that is your goal it attacks a related, important aspect: rules that connect the data and the graphic. Obviously then, it might be of great interest to computer scientists and theoreticians. The grammar described in the book is also object oriented in design which helps to make the code quite elegant. In my instance, I am a "power-user" of the SPSS statistics software package, and the ideas in this book have been partially implemented in SPSS since around 2005 (Version 14.0). This is absolutely not a user's guide or a how-to manual for using SPSS. If, however, you already have a working knowledge of the latest SPSS graphics and its related programming language then this book will put that whole approach in context, and massively increase your understanding of the technology "behind the scenes". Many technical audiences might benefit regardless of their using or not using SPSS Statistics. Unlike Tufte, you will find no hand drawn graphics, or illustrations crafted in Adobe illustrator - they can be both beautiful and effective but such graphics would not belong in this book. All of the graphics here are created using a computer and a grammatical description and arise directly out of a data set. Making a graphic elegant and clear is the work of the designer, the purpose of the grammar is to insure that the graphic is tied to data, and to separate graphics that make sense from graphics that are non-sense. To paraphrase the book, graphics that are grammatically correct may or may not be ugly, but can never be meaningless. In a brilliant final chapter the author explains that the famous Minard's Napoleon's march chart (see Tufte) has some incorrect data, and that Minard took some liberties with the data. The conclusion: a beautifully designed and crafted graphic does not assure 100% accuracy. Whereas to quote the author, "a graphic without a concept of a variable is not a statistical graphic". So, if you make the variables available (in a data set) to your audience you make it all that much harder to mislead with graphics. Although Tufte's books have a VERY different style, avoiding lying with graphics is an important, recurring theme. So if you are technically inclined, and especially if you are an SPSS user, seek out this book. It is a great companion to the literature on chart aesthetics. Before you dive in, note that this book is several hundred pages long. SPSS user's should look over the "GPL Reference" first to determine if this is a topic you are willing to dedicate time to.

A good book with lots of actual graphical examples to illustrate the different concepts. Coverage of many different kinds of graphs and how they are related and transformed into each other.