E.J. Wegman, review of The Grammar of Graphics, Journal of the American Statistical Association, 2000, 95, 1009-1010.
Destined to become a landmark in statistical graphics, this book provides a formal description of graphics, particularly static graphics, playing much the same role for graphics as probability theory played for statistics. That is, it gives a formal framework for developing graphics in a way similar to the way probability provided a formal framework for the development of statistical theory. Graphical methods have been part and parcel of statistical methodology for some time, but have in a sense been developed as ad hoc methodology. Wilkinson provides a synthetic framework in which scatterplots, bar charts, histograms, pie charts, and other, more sophisticated tools are shown to be special cases. This excellent book is highly recommended for anyone - statistician, engineer, or social scientists - with an interest in statistics. It certainly transcends the narrow interest of those doing research in statistical graphics.
The book comprises 15 chapters: Introduction, How to Make a Pie, Data, Variables, Algebra, Geometry, Aesthetics, Statistics, Scales, Coordinates, Facets, Guides, Graphboard, Reader, and Semantics. The main substance of the book begins with the chapter on Data and runs through the chapter on Facets. Each chapter begins with a discussion of the etymology of the chapter title. I found these to be particularly satisfying discussions, as is Wilkinson's interesting exploration of the evolution and variants of Mies van der Rohe's well-known phrase "God is in the details." These discussions set the context for the book.
In the Preface, Wilkinson remarks on the inspiration that he derived from what was once the AT&T Bell Labs statistics group. This also would seem to cause some unevenness in his treatment of specific techniques. For example, on page 224, he attributes to John Tukey the proof that as the power p approaches 0, the power transformation, suitably scaled, behaves like a log transformation. Not only is this a trivial exercise in elementary calculus, but most of us would know this rescaled power transformation as a Box-Cox transformation. Neither of these scholars is mentioned or referenced.
The foregoing complaints notwithstanding, The Grammar of Graphics is an urbane, erudite, groundbreaking book laced with a nice sense of humor. I highly recommend it.