1.4 Reading about graphics

1.4.1 Books and articles on data graphics

There are many good sources on how to draw graphics and what principles and guidelines should be followed. Despite the plethora of good advice available, it is not always taken, judging by the graphics that are published. More relevantly for this book, there is far less advice on how to interpret graphics. It is often surprising to find graphics that have been beautifully drawn with a great deal of effort that are then barely discussed at all. When looking at graphics in a book or on a website, check whether there is text accompanying them that explains what information the authors believe their graphics convey.

Classics of graphics literature include books by Tufte (Tufte (1990), Tufte (1997), Tufte (2001)) and Cleveland (Cleveland (1993), Cleveland (1994)). Going back further, some of Playfair’s work has been republished (Playfair (2005)) and Minard’s work across the later part of the nineteenth century has been gathered together attractively in Rendgen (2019). There is a useful overview of the history of graphics in Friendly (2008) and this is covered in more detail in his book with Wainer (Friendly & Wainer (2021)). The book “Design for Information” (Meirelles (2013)) includes many fine historical and modern graphics in a structured approach to visualisation for design students.

Graphics advice has always tended to be just advice without theory. Two important exceptions are Bertin (1973), an English translation is available, Bertin (2010), and Wilkinson (2005). Wilkinson’s work has proved especially influential through Wickham’s implementation in the R package ggplot2, described in Wickham (2016) and Wickham (2023) and on the web. Unwin (2015) concentrated on applications of this implementation and the graphical theory behind it using real datasets.

The graphics in this book are part of what is nowadays referred to as Data Visualisation, and are only a part of the wider field of visualisation. There is much work in Scientific Visualisation and Information Visualisation that overlaps with what is covered here. The Computer Science approach to Visualisation is discussed well in Munzner (2014) and there is an extensive technical literature that is too broad to discuss here. Cairo has written popular books, deriving from his journalistic experience (Cairo (2012), Cairo (2016), Cairo (2019)), and these include some attractive examples.

Kirk’s book (Kirk (2016)) offers a range of different graphics and concentrates on project flow more than traditional books do, discussing design, development, and cooperating with clients. Wilke’s book (Wilke (2019)) is more about graphics for scientific reports and publications, looking at graphics from a practitioner’s point of view. He recommends that every graphic has a title, good advice for presentation graphics, not so relevant for exploratory graphics where the message is not yet known. Healy’s well-written book (Healy (2018)) includes some real applications, and also some invented ones. None of these books include much on what can be seen in their graphics, yet the authors have a lot of experience and there is doubtless more they could have written. Assuming readers see what you can see may flatter them, but does not assist them. “How Charts Work” (Smith (2022)) is on working with data graphics to present your ideas, mainly using examples from the Financial Times where Smith works. Chang’s book (Chang (2018)) goes into the details of drawing graphics in R and has excellent complementary webpages. Murrell’s book (Murrell (2018)) provides a clear overview of the technical structure underlying R graphics. Another approach to writing about graphics using R is provided by Rahlf (2017). The book is built around 111 examples of publication-worthy graphics and in each case the R code needed to draw the graphic is provided. As little advantage is taken of packages for R, the code is often lengthy, but it demonstrates how even the most elaborate (and often attractive) plot designs can be reproduced using base R. Again there is little discussion of what each graphic shows.

Much interesting and attractive work has been carried out on producing special individual graphics developed using sketching. Two prime examples of books describing this process are Posavec & Lupi (2016) and Bremer & Wu (2021). Both present striking and attention-grabbing graphics.

Kandogan & Lee (2016) describes research associated with designing automatic visualisation systems and claims that the authors found around 550 guidelines in their literature review. In a related vein, Wood et al. (2018) introduces literate visualisation, a scheme for integrating the process of writing data visualization code with a description of the design choices that led to the implementation. The authors have developed a literate visualization software environment, litvis, for carrying this out.

It is impossible to cover all the literature, but it would be a pity to omit mention of some of the more unconventional contributions. “Tidying Up Art” (Wehrli (2004)) demonstrates a novel way of deconstructing complex works of art into summarising statistical forms. The book “Info We Trust” (Andrews (2019)) discusses ideas about information and graphics in an amiable and personal way, using the author’s own drawings.

Statistics textbooks have traditionally had a chapter on graphics, usually just describing which common graphics are available rather than explaining anything more substantial. Yet data graphics complement statistical analyses and statistical models are one of the ways that can be used to assess results found from graphics. There are some excellent books on statistics. If you need more detail on a statistical topic, look for it in one of the books you personally like.

1.4.2 Websites and other media

There are numerous websites discussing data visualisation, some more active than others, and several of the books mentioned have supporting websites. For a number of years Kirk’s website (Kirk (2023)) included monthly digests of what he considered the best of data visualisation on the web. The old selections are still worth a look.

Most printed news media, indeed all kinds of organisations, offer graphical content on the web. They provide many excellent examples—along with some not so excellent ones—and they are a valuable source of material for studying graphics.