2025/11/20 by Christian Hennig · 1 voice
Computer Science · #Data Visualization and Analytics #Computer Graphics and Visualization Techniques #Data Analysis with R
paper · pdf · doi:10.1093/jrsssa/qnaf194
openalex publication_date 2025/11/20 · openalex created_date 2025/12/03 · openalex updated_date 2026/07/30
This is an unusual book about data visualization. It does not give much explanation or systematics of what kinds of plots there are and what can be done, and neither does it explain how to use software to do the things shown (the book has a github page with all the code and data; all is done in R). There is nothing on the history and development of data visualization either. Statistical models and inference are rarely mentioned. Very occasionally we see confidence bands and simple summary statistics, but the visualizations all show the data directly rather than derived quantities such as residuals. The book is on exploratory graphics as opposed to presentation graphics. It focuses on why to do what, and better understanding of what it means that can be seen in the plots. Furthermore, it emphasises the crucial role of knowledge of the subject matter background and the aim of visualization strongly. As such, this book is rather unique and will give the reader a lot that cannot be found elsewhere. After the Introduction, there are two chapters. The first one, Graphics in Action, presents analyses of 26 real data sets. This occupies 285 pages, the vast majority of the book. The second chapter, Advice on Graphics, discusses lots of aspects of statistical graphics in a more systematic manner. This includes, among other things, considerations regarding colours and ordering observations and variables when setting up graphs, but there are also sections covering the human perception of graphs, attention, cognitive biases, and checking of potential inconsistencies. The role of visualization together with background knowledge to assess data quality is also emphasized. Devoting so much space in the beginning to data examples is a bold decision. The data analyses can be read independently of each other, so a reader can sample some of these before going to the second chapter; reading them all in one go is a bit tedious before getting to the more general and systematic exposition, although they are nicely written. The good reason for presenting all these analyses is to drive the point home that how to do and understand data visualization always happens in crucial interaction with the specific aims of a study, the data collection, and the subject matter background. The first chapter provides a lot of material that serves to illustrate and motivate the main points made in the second chapter. It highlights many issues that can be relevant in data analysis in practice and that are very rarely treated in other textbooks if at all. The data sets can also be a valuable resource for teaching. As is probably unavoidable with such a book, occasionally in the first chapter, I would have liked to see different aspects of the data discussed, and in the second chapter some issues would have deserved more elaboration rather than just mentioning that they are also important. The book is not short at 417 pages before references, so some selectivity was required. Given that this is a book on data graphics, the author is surprisingly stingy with axis labels. One rationale regarding the selection of data sets was certainly that the readers should not need all too sophisticated background knowledge of the topics. This explains to some extent the distribution of subject areas covered by the data examples. There is no genetics data set but six on sports. There are three data sets on animal species and two on cars, but no ecology, climate, or agriculture. Science data sets are mostly historical. In any case, I highly recommend this book. I learned a lot from it and it was a joy to read. The second chapter is very useful when producing data graphs. The first chapter helps to get more out of looking at graphs, besides already mentioning many important issues for graph production. The focus of the book on understanding makes it an excellent and unique reference that I will happily pass on to my students.