By Stephen Hampshire, Client Manager, TLF Research
Review in a nutshell
Excellent primer on the key principles of accurate and effective data presentation. Useful summary of the graphical toolkit available to designers. Lacks focus on the importance of getting the right data in the first place.
Data visualisation (dataviz), information design, and infographics are a big trend at the moment. But how many of these "infographics" really add to our understanding of the data that lies behind them? The best bring insights that were invisible without them; or communicate a message in a clear, powerful, way. Many do nothing but ornament a simple factoid with unnecessary graphical embellishment. The worst get in the way, and distort or conceal the truth.
Brian Suda's book is in many ways a modern, graphic-focused, version of Darrel Huff's classic 'How to lie with statistics'. As much a "what not to do" as a "how to". Reviewing it was a slightly surreal experience. I read with an uncanny sense that I had somehow got hold of a copy of my own, as yet unwritten, book. Somehow Suda had managed to plagiarise my future self. On reflection, this probably owes more to a shared reading list than to time-travelling.
In many ways, the key strength of Suda's book is not that it offers much that is new (it doesn't); but that it provides a very clear, layperson-friendly, review of key recommendations from the likes of Tufte and Few. That doesn't mean it's only for people new to thinking hard about presenting data - it's a useful book for all of us - but it definitely has more to offer those who are not already familiar with the literature.
It's also worth saying that the book itself is very nicely produced. Mark Boulton's 'Five Simple Steps' publishing house now offers a small range of books focused on different aspects of design, all of which are themselves designed to a very high standard. But what about the content?
Although it's divided into 5 parts, to my mind the book splits into 3 sections:
- The principles
- How to lie with charts
- Chart gazetteer
The overview of principles contains nothing that will be new to anyone who has done their reading on information design. The influence of Tufte and Few is obvious. That's no criticism though, the principles are sound and it's useful to have them collected and summarised. For anyone not familiar with the works of Tufte and others, this offers a clear and accessible introduction...ideal for leaving on your boss's desk, perhaps? For the rest of us, it offers a nice summary and some new thoughts - for instance on using MD5 hashes to create unique colours programmatically.
How to lie with charts
Suda offers a good summary of well-established tricks like using implicitly 3D bags instead of bars, or chopping axes.
He also covers some more subtle cheats, like the difficulty of balancing false negative and false positive errors. Anyone wanting an informed opinion on the debate over breast cancer screening would do well to read that section.
Two sections of the book are dedicated to discussing the features and uses of a range of common and not-so-common charts, from line and bar charts to "sound charts" and Chernoff faces. I found this the strongest part of the book, with excellent coverage and useful advice on when to use (or avoid) each chart type. Suda also brings good practical advice on some additions and tweaks that add value to each chart type in certain situations.
The missing section: what to chart
One of the relative weaknesses of the book is that it concentrates on how to chart at the expense of what to chart. In his introduction, Suda says "The main purpose of this book is to visualize and design for data in such a way that it engages the reader and tells a story rather than just being flashy, cluttered and confusing." This is an excellent goal, but by deliberately ignoring the data itself, Suda has missed at least 50% of information design as a discipline.
When he talks about "Sins of omission", for instance, I would expect a discussion of the importance of correcting for inflation when showing the long-term trend of anything to do with money. Charting bare financial data without accounting for the fact that ¬£10 in 1970 bought you more than ¬£10 in 2011 is far more misleading, in my view, than using silly 3D pictures of bags of money or chopping the axis.
Good information design starts with interesting data. It's about releasing the story in the data, not embellishment. In my view there are three stages to producing a good infographic, and Suda only considers the last in detail:
1) Get good raw data
2) Treat and analyse it to tell the right story
3) Display the story in the most compelling way
Suda has deliberately avoided any discussion of this, but the book ends up feeling unbalanced as a result. Nonetheless, this remains one of the better summaries of the toolkit available to information designers in the 21st century. I'll definitely keep it as a desktop quick reference for whenever I can't quite find the chart I want to tell my story.
Edward Tufte, "The Visual Display of Quantitative Information"
Recently revised and updated, this is an uncompromising classic.
Stephen Few, "Show Me the Numbers"
Elegant and to the point, with a greater focus on business.
David McCandless, "Information is Beautiful"
Flirting with the "decoration above insight" end of the dataviz spectrum, but helps you to think beyond charts.
Darrell Huff, "How to Lie with Statistics"
A classic, but still very relevant. The charts may look prettier now, but the tricks are mostly the same!
Edward Tufte: edwardtufte.com
Stephen Few: perceptualedge.com
David McCandless: informationisbeautiful.net
Nathan Yau: flowingdata.com
Kaiser Fung: junkcharts.typepad.com