Information is Beautiful - Review
Alex Lea, Research Manager, Research and Information Team, Leicestershire County Council
How do we make complex and not so complex data more accessible to non-technical audiences? Moreover, how do we draw out key messages from data so that the right people can get to grips with the key issues they need to do their job?
Our role as a research team in a local authority is built around these questions. We need to present data in a way that not only gets the audience’s attention, but also leads them to the conclusions that they would come to if they had the time to sit down and work through the data themselves.
Which they just don’t have.
But how do we present this data? What can we do to make decision makers sit up and take note and make sure that they base their decisions on sound evidence?
People are often scared of data, or don’t understand it. That or they just switch off, due to the statistics overload that we are subjected to with the internet and 24-hour news. How do we make sure they pay attention to the things that we, as researchers think are important, and that the data is showing us? If the data is presented well, then it should speak for itself. We merely amplify the messages through effective visualisation. But how can we do that? In some ways the possibilities are endless.
We don’t like pie charts. So there’s one less thing to worry about.
But there are an endless number of ways to display data, as David McCandless demonstrates in his book Information is Beautiful. There are bar charts here. Along with line graphs, bubble graphs and treemaps. Used sparingly and in conjunction with other types of charts and infographics in order to take complex datasets and tease out the key messages. Information is Beautiful is concept design for what we do at work. It’s not meant to be a guide to how we can display data, but it shows us what can be done to make data more visual and easier to understand. As McCandless says, it’s a map book, not a geographical map, but an information map to help the reader navigate through the terrain of the data.
If you’re looking in the book for hard facts, you may want to go elsewhere and find more reputable data sources. The sources used are often questionable (Wikipedia) or trivial (TV and Movie time travelling) but in many ways, the data itself is not the important thing - what matters is the way it’s displayed.
Imagine you were presented with a dataset containing a few hundred entries relating to various dietary supplements, minerals and vitamins. Presented with the raw data, you could look down and across and get a picture for which ones have proven health benefits and which don't, which are used by large numbers of people and which aren’t. You would however, only have a partial view of the pattern developing.
Now plot each entry on a page from high to low based on the proven benefits. Now make each point into a circle whose size is dependant on the number of people who use the supplement etc. As if by magic you have a complete picture of the health benefits and popularity of around two hundred supplements. You can instantly see that the more proven supplements have higher usage (generally speaking) and that those lacking evidence are less popular. However, you can also make out the outliers;
Why aren’t more people using evening primrose oil for PMS? Why are so many people using ginkgo biloba and ginseng?
On this basis, why am I still using ginseng?
Information is Beautiful is full of different ways to display data. Within our team, many of these ways are already familiar with us and those that read our work. We’ve used consensus clouds for years and the technology is out there to use for free, right now. Non-researchers have even started using it, which probably means we need to start upping our game.
Flow charts can be put together easily if clumsily in off the shelf software and heat maps can be created at a basic level using spreadsheet packages and conditional formatting. I suspect that most of the graphics in the book have been put together using a graphics package, and in some cases, I would question the general accuracy, but it serves its purpose. If you really want to know the figures for stocks of Atlantic Cod, go and look at the data or ask a researcher to tell you. But if all you want to know is the key message - can we still realistically continue to fish it- then look on the graphic.
Answer, “No”.
That’s the key message. That’s what policy professionals and decision makers need to know. That’s what people want to know to sound informed around dinner tables up and down the land.
Information is Beautiful isn’t a textbook. It’s not necessarily going to give you cold hard facts on interesting and not-so-interesting subjects such as coffee-related drinks and daily water usage, types of facial hair and wine vintages. What it does it take you gently by the hand and points out the interesting bits in a visually stimulating way.
The book isn’t without its flaws. Some visualisations are just plain strange (Types of Facial Hair) and it’s not quite clear what they are meant to show. Strictly speaking, if a dataset can be explained in a few words or sentences, you don’t need a visual. Here McCandless wastes two pages pointing out that dictators often have facial hair and kill large numbers of people (and the data on genocide is incomplete).
At times you do get the impression that in a drive to present data in a visual way, McCandless has focused too much on the visual and not enough on the data. In that respect, the book at times strays too much towards the artistic rather than the informative (Food Colourings Linked to Unpleasant Deaths or Creation Myths) whilst others are too data driven and not graphic enough (Dictators’ Wives is just a table peppered with some small graphics to illustrate shoe ownership, amongst others). Others are simply missing labels and texts, although I’m assuming that is a printing error (Note: I’m thinking of pages 62-64 - Alex). It’s sometimes a mixed message which leaves the reader scratching their head, but given the often light touch of the author and the sheer number of different types of visualisations (see page 128), a few duds can be excused.
Information is Beautiful treads a fine line between a study of how it is possible to display data and a coffee table art book. In some ways it often misses on both counts; too arty in places to be practical for professionals, too technical in places to be interesting to the layperson. But taken with a pinch of salt and an open mind, Information is Beautiful is a prime example of why a default bar chart is such a travesty. Convincing those that consume and use data that there are better ways to display data which make their job easier is a difficult task. Probably because it involves time and effort, and moving out of your comfort zone.
People like pie charts. They feel at home with them, even though they are terrible in almost every way. McCandless shows that data can be interesting, patterns can be seen and conclusions can be made in an effective and unthreatening way.
And all this without the use of a single pie chart.*
*I’ve checked and he doesn’t.