In the last post, we promised to share some tips from Edward Tufte, the recognized pioneer and master of graphically displaying information.
Tufte has produced several books on graphical excellence that bring to life the gap between good and bad data visualization far better than a blog ever can. There are, however, a few of Tufte’s maxims that are both easy to share and highly relevant for asset managers using graphical representation in their client reporting.
Show me the data
Tufte’s whole philosophy subscribes to the “less is more” school of thought. When it comes to data visualization, the most important thing is the data, so let the data play central stage without clouding the issue with pretty pictures or unnecessary embellishments.
Sometimes, that means not using a chart at all. When you’re talking about small sets of numbers, tables often work just as well, if not better, than charts or data graphics.
Data visualizations are fantastic at conveying complex things succinctly but don’t be tempted to overuse them. Turning a simple chart into a complicated graphic contradicts everything that data visualization is meant to achieve – clarity, simplicity and precision.
Focus on the facts
Focus on the key facts contained within the data. By graphically representing a set of numbers, you should help people get to the message they need to see more quickly than they would otherwise.
Data distortion is deadly
Beware the danger of data distortion. Our minds play tricks on us — that’s why optical illusions work. At every stage of producing data graphics you should sense check whether the data visualization in front of you still conveys the truth revealed by the data itself.
The purpose of data graphics is to “reveal the data” not muddy the water with distorted presentation.
Squashed up data = second rate interpretation
As with most forms of graphical display, empty space is a hugely underrated commodity. Whether it’s removing gridlines that serve no purpose, dampening backgrounds to let the data take center stage or simply ensuring that data labels can be easily read, adding space for the data to breathe is essential.
Data visualizations really come into their own when conveying massive data sets that would be impossible to succinctly convey in more conventional graph — so long as you don’t start squashing too many facts into a small space.
As Tufte says, “every pixel should speak for itself.” Every bit of your graphic should add something to your story — or be removed.
Excellent statistical graphics
Of course, Tufte would be far more likely to illustrate these points than merely write about them. In The Visual Display of Quantitative Information, a book he self-published originally in 1982, he provides a huge collection of charts, images and data visualization graphics to explain what he considers good and bad statistical graphics.
According to Tufte, “excellence in statistical graphics consists of complex ideas communicated with clarity, precision, and efficiency”.
He also hates – for obvious reasons – data visualizations and charts that lie.
The phrase “lies, damned lies, and statistics,” popularized by Mark Twain, surely exists for a reason, and while lying might seem an emotive accusation to throw at chart producers, the risk of distorting data is very real and highly dangerous.
“The representation of numbers, as physically measured on the surface of the graphic itself, should be directly proportional to the quantities represented” Tufte said, and developed his own “lie factor” equation to test a chart’s integrity:
The Lie Factor = the size of an effect shown in a graph divided by the actual size of the effect in the data on which the graph is based!
Pie charts are less effective
The ability of statistics, graphs and charts to hinder rather than help understanding is a topic that has held people’s interest for ages. But just this week attention has re-focused on the use (and abuse) of pie charts. On Monday, Walter Hickey published an article, The worst chart in the world, calling for more scrupulous use of the pie chart. I’d strongly recommend that you go and read the whole article but, in brief, his argument against pie charts goes something like this:
1. The “point” of a pie chart is to show the relationship of parts to a whole – something they are actually remarkably bad at doing.
2. Most people find it harder to compare slices of a circle than they do the rectangles found in bar charts. Hickey illustrates this by asking how many people can quickly put the following segments in size order?
The reason we find it hard is the same reason “you probably found trigonometry and radians a lot more difficult than you found basic rectangle geometry,” Hickey concludes.
3. Any 3D pie chart is especially prone to distorting data because altering the perspective makes some segments appear larger than they actually are.
4. Pie charts are usually very hard to interpret without data labels showing percentages and Hickey believes that “the most elegant charts don’t require data labeling.”
5. Finally, there is one thing that a pie chart is very good at – comparing two or three different data points with very different amounts of information.
The Washington Post followed up with an even simpler summary of Hickey’s points, conveyed in two pie charts:
The same data displayed in a 3D pie chart looks very different:
Tufte would agree, as demonstrated by one of his recent tweets:
Pie chart users deserve same suspicion+skepticism as those who mix up its/it’s, there/their. To compare,use little table, sentence, not pies. — Edward Tufte (@EdwardTufte) January 10, 2013
But to be fair, not everyone agrees. Here are just two examples of the many articles written in response that defend the pie chart: Pie charts are fine and you all need to calm down and In defense of pie charts.
But for a deeper dive into applying graphic standards to real-world examples, see the excellent book The Wall Street Journal Guide to Information Graphics by Dona Wong, a student of Tufte’s at Yale who went on to lead the information graphics department at the WSJ.
With so many different opinions and options, fund marketing teams need to do their homework. If you aim to create factsheets and client reports with true graphical excellence, making visual sense of complex financial data, learn from the masters.
Next time, we’ll be returning to Tufte, honing in on graphical integrity and discussing how to ensure that the charts you put before clients really do say what you want them to.