Data Visualization Aesthetics: Why Information should be Beautiful

Aesthetics in data visualization can be a controversial topic. On the one hand we have the sage advice of Edward Tufte ringing in our ears about the perils of “chart junk”¹ that obscures meaning and impedes comprehension. On the other hand there’s people like David McCandless and his popular book/blog Information is Beautiful² which accepts making data “just look cool” as a valid objective. So what are the actual effects of aesthetics on data visualization and how important is it?

In terms of importance, C. Chen identified aesthetics as on of the top 10 unsolved problems of information visualization³. “Solving” aesthetics may indeed be beyond our grasp (and may be forever), but we do have some relevant information that we can examine.

When first viewing a visualzation, beauty arouses interest and provides an incentive for exploration. In our attention management economy, just getting people to look at the visuals can be sufficient motivation to focus on aesthetics. Once they are looking, the next question is what difference does the presence or absence of beauty make?

Nick Cawthon and Andrew Vande Moere⁴ used 11 different static representations of the same data and had subjects rank them for beauty and complete a number of tasks using them. They found that while perceived beauty was not correlated with improved efficiency (task completion time) or effectiveness (error rates), it was correlated with ‘erroneous response latency’. This means that users displayed more patience and spent more time trying to figure out the answer when they found the visualization aesthtically pleasing.

Finally, aesthetics has an impactd even when not using the visualization. Tractinsky et al⁵ found people perceived the usability of a bank ATM interface to be better if they found it aesthetically pleasing. This was both before and after using the system, and regardless of the actual usability of it. Think about that: people thought the beautiful interfaces were easy to use even when they weren’t!

This effect is something social psychologists and marketers have known about for a long time. In 1972 Dion et al found⁶ that when judging other people we tend to assume beautiful people also posess other desireable but unrelated traits. This is often referred to as the halo effect⁷. Interestingly, Reeves and Nass found⁸ that we treat our media and technology in a fundamentally social way — as if it were human. The exact reason for this is unclear. It could be due to a halo effect causing users to conflate usability and beauty; or it could be because the beautiful visuals elevate their mood, making them more likely to regard the system’s other aspects positively; or it could be some other reason. The net result, though, is that by improving the aesthetics of a product we can also improve its perceived ease of use.

Many of the points here about the importance of aesthetics for data visualization apply equally to any other media. Beauty will act as an incentive for exploration for a magazine cover as well as a chart. The halo effect has been applied in a wide range of domains and is well known in brand marketing. But it’s the unique challenges of data visualization that make these points especially pertinent. Data visualizations exist to allow people to understand data in ways they couldn’t otherwise. By their nature they ask people to slow down, look closely, and think rationally and critically — things people are often disinclined to do! We’ve been taking advantage of their effects on our perecptual psychology for a long time. By taking advantage of their effects on other aspects of our psychology we can further improve their performance.