Visualization is a medium, you should treat it as such
Delaney Turner 270003RQ8K Delaney.Turner@ca.ibm.com | | Tags:  ibmsoftware information-insights
0 Comments | 2,962 Visits
The following is the third in a new six-part series on Advanced Data Visualization. Over the next three months, IBM visualization experts will explore new and emerging visual techniques and the underlying technologies you can deploy to better understand your data to transform insights into better business outcomes.
Frank van Ham is a well-known research scientist and an IBM Master Inventor with over a decade in experience in designing and deploying interactive information visualizations. Some of his past projects include Many Eyes, a site for collaborative visualization and SequoiaView, a visual disk browser. Frank currently works with the IBM Business Analytics division on integrating visualization into IBM's product portfolio.
In a past blog post, Graham Wills argued that visualization is not just a novel interface technology, but it’s a communication medium in itself. Communication media are not content neutral; their impact very much depends on the person composing the message, the form of the message, as well as the intended recipient of the message.
Once you realize that information visualizations are subjective carriers of a particular message, there are a lot of lessons from communication theory you can apply to your daily visualization design. In this post I will touch on a couple of these.
Know your message
For every data set there are many different messages I can choose to highlight in a single visualization. You'll be unlikely to find a single visualization that conveys all features of your dataset in a single image. Designing an effective visualization usually begins with making choices and deciding what the message is that you want to convey to your viewers.
The flipside here is that this means that you also have to decide which features of your data are irrelevant to the message you are trying to convey and eliminate them from your design. Another factor involves deciding what you want to achieve by showing that particular feature: When someone is doing data analysis, they likely want to obtain and objective perspective on the data and have that perspective be as detailed as possible
If I want to present a set of findings to some of my stakeholders to convince them to pursue a particular course of action, I might want to make sure that they’re only seeing the high level message I intended to communicate and not all the nitty-gritty details.
Know your audience
Like any other communication, many aspects of visualizations are dependent on the audience that will consume your visualization. For example, if your intended audience has an already established mental map of the domain of your visualization, they will be very unlikely to accept an alternate but equivalent representation. The world map below might be a slightly contrived example, but especially when you are deploying visualizations with highly skilled domain experts, you’ll find that they often have their own set of mental models and are unwilling to change them.
Color is something else that is highly cultural dependent. The typical green/red color scheme for financial data might be valid in the western world, but in the Far East you’ll find that the scale is reversed.Similarly, if you’re building a visualization for a heart monitor, you’ll find that darker red is actually good because medical specialists associate it with well-oxygenated blood. Talking to your audience before designing and deploying your visualization will help in identifying these potential stumbling blocks.
It’s all good, stocks are in the green.
Be a skeptical viewer
Visual media can be much more convincing than other communication media, because humans are inclined to believe what they see. And exactly because visualization is a medium, you can distort the truth just as easily as in any other medium, sometimes without recipients realizing it. The most obvious occurrences of this are playing with the data axes (using a non-zero axis origin to exaggerate growth, an inconsistent axis scale throughout the graphic, or mapping a single value to area without clearly indicating this).
Doing a Google search for lying with visualization will yield you plenty of instances.Other less obvious cases involve using inappropriate accuracy for values that have a high amount of uncertainty. Graphs that show forecasts should indicate the uncertainty for some of those predictions.
Graphs that visually show results from a sample or poll should include visual representations of the uncertainty (instead of listing the sampling uncertainty in tiny font beneath the graphic, I’m looking at you, pollsters). In many cases, it’s likely that you are being fed one particular interpretation of the data to support (or create!) a story and it’s important to be mindful of these cases.
If my unit of communication is a sentence, I can try to cram as much information in a single sentence as possible, for example by adding extra clauses to a sentence or by adding concatenations, which often results in sentences that carry a lot of information but are not easy to digest, because the reader has to parse the structure of the sentence to be able to understand it.
In visualization design it might be tempting to have a single graphic show multiple aspects of a single dataset, but in many cases it’s better to break down a complex message into multiple sections. My colleague Graham Wills has offered a number of approaches to reduce chart complexity in a previous post .
Chart embellishments are another feature that should be used with care: It pays to have a graphic that grabs the audience’s attention, because a message that is not received is not a very effective message by definition.
On the other hand, if the features you are using to engage the audience obscure the message you are trying to convey, you will not communicate effectively either. In this post I’ve argued that visualization should not be seen as a content neutral information carrier, but rather as novel communication medium.
Almost all of the caveats that apply to more traditional forms of communication (text, speech, images) apply to visual representations of data as well. Next time you view an infographic or a visualization, try to distill the message the authors was trying to convey consider if that message is conveyed effectively and more importantly, accurately.
Continue exploring visual analytics on IBM Many Eyes
Why stop the insight with this article? Visit IBM’s hub of visual analytics, IBM Many Eyes, and join over 100,000 like-mined visualization enthusiasts, academia and professionals. The Many Eyes web community democratizes data visualization by providing a simple three step process to create and interact with a visualization using your data set. Then share or embed your visualization across the web or your social network.
Read previous entries in this series: