Last month a group of Juicers attended a lecture at Georgia Tech entitled “More Than Insights: Beyond Exploratory Data Visualization” given by Hanspeter Pfister, Professor of Computer Science and Director of the Institute for Applied Computational Science at Harvard University.
Pfister cited the rise of the infographic, as well as an increased general interest in subjects like data storytelling and data journalism as evidence that more and more people are becoming interested in using visualization to communicate and explore information. But what comes after information is shared?
“After insight comes the message,” Pfister explained. “The information is the ‘what’, the message is the ‘so what’ - the ‘why should I care?’”
Being able to address the “so what” brings a whole new set of challenges to data communication, Pfister told the audience. He explained that we’ve only just begun to scratch the surface of what is possible, that we actually don’t know as much as we think we do about these subjects, and that much more research is needed to even begin to understand these intricacies. To illustrate his point, he used examples from three different subject areas: data visualization, data storytelling, and data tools.
Pfister cited a study that he had participated in along with Michelle Borkin on what makes a visualization memorable. In the study, participants were shown a string of various visualizations and told to respond if they remembered having seen it previously.
So what did the researchers find made a visualization memorable? The charts were found to be more memorable if they contained human recognizable objects (such as dinosaurs or faces), if it was colorful, visually dense, or had a title, labels, and/or paragraphs.
Are these descriptions setting off alarm bells and making you scream internally? It’s probably because these characteristics are the exact design elements we’re taught to avoid. To further prove this point, Pfister shared that the least memorable visualizations were what we’d think of as more “Tufte-compliant.”
So the question on everyone’s minds: do we toss out the old guidelines in favor of brighter, busier visualizations? Not necessarily. Pfister shared that he believes the answer may lie in “something beyond [Tufte] that we haven’t explored that much.”
Pfister then brought up the ultra-new method of using comics to communicate data. Ultra-new because, as Pfister pointed out, there are few actually using comics to communicate data, there is no real definition of what a data comic actually is, and there are no real tools to create data comics.
A data comic, he explained, is communicating data in a way that comic books typically communicate stories. He explained that the four essentials for data comics were visualization flow, narration, words, and pictures, and demonstrated how all of these work together by displaying a data comic that showed the various power struggles that contributed to World War I.
It’s hard to do the comic justice by just talking about it, but to give you some idea of the effect it had on the audience, I would like to use one audience member’s own words: “It’s like a punch to the brain.”
Viewing the information in the form of a data comic was a faster and clearer way to communicate the information than any textbook could have done. It was evident from this example that data comics are more likely to play a larger role in the future, but, Pfister questioned, how will it fit into data storytelling overall?
The last subject Pfister hit on was data tools. He explained how the majority of popular data tools are relatively easy to use, but lack ability to customize visualizations easily. On the other side of the spectrum, however, are tools that are more expressive but lack ability to add insight. He argued that data scientists not only want but deserve better tools, and because of this there should be a product that falls somewhere in between Excel and InDesign.
The answer that Pfister and a team of individuals, in collaboration with Adobe, came up with was a program in which the user puts data into a spreadsheet, then uses guides that constrain the data to create a visualization. It was an interesting way of displaying data, but will it satisfy data scientists’ quest for the perfect tool? Only time will tell.
It was clear from Pfister’s lecture that more research needs to be done in all of these areas before we can truly say for sure what the best methods of communicating data are. It’s an exciting time to be in visualization, and we’re excited to see what the future brings. In the meantime though, check out our design principles for what we’ve found to be some pretty effective strategies for communicating data.