Friday, February 1, 2013

Storytelling with data visualization: Questions and challenges

I've just finished a conversation with Robert Kosara at the Computation+Journalism Symposium 2013, at Georgia Tech. It was fun. The title of our panel was Storytelling with Data Visualization: Platforms vs. Craft. We both started with some remarks. I posed several questions that I believe will be relevant for the future. Take a look at the slides above. Keep reading for a brief summary of what I said.

• Slide 1:
Storytelling is a buzzword in visualization and infographics nowadays, but I have serious concerns about how the word is used, about how carelessly we talk about it. So my first question is...

• Slide 2: ...Do we really need storytelling in visualization? If we do, what kind of storytelling? The very word "storytelling" means different things to different people, and many have expressed valid reservations about it.

• Slide 3: I am a storyteller, a journalist, and a designer. Storytelling is what I do: I get information and I arrange it into narratives. I do infographics for a living. I also teach how to create them. But I see a challenge: It seems that audiences tend to trust pictures, graphs, maps, more than they trust text. So perhaps visualization and infographics designers need to be even more careful than other kinds of journalists. Are we doing enough to guarantee that what we present is accurate?

• Slide 4: Here are a few objections that some people have posed: Stories can deceive, stories force readers to see mainly what the designer wants them to see, storytellers tend to value anecdotes too much, etc. Besides, if you are an outsider (a scientist, a statistician, a cartographer) and you want to learn storytelling techniques, you will probably stumble upon...

• Slide 5: ...Books like these, with titles that include words like "convince", "impact", "persuade". Where's "accuracy"? Where's "depth"? You may say that persuading and convincing are not primary journalistic values, that we value efficiency, seriousness, and accuracy above all, but are you sure of that? Have you read a newspaper, a magazine, or a news blog lately? Or watched the evening news?

• Slide 6: At the same time, I believe that journalists are not really aware of cognitive biases: The human brain tends to impose narrative structures to unrelated data, it tries to see patterns and trends that may be inexistent, it infers causation when it perceives concurrent phenomena. We have to fight against those urges using the tools of science and analytical thinking.

• Slide 7: I don't have to remind you of Jonah Lehrer, right? His last book, Imagine, was withdrawn from bookstores because he made up quotes, but I believe that this was not his main journalistic sin. That was just minutia. What Lehrer and other well known writers don't understand is that there's a huge danger in trying to lightly weave sparse data and preliminary research into cohesive narratives. This is, unfortunately, what magazine, newspaper, and TV editors and producers demand: "Get to the point; tell me the story!". Well, sometimes the "story" is multifaceted and nuanced. We should reflect that complexity in what we present. This can be applied to all kinds of journalism: Written, spoken, visual, data driven, etc.

• Slide 8: Simultaneously, researchers tell us that we cannot flee from storytelling, because stories match the way the human brain understands information. No matter how hard you try, you cannot avoid telling stories to yourself.

• Slide 9: The other side of the coin is that visualization designers who try not to build stories or linear narratives, but to create tools for readers to discover their own stories, also face risks. How do you know if your audience is getting things right if they are left on their own, for instance?

• Slide 10: Yesterday, Philip Meyer, of Precision Journalism fame, said that we should not teach J-students tools, but how to build those tools. What does that mean? Do all J-students need to become full-blown coders? Or can they learn the logic of coding by studying some basic-level scripting language? Or is it more about conceptual skills?