I propose the following law:
"The longer an innovative visualization exists, the probability someone says it should have been a line/bar chart approaches 1"
I’ve seen the "shoulda been a line chart" suggestion many times, most recently in Randy Olson's excellent discussion of an award-winning visualization from the The Wall Street Journal. However, discussions like this forget a fundamental challenge when we communicate with data: Getting people to engage is sometimes as important as building the cognitively most valid method.
Here's how things unfold. Somebody publishes an innovative visualization that communicates a really powerful story (note that by "innovative" I don't mean deceptive: I mean visualizations which use correct methods but in an unusual way). The originality and clarity of the design catches people -- they get the message, learn something and share it. The visualization goes viral: People engage with the original, striking style, and as they share it, more and more people are informed and educated.
Which one of these is the correct way to tell the story of vaccinations? Which would you engage with?
In the WSJ diseases charts (left), the designers chose not to use a straightforward timeline. Instead, their use of color, a grid layout and a thick black line punches home the message: Vaccination virtually eradicated serious disease.[1]
Following the viral success comes reflection. Visualization experts begin to look at the design and question it. Thus begins the end phase and it ends up with someone ultimately saying, "Yeah, it's a great chart, but it would have been better as a line chart."
No. It wouldn't have been.
Saying it should have been a line chart forgets two important aspects of communication which are sometimes as important as complying with the "rules" of data visualization.
Data storytelling can be beautiful as well as functional.
You and I have seen tens of thousands of line charts. You may have seen 20 already today. The novelty of the WSJ design grabbed people's attention. The novelty created the success in a way a line chart, even if it told a story in a purer way, would not have done. What was your first reaction when you saw the WSJ viz? I was drawn in by the novel, beautiful design, and then further delighted by the density of data I could explore. The impact of vaccinations was punch-in-the-face clear. The very fact that it was not a timeline was what made me want to interact.
This doesn't mean all visualizations should be novel. We'd be overwhelmed if everything was novel. My point is that saying "it should have been a line chart" ignores the initial response people had and the reasons they wanted to interact and share it.
On a recent episode of What's the Point, Giorgia Lupi expressed this perfectly when she said, "Beauty is a very important entry point for readers to get interested about the visualization and be willing to explore more. Beauty cannot replace functionality but beauty and functionality together achieve more. Beauty is an asset."
This doesn't mean you should never produce a line chart, but would the WSJ article have been so successful had they done it Randy's way?
The perfect chart does not exist.
Randy acknowledges this in his article. We both agree you need to craft accurate charts and focus on the story. A rich dataset can tell many stories. In this case, even when you have chosen the story you want to choose ("vaccinations end disease"), it can be told in many different ways (line chart or highlight table).
Randy's line chart hits a home run in terms of reducing the data to a clear, simple articulation of the data. However, the WSJ story shows the impact of vaccinations equally well, with the added advantage of displaying as much of the data as possible.
I really value Randy's critique. Of course we should encourage rethinking and reworking of great visualizations. Enrico Bertini, assistant professor at the NYU Tandon School of Engineering, said recently at Tapestry, the data storytelling conference, "Use data visualization to create ideas not truths."
Instead of saying what a chart should have been, we should explore what other stories the dataset could say. This doesn't make one version right or wrong, it just shows new perspectives. My MakeoverMonday project with Andy Kriebel reinforces this: Each week we ask the community to remake a chart and tease out new perspectives from simple datasets.
If Tynan DeBold and Dov Friedman had told the story with a single line chart, would it have been so widely shared? Would it have won awards?
[1] I do agree with Randy's point that their choice of color palette could have been better. I also understand that the color palette was imposed on the design from people outside the WSJ's graphics team.