Topics in Dataviz: A Primer for Getting Started
People in the Data Visualization Society Slack channel have been discussing their personal rules for making great charts. Such a broad topical naturally welcomed an amazing variety of responses. What follows is an attempt to highlight the main points which emerged during the discussion. Many of these are topics of significant debate in the visualization community, and we will try to dive into each more deeply over the coming months.
The first draft of this article had numbers next to the sections below. But we found that this is counter intuitive as different people may approach their work in a different sequence after the foundation stage. We recommend that you read this article in the sequence that resembles your own workflow.
Know thy data …. and own the story you can tell with it.
Data is the material we have at our disposal. Knowing the data is a vital part of determining which story you can tell. A good question to ask yourself is whether you know the system the data describes well enough to detect odd trends or errors.
The story you can tell
Humans are driven by the search of a story. A story helps our brains put things in a broader context and predict what may happen next. It reduces stress levels and helps focus attention. Based on your data, what story can you tell? Will you be the narrator, or will your audience make up their own story based on what you give them to explore a topic.
Ask brutal questions
Try to answer the fundamental questions — Who, What, Where, When, Why, How. Then stop and ask the hard one: Why should anybody care? Don’t take it to nihilistic levels but try to answer it before moving on; it will save you pain and heartache down the line.
I think it’s important to begin with basic editorial questions, as you would with writing. Every piece has an intended audience, with their assumptions, technical background, and needs. ~ Jasper McChesney
A proverb in the writing community compares a story to a dog. A good dog has a head, a middle and a tail. I’d argue the same is true for visual communication. When you are writing the reader follows your story along a single line you created for them.
When working with a visual medium you are dealing with a different beast altogether. 2D space is pretty flexible but establishing flow is tricky. But the foundations are very simple. You need to start to think in levels and layout, they can be your biggest friend or your biggest foe. I would recommend trying to map your story to different scales very early in your project.
- Structure your infographic in such a way that somebody that looks at it for 5 seconds can take away the major point you are trying to make.
- Pick a layout that guides people on how to consume the story you are trying to tell. Should they start at the top left and end at the bottom right, should they explore it by going from the center to the various edges and back?
Visualization is the transformation of abstract information into a visual system. The best visual systems come with context built in. The easiest way to communicate context to your audience is via metaphors.
If you are using multiple plots to tell a story, make sure that you are using the same/similar visual language for shared components. Keep the background and mindset of the audience in mind before selecting an element. Yes, everybody loves the idea of cat emojis as glyphs, but they may not be the best choice for work you are doing for the World Canine Society.
Make sure you use an appropriate color schema. One thing to consider is the cardinality (number of sets/groups/elements) of your data. Make sure you pick a color schema that has the same number. If you don’t, chances are you won’t make full use of the relationship of colors selected for the color schema. Other than that I would recommend playing a round of color golf. It’s a game where you successively remove a color until the visualization has become incomprehensible. Oh and one more thing, when choosing a color schema keep in mind that not everybody sees the full spectrum.
I try to think beyond the chart — my dashboards are usually a composition of charts, so I need to ensure they all work together to create an effective message that is accessible and understood by the end user/audience. ~ Bridget Cogley
What will it feel like
Before you start to dive into doing, consider preparing by sketching our what you have in mind and explore ideas. Sketching is great because it allows you to quickly explore ideas, make variations, drop them and start fresh. Another aspect sketching can help with is information density. If your sketch looks too cluttered, split it. There is another aspect that sketches can deliver where code has a harder time, you can let your imagination run wild.
I was born in 1935, and as far back as I can remember, I was sketching designs. My first subject was an aircraft, which I imagined myself piloting. — Norman Foster (Architect)
Interactivity doesn’t encourage people to explore, top-level representation does. Interactivity should do one thing and one thing only, support you in telling the story. If it becomes too much of a distraction you put them on the slippery slope that takes them from getting insights to getting entertained. Also, consider how the interactivity will age. Will the “cool transition” still help get your story across when the eye-catching novelty is gone. A brilliant example of how to do this in the right way is a recent article on the Japanese Cherry Blossoms (thanks Bill for sharing this).
The Visualisation should draw your audience in without text. But once you have peoples attention Text will help them understand what they see. Don’t think book chapters but short concise passages/labels/legends that help reinforce points or clarify. Consider Ernest Hemingway your go-to author for style and brevity.
No text written in Switzerland in the realm of graphics is allowed to skip the topic of Typography. You may think that this is a minor point: the wrong fonts can ruin a perfectly fine visualization. If you are not convinced I challenge you to use Comic Sans or Webdings in your next project update to your supervisor/sponsor/client. I think CERN is the only place where you could pull this off without a scratch and then only if you find a new particle.
Ah fonts! Do you think it’s common for people to think about the font they are using? I’m very conscious of it and assume everyone else is. However, I often come across fonts used because they ‘look nice’. Bad luck if you can’t read it 😳 ~ Wendy Small
Putting it together
The important thing is to find YOUR process, exercise it, review it, and update it.
Dataviz is not magic, it is planning, doing, failing/succeeding and showing up the next day to do it all over. Document what you do, experiment, and be ready to debate decisions you made. Allow for feedback cycles in your work. Be obsessive about details of one of the points above, or do the bare minimum. Take pride in your craft, but don’t take it too seriously. Above all else have fun!
With that I’d like to thank you for your attention and hope that this helps you get started on your next project, please leave me some feedback below or on twitter.
Have a lovely day.
It would have not been possible to produce the above post without the below-listed people generously sharing their insights and time.
Duncan Swain, Steven Braun, rlage, Nic Moe, Will Chase, Jacque Schrag, Jasper McChesney, Matthew Montesano, Regis O’Connor, Paul1980, Jason Forrest, Nicolas Kruchten, jameslytle, Wendy Small, Carlos Reinado, lord, Tricia Aung, terence, Corwin S, Bridget Cogley, Joey Cherdarchuk, Elijah Meeks, Lilach Manheim, Regis O’Connor, Enrico Berini, Adam Pearce, senthil, Daniel Zvinca, Zan, Viola Bernacchi, Carl Manaster, Wendy Small