21 Lessons from David McCandless

On April 21 I spent a half-day at a hotel in Soho listening to David McCandless, the London-based data journalist, designer, and author of Knowledge is Beautiful and Information is Beautiful. The workshop was a bit pricey, but tax-deductible for self-employed folks like me, and when I listened to David and talked to the creative directors and data geeks around me it was clear that I had found my tribe.

David’s manner is open, low key and thoughtful. This is a guy comfortable in his own skin, nothing to prove, no need to hear the sound of his own voice, plenty to show and say, doing his best to be of service to the audience.

I took notes. Not very good notes, but from them came this list of 21 tips from David:

Rules for visualizations

1. The word media comes from mediate. That’s what a good visualization does: mediate between facts and audience.

2. A good visualization has four hooks: trust (which comes from data), interest (from the story), a goal or purpose (which cuts extraneous detail and makes the visualization efficient) and impact (which comes from the design).

3. Four more elements: the what (the data or information), the “exactly what” (the concept), the why (goal or purpose) and the how (the design).

Finding your story

4. Start building your visualization with a question. (Here’s one: Do horoscopes all say the same thing? It led to scraping text from horoscopes and doing a 12-sign word cloud in the shape of wheel, with the shade and size of the text indicating amount and frequency of the advice.)

5. The biggest story is always fear. (It’s that fast thinking again.)

How to

6. The sequence: First an idea or question, then a concept (idea rounded out and made explainable), then an iterative circle of sketching, researching and designing.

7. Start the design with a sketch, the rougher the better. Keeping it rough allows you to iterate without fear.

8. Stay in sketch mode as long as you can.

9. Even if the type of chart seems obvious, sketch your data in different ones. The star chart. The bar chart. The bubble chart. The cycle diagram. Keep on going.

10. To make the data more interesting, ask more questions. Add more layers. Repeat as necessary.

The data

11. Once you understand the data, the rest is easy. 80% of the work is the data work. (It’s 90% in the suffering-statistics-storytelling breakdown I wrote about here.)

12. When you have data left over, consider how it could be encoded and added to what you’ve already got.

13. Amounts may be interesting. But comparisons among amounts is where the story lies.

14. Six dimensions, often conflated, of what people call “big data”: gathering, handling, structuring, examining, discovering and delivering.

The qualitative data

15. Use concept maps for qualitative data. (Concept maps have the advantage of being able to hold contradictory ideas.)

16. Visualizing quantitative data is all about finding a structure. Once you’ve found a structure, you can visualize it.

The image

17. Consider which variables map to which elements of visual language (color, shape, position, pattern, frequency, proximity, opacity, etc.).

18. Anything that requires a legend to explain is unnatural. No legend, natural. Unnatural isn’t bad, but take it too far and the visual becomes distracting decoration.

19. Think about where you’ll want to zoom in (for detail) or zoom out (for context and comparisons).

Play

20. Learning is all about playing. Playing is all about trying things out.

21. Play. Play. And play some more.

Author: Dan

A writer, editor and aspiring web programmer, hoping to turn the madness of crowds into the wisdom of crowds.

Leave a Reply