SXSW: the most influential visualisations of all time

thevizzes

For the slides – head to the end of this post

For Tableau-versions of the vizzes, head to this post

What do you think are the most influential visualisations of all time? I asked that at SXSW today: what a great time it was delivering the session to everyone. If you want to watch the session again, you can watch an older version I recorded with Data Science Central.

I’ve also uploaded the Tableau workbook I used to recreate the visualisations (see this post). This in itself was an interesting exercise. Do they look better in a modern data visualisation tool? I don’t think so (as you can see in my exploration of Nightingale’s chart earlier this week). What the originals have is the depth of the personal touch. Being hand-drawn they contain the soul of the author of the viz. Each dot of ink will have been considered carefully:

–          What layout will best communicate my message?

–          Should I use colour?

–          What thickness should the axis be?

Today, not enough people stop to take the time to make those considerations.

My slides are also available on slideshare. And if you want to go do some further reading, I have a bundle of links on Bitly.

I’m interested to know what you think of my choices and if you would choose different ones.

The influential vizzes. In Tableau

So: can you recreate my list of influential vizzes in Tableau? Sure. Mostly. Here’s an interactive gallery. Whether you SHOULD recreate these vizzes or not is a discussion in itself. I leave it to you to debate this in the comments below. The supporting slides and more info can be found in this post: “SXSW: The most influential visualisation of all time“.

William Playfair

Playfair nailed it. It’s a great viz that also looks great in modern tools. Well labelled and nicely designed. What can I add? Um – interactive tooltips?

Florence Nightingale

Tableau can’t do radar chart or rose diagrams. That’s probably a good thing. I tried many different views but the one that had the most impact, I believe, is a stacked area chart.

Charles Minard

Thanks to Kim Rees of Periscopic – she did this one. It’s great. Again – Minard’s has more soul.

Hans Rosling

Rosling’s Gapminder is a modern tool. Tableau doesn’t animate so well in the browser, so while we can approximate the experience it’s not nearly as smooth an experience as the real thing.

John Snow

This one needs animated pages to work properly. My verdict? John Snow definitely did a better job

Florence Nightingale didn’t need to worry about best practice

Florence Nightingale’s rose diagrams are one of the most famous and influential visualisations of all time. The story behind her and the creation of the charts is a fascinating one. I tell it in my “Influential visualisations talk”.

As part of the talk, I talk about how people criticise her diagrams for not being best practice. Phooey, I say to them. Why? For two reasons:

  1. She didn’t have the benefit of books and blogs about data viz best practice.
  2. She was trying to convey a simple message: soldiers were dying needlessly. The message was what counted to her.

I got to thinking, though, what would the charts look like using “best practice”? The answer: boring. Sure, they’re accurate, but they aren’t engaging. And it brings to mind something Andy Kirk and Alberto Cairo often talk about: while scientists tell us what to do, creatives tell us what we could do. Nightingale was a creative data rockstar.

So here are my attempts at reworking them. Do you think any of them are more effective at communicating the message than Nightingale’s?

area percent of total
Area chart as a percentage of total? Not bad. Lots of blue. But boring,
area line
Area chart of the numbers. I like this one, The text floats nicely and boy, there’s lots and lots of blue (avoidable deaths)
bar
Bar chart. Very effective and removes time from the view – many people argue that time is redundant in the originals. It’s not very engaging, though!
pie
Pie. Dare I admit I quite like this one? It really emphasises just how much blue (avoidable deaths) there are in relation to the whole.

As a final note – I concede I’m a little confused about whether Nigthingale’s originals were sized by area or by distance along the radius. Two great articles seem to contradict this (at least as far as my tired mind can interpret). One article that suggests the area is correct is on the excellent Understanding Uncertainty blog. Whereas Dynamic Diagrams thinks the area is incorrect.