Another blog post – another question. But this one might suggest that I’m losing my love of data visualisation. Losing the momentum that drives me to produce lots of work. An introspective question it might be, but don’t worry, it’s not an exasperated cry for help. Let me explain.
Earlier this year, not for the first time, there was a lot of talk about storytelling in data visualisation. I read about it extensively, I blogged about it here, I was invited to give a conference talk about it (and I did just that). I spoke about storytelling as a narrative, as an anecdote, as a news story, and spoke a little bit about the pros and cons of each definition. And then mostly, my talk became: “Who cares? Here are some great visualisations.”
But recently the debate has been raised again. Alberto Cairo tweeted this …
— Alberto Cairo (@albertocairo) September 11, 2017
Is he right? He usually is. I tend to think that if it’s open to so much discussion, interpretation and misuse then there must at the very least be caveats in the use of the term. Are so many others wrong, in different measures?
Then recent MakeoverMonday correspondence got me thinking. I’m always full of admiration for Andy Kriebel and Eva Murray running the project, because of their weekly commitment to (a) sourcing and preparing data (b) delivering it to the community (c) creating their own visualisations (d) blogging about the thought processes behind their own visualisations, (e) offering constant advice and feedback and (f) weekly recap with advice and favourites. There’s probably at least a (g) to (z) that I’ve either forgotten or didn’t even know about, their commitment is amazing. And I don’t always agree 100% with their advice in (f) but I know even if I don’t they are usually right.
But it occurred to me that the oft overlooked part of the above is (d). The community have very little to do: download, create, tweet, refine if they wish. And learn every week. But in amongst all the work needed, Andy and Eva still always do (d). Their vizzes may be among the simplest each week, but they are always instructive, and they will always say what they wanted to show, what question they wanted answering, and how they decided to viz as a result.
For example (excuse the paraphrases here – full versions and visualisations on the makeovermonday website
Andy (week 37 – UK bike theft), gave his goals as a list of questions he wanted to answer:
- What are the worst areas in the UK?
- Is bike theft increasing or decreasing overall and in specific areas?
- Are there as few positive outcomes as it seems?
- Where should I avoid locking up my bike?
- Is there any seasonality in the data? My hypothesis is that the number of bikes stolen would reduce in the winter months.)
And then he created a visualisation that answered them. You can’t argue against that being perfect preparation for a good, successful visualisation.
Eva (week 30 – how thirsty is our food?), gave a list of goals, showing her deliberate intention to focus on a specific issue and keep it simple:
- Mobile design
- Focus on two food items: Starchy roots (potatoes!!!) and Bovine meat (dead cows and calves)
- incorporate the story of Chris Voigt, who ate nothing but potatoes for 60 days straight, to contrast the differences between what you can consume on a plant-based diet and what you surely wouldn’t attempt to do with meat; include just numbers, bars and text
These aren’t stories, but they perfectly set out the point, or the goals, of the visualisations. Sometimes you can over-think to the extent that it seems they are pulling a small element of information from a rich dataset. So you may or may not agree they are telling a story, or at least the full story. But they tell you the point of their visualisation.
Another person whose work, and way of working, I admire is Colin Wojtowycz. A recent starter, he takes time to curate his visualisations but always explains what he is trying to do, and in doing so offers great learning and expands on the visualisation. Results are always good, but richer for this. you can read about his process here, and the point behind his visualisation here.
Colin’s points include maximising impact (the main point), improving on weaknesses in the original visualisation and finding a story. The latter being just one of the objectives in a series of points.
So you can have the best intentions of telling a masterful story but the minimum (and often all you need) is a point to your viz. Otherwise it is a waste of time and effort. The converse to this is a viz where data and/or charts are just “thrown on a page” because it was possible to do so. Viz without a point can be confusing, non-memorable, and, well, pointless. And pointless is no good unless you’re on a gameshow.
I want to start with an example of mine which is a bit like storytelling in reverse – a viz inspired by a story. Here’s my visualisation on the Ski Jumping long hill competition from Calgary 1988 (click image for interactive version).
There’s a real circular argument to this – the viz itself was inspired by watching the “Eddie the Eagle” film, a recent film inspired by the exploits of Eddie Edwards, the plucky British competitor who finished a distant last. So I decided to visualise the data. Is this storytelling? There are plenty of arguments to say that it is, but then even I’d be the first to admit that a one-page visualisation doesn’t have the same storytelling appeal as a two hour Hollywood movie (even if a certain amount of “poetic licence” is used in the storytelling of the latter). But there is a point. The points behind my visualisation were
- Show the disparity in distances including the impressive winning distances and the large gap to Edwards at the back of the field
- Demonstrate the trajectory of each jump in relation to each competitor
- A personal, technical point: experiment and demonstrate that I can trace out an arc that is a little different to what I’ve done before.
Although it does each of the three things above, it deliberately backs away on focusing solely on Edwards. That would be more like the “Hollywood” story. But I was delighted by online feedback from Matt Francis (and backed up by visualisingdata.com’s Andy Kirk)
I like the way that You have “hidden in plain sight” the star of the data by not highlighting it.
— Matt Francis (@Matt_Francis) September 18, 2017
I was particularly delighted, not just because it’s great feedback, but because that’s was exactly what I was trying to do. In this case, arguably the data tells the story without me doing so – I’m crediting my audience with being intelligent enough to find it (it’s not difficult) without the need to make it my story. That’s the point (pun intended!)
To reiterate, it’s not always necessary to state your point in the viz or the post/blog promoting it – it might or might not be obvious but so long as there was a point driving the viz it should help it improve impact or insight to the desired audience. Here’s another one of mine (again, links to interactive version):
I had several points here
- Try an area bump chart but on a dataset where it’s relevant and able to show interesting trends.
- Show that although Western industrial economies have been more stable on CO2 emissions in recent years, the overall total is continuing to grow.
- Highlight an important environmental issue
I think I succeeded – (a) proved to myself and a few interested parties that I could do this clever new chart type well (albeit by copying/adapting the great work of those before me who had devised the concept in Tableau), (b) the visualisation achieves this because the chart type does this nicely and (c) is done not just here but in a wider context by VizForSocialGood
VizForSocialGood is an amazing project which deserves more than the time I can give it. Go there. If you research the brief and immerse yourself in the project, your viz will have a point. A very good one. Really recommended for involvement with great datasets that make a difference. If you are looking for a genuine point behind your visualisations as well as an opportunity to make a difference for some high-profile organisations (and some equally important low-profile organisations that need help) then it’s the best place to start.
Professionally – my work involves the production (and reproduction) of dashboards which are produced as proof-of-concepts from teams of Higher Education professionals, with a wealth of data sources at their disposal. Every project and dashboard (or set of dashboards) produced requires a user story (for those of you who know Agile methodology you may recognise the term). Despite the use of the word story, this is not what I’d consider true “storytelling”, but it’s an excellent way of predetermining exactly the point of the visualisation. Like the examples from MakeoverMonday above, it encourages the participants to set out the goals of their visualisations, the questions they are hoping to answer as a result, and the context and reasoning behind these goals.
A composite / generic and fictitious example of such a user story might be :
“As a chief planner, when preparing courses, I want to see the most common destination industries for my students and those of my competitors, so that I can tailor my courses appropriately.”
This approach (though I can’t back it up visually with examples in this arena) sums my argument up personally. We’re never looking to tell a story, but we have goals of data we want to show and explore, and questions we want to ask. A point to the whole visualisation, story or no story.
And finally … what about “data art” type visualisations? Those kind of visualisations that may not have an obvious story behind the data, but attempt to catch the eye and focus on beauty rather than function? (I have a whole different, abandoned blog post on that – it’s such a well-discussed argument that I felt I had very little to add, and my influence is far too small for Stephen Few to notice, or care, that I think he’s wrong!).
Well, what about them? In their case, the art itself can be the point. The discussion point. The aesthetic point. Here’s a recent Hitchcock-themed visualisation I created.
The principal point of this wasn’t an analytic point or a story to tell, but to demonstrate that iconic film art can influence visualisation in an engaging way just the same way as record cover art which is a subject I’ve had fun with in the past. OK, as it happens, in this case, there is an additional point to this – to highlight gender disparity in earning power between Hollywood males and females by the relative size of circles. But you could argue (easily and successfully) that there are better ways to do the latter, though it at least formed the analytical half of my “point”.
So my conclusion is not to take a particular side on the storytelling debate, but to state that every visualisation should have a point: to answer a question, to provoke further questions, to tell a story, to demonstrate a skill, to create impact, to promote an issue, to achieve a goal, or whatever that point might be.
Do that well and you should feel that your visualisation is worthwhile. If your release of a visualisation leaves you a little “flat”, revisit and determine whether it really had a point (and whether that point was obvious). If not then it will probably come across as just data thrown on a page. Pointless.