How I came to know so little about Data Visualisation
Best Practice in Data Visualisation has never been better documented, and yet my own thirst for knowledge on the subject remains unquenched and I can’t find the answers I need. How did I come to know so little?
I’ve been working in Data and Data Visualisation for my whole career, but my formative years in data visualisation started approximately 6 and a half years ago as I started to use Tableau and started to learn more about Data Visualisation best practice as a result.
I learned about Gestalt Principles, Pre-attentive attributes, and cognitive processing and how they are applied to data visualisation. I learned about Tableau’s colour palettes and how they are researched from amazing researchers like Maureen Stone (and in turn learned about red and green palettes and colour blindness). I used these lessons to build rules and best practice around my visualisations, for example using the line charts to show change over time and avoided pie charts to show parts to the whole (because we can’t accurately judge angles).
Stephen Few laid down the law and in places I learned the subtleties behind the rules and learned when to break them thanks to research such as the science behind Pie Charts from Robert Korsera.
At the same time, I learned about design and the importance of engaging with an audience with visualisations. David McCandless provided endless sources of inspiration in his books, and Alberto Cairo taught me how to respect both accuracy and beauty.
I’ve used my knowledge to build a public portfolio of work that ventures from best practice all the way through to unusual charts, with mixed success.
I’ve also worked as a consultant building business visualisations for clients and advocating the sharing of more of these “real world” visualisations on sites such as everydaydashboards.com.
The Important of Rules
My passion for the rules I’ve learned through the course of this journey has been feverous. I’ve given talks and lectures on best practices in Data Visualisation; I’ve talked at conferences about the benefits of keeping charts simple and why it aids our learning, the community, and our visualisations. I’ve blogged at length on the subject. I’ve sparked numerous Twitter conversations and more than a few arguments on the subject of best practice in Tableau Public (probably much to the annoyance of many).
You see I’m a very rules-based person, the rules and reasons behind best-practice data visualisation appeal to me. I’m relatively intelligent and can understand the science behind them and that comforts me.
The purpose of Data Visualisation, in the words of Steve Wexler, is:
for the largest number of people, provide the greatest degree of understanding with the least amount of effort
I can subscribe to this and the rules I’ve formed around data visualisation help me find ways to build visualisations that meet this aim.
But I’ve also broken the rules on many occasions, knowing that many of the charts and visualisations I’ve built aren’t best practice. I’ve enjoyed the process of creating visual metaphors with strange chart types and exploring new ways of telling stories that can engage with audiences in different ways. Breaking the rules never felt alien or wrong, because the rules gave me space where I knew what I was doing and I enjoyed the creative process.
The feedback around these more novel approaches has always been kind, and have often resulted in the visualisations getting more attention than my more stayed approaches, but I’ve always been careful to be conscious that any feedback is artificial given my audience is weighted towards data visualisation practitioners and peers, rather than a representative sample of the public.
Ultimately though the rules have helped me form opinions, rightly or wrongly, around what “good” means in terms of data visualisation. Can I understand it quickly, would a lay-person? Is a complex story being made simpler and easier to understand through the visualisation?
Rightly or wrongly I’ve often used these opinions to discuss with others the merits of their visualisations, enjoying the debates and discussions and always, hopefully, approaching them with the best intentions and always with a willingness to learn (rather than assuming my opinions are correct).
A Journey into the Unknown
It’s against the backdrop of this journey over the last 6 or 7 years that I find myself now in a position where I doubt my opinions and learnings more than ever.
I continue to learn new chart types, for example, UpSet plots and Voronoi Treemaps, and continue to debate about the effectiveness of different graphs. However, my assumptions and attitudes to what is “engaging” and what an audience can understand continue to be eroded and my need for better “rules” only grows.
I crave research or discussions into whether our assumptions and “accepted truths” in data visualisation are correct in order to help me navigate my own data visualisations and the opinions I hold so dear.
Will a non-expert audience be put off by a novel chart type? How much patience will they have to understand it? Will a novel chart type in itself encourage them to spend longer understand a visualisation?
How do audiences engage with visualisations? Is it the subject that entices them to look further or the visualisation?
Can a “dull” subject be made inspiring and interesting for a lay audience using a compelling visualisation? Will they be encouraged to “lean in” and spend time? or will the subject immediately put them off?
Do charts and graphs always aid comprehension, or are there times simple textual stories without visualisations are better?
Do best practices lend themselves to less engagement? Do visually appealing charts automatically lead to more engagement and understanding?
Even fundamentals such as whether bar charts axis starting at zero are bad practice are still debated and seem less clear than ever before.
It’s clear the answer to a lot of these aren’t binary answers. The devil in answering these are in the detail of the audience, the charts and graphs we use, and the information being conveyed. However, I still feel there is a significant lack of scientific study and research that helps answer even the most fundamental questions around how audiences engage with data visualisations that can be applied to our day to day work.
Whether that gap between the science of data visualisation and the art of design and engagement can ever result in satisfying research and clearer rules, I’m not sure. Similarly. someone may be able to point me to existing research that answers some of these questions.
In the meantime, I can only feel the questions I have and the uncertainty around my field growing stronger, and my “expertise” ebbs away with every viz I see.
Where does this leave me?
My own journey into this field is clearly just beginning, and I have none of the answers I need to make any clear judgments or opinions, in fact I have much less than I thought I did a year ago. However, in the absence of clear rules and research I think my best bet is still to stick to these propositions that have tended to serve me well:
subjects that are interesting will naturally lead to more engagement
keeping things simple is the best way to ensure you educate people
don’t rely on gimmicks and novelty alone to engage an audience, you need to have more
best-practice isn’t a bad place to end up
From here I’ll keep learning, and keep discussing how we help beginners and experts (or enthusiastic amateurs like me) navigate the “rules” of Daa Visualisation. Thanks for your time.
Thanks to Josh (@data_jackalope) for the feedback