Data & Narratives
Data visualisation for communication
A bit of background
Experience in data journalism, teaching, engineering.
My data visualisation approach*
Data visualisation with code
This workshop is about the craft of data visualisation
Tools do matter
Inspiration & Motivation
Reading a visualisation
Principles and examples
Data visualisation design process
Narratives and editorial thinking
Vocabulary and techniques
Exploring data visualisation tools
Visualisation as a language, informed by data journalism practice - to make real world impact.
My framework for data visualisation
Outside of New Zealand - an innovative landscape of data visualisation practice.
Computer-aided-reporting back in 1980s
A place to experiment, get real feedback
Provide value, compete with clickbait
The innovators - Mike Bostock, Rich Harris
NYT graphics dept was my starting point for inspiration*
Learning from data journalism practice
A comparison between practices and paths overseas versus NZ
Who gets to breathe clean air in New Delhi? (NYT)
Shifting smoke - Wildfires in US (Reuters)
COVID simulation (Washington Post)
2021: The year in visual stories and graphics (NYT)
This is what fuels the West’s inferno (Washington Post)
John Burn-Murdoch at FT
Data visualisation history
William Playfair, Charles Minard, Florence Nightingale, John Snow and many others
Data visualisation before we started thinking with templates
Rules are invented and meant to be broken
This could be a whole talk in itself
W.E.B Du Bois and the data portrait of black America
Invented the bar chart, line chart, pie chart.
We have [made] an honest, straightforward exhibit of a small nation of people, picturing their life and development without apology or gloss, and above all made by themselves.
~ W.E.B Du Bois
the greatest data storyteller
Whenever the data hype gets too much for me, I revisit Rosling’s talks and books.
Numbers, but not only numbers. The world cannot be understood without numbers, and it cannot be understood with numbers alone. Love numbers for what they tell you about real lives.
Curb your dramatic instincts
Reading a visualisation
(or, how to invent new charts)
A mark is a basic graphical element in an image.
Marks are geometric primitive objects classified according to the number of spatial dimensions they require.
A visual channel is a way to control the appearance of marks, independent of the dimensionality of the geometric primitive.
Every graph is made up of marks and channels
What happens when you click the bar chart button
Your data gets mapped from a spreadsheet column to a geometry
Understanding this is the key to become better at data visualisation than your BI software
Let’s break down this visualisation
At the start, you will find this slightly tedious but with time, it should become a habit.
Marks & channels?
Marks - point
Channel - vertical and horizontal
What changed here?
Channel - Sized to population
One step closer
New channel added: colour
Channel - Colour with opacity
What else is there
Gridlines, better legend, title, annotation
Look at the original again
Effectiveness of different visualisation
Reading a visualisation
Take a visualisation at work, or in the media and break it down to maps and channels.
Think about why some of the choices are good or bad.
Can you think of another way of creating the visualisation using marks and channels?
Each visualisation includes editorial decisions*
*It depends whether you are making it, or the software defaults. More on that later.
Map shows land area
Mark - line
Channel - length
States where votes are split
Map shows electoral college votes as area
Split votes are now part of the map
Mark is now Circle - size proportional to the lead
Map serves as background now
Mark is now angle -
Showing shift in margin
Also, length to show the size
Let the data speak
...and see it ramble on and on.
One data set visualised 25 ways
Sketching is every data vis creator’s superpower*
Sketch your variables to marks and channels
Pick a data set that you are reasonably familiar with, start with paper and pencil, sketch out different ways that you could visualise the data set.
Patterns, emotions and impact
It’s about patterns
And at other times
just plain confusing
Auckland Council Unitary Plan Maps
Tasks for the user
The journey to perceptual inference
A different approach
Chris McDowall’s maps for the Spinoff
Leveraging the idea of suburbs
Focus on overview and filtering
Designed to work for small devices and quick scan
Principles before process
(Decide what matters to you and your audience)
To data and to the audience,
Is always a worthwhile goal
Serve a purpose, answer a question, develop understanding, provoke more questions.
What’s your data visualisation design process?
Communicating data well is combination of art and science. It is rewarding for you and the end user, when done well.
If you do not have a process and are chucking data into a tool, or a programming language, you are not going to do it well.
The Latin word data is the plural of 'datum', "(thing) given”.
Users come to a data product with needs and requirements
Design process is an interplay between given data and user needs
What happens when there is a mismatch?
Your organisation has dashboards that exist in perpetuity without anyone knowing why they are being maintained and who exactly is benefiting from them.
Visualise with intent
‘Good design is honest. It does not make a product appear more innovative, powerful or valuable than it really is. It does not attempt to manipulate the consumer with promises that cannot be kept.’
Andy Kirk’s seven hats of visualisation design
Write down your own design process
What is the process you currently follow? How can you make it better?
Don’t build a dashboard for the sake of building one, try and solve an actual problem.
Exploratory data analysis
If we need a short suggestion of what exploratory data analysis is, I would suggest that
It is an attitude AND
A flexibility AND
Some graph paper (or transparencies, or both)*.
~ John Tukey
A note about data analysis
Doing this effectively and well is the key to your design process and discovering narratives.
The real glamour of of data jobs - cleaning up a dataset
How you do exploratory data analysis matters
Your tools will dictate the speed of discovery
Make lots of charts, make them early
This here is your data visualisation playground - use it
Narratives & editorial thinking
What do you mean when you say data storytelling?
You’ve to understand your audience, not just your data.
Then, the communication becomes about them and not you.
No singular stories
Data unlike stories is not static
Your data has to say something
Narratives open a pathway to more than one interpretation
Your analysis, visualisation and design choices dictate the narrative
Data is never ever neutral
What view(s) of your data is most relevant? In language terms, what question should your eventually chosen charts answer?
What data items and values will you include and exclude? What is most representative of your subject.
Are there any features of your data you would wish to emphasise? This is especially relevant to explanatory visualisations: if you have something to say, say it.
For sale: baby shoes, never worn
Communication vs interpretation.
“The story is triggered in our mind, when we read this passage and start to infer meaning, implication and context. A story is being presented only if it is accompanied by some explanation of the meaning of the data. Otherwise, any story derived is what the views form themselves.” ~ Andy Kirk
Martini Glass Structure
You are gearing up towards a strong ending. You want your audience to leave with a point.
Structuring visual narratives to feed the curiosity - Gurman Bhatia
Familiarise with data, setup for exploration
Narrative Visualisation: Telling Stories with Data(2010)
How to do this effectively and consistently
Easiest way to prototype a narrative is by sketching it out
Do this with somewhat realistic data
Your narrative is informed by what your data has to say
Strike a balance between your expertise and user requirements
Vocabulary & techniques
Techniques beyond a single chart
Here is an example of NZonAir survey data on media usage.
What if we wanted another categorical variable
This is something you absolutely shouldn’t present.
One of the most effective techniques when comparing across different categories
Filter to emphasise your point
Keep the whole data set in the background to show comparison
Make interaction meaningful
Once you have a visual defined, manipulate the same to show different facets of the data.
Averages hide the data
Datasaurus by Alberto Cairo, via Autodesk
Show the distribution, include the average
We come back to the Marks and Channels chart - think about how this differs
It is the most domination visual element of your visualisation.
Design a colour palette
Organisations use brand colours.
Start with defaults, explore from there.
Colour imbues meaning
Use it to make impact
Your charts can have a brand feel without exploiting the data elements
Please read the blog posts from Lisa
Experiment by creating alternate versions
Data visualisation as a brand