Communication of statistics through visualisations
Statistics explained visually
December 6
2017
Maarten
Lambrechts
Eurostat
Luxembourg
Maarten Lambrechts?
Today
What do users want?
Why data should be visualised and how
Lunch
Pitfalls in datavisualisation
Making make-overs
9:30-10:45
11:00-12:30
12:30-14:00
14:00-15:30
15:30-17:00
Disclaimer
Who in the room has made charts for Statistics Explained?
My apologies
What do users want?
I
Answers to their questions
Lisa
16 year old schoolgirl
Addicted to her mobile phone
Assignment on SDG poverty in EU
'Pfff.... Poverty? What do I know about that?'
'What? A scatterplot? I thought someone spilled her latte over that page 😂'
Jonathan
37 years old journalist
Reads 3 economical newspapers each day
Needs story for a special on European SDG targets
'Ok, chef, article will be ready in 2 hours'
'Yes, I'll find the data for the chart to illustrate the article'
What are sustainable development goals?
What is poverty?
Is poverty a big problem in Europe?
Where in Europe is poverty the worst?
What is the goal and how will we reach it?
Is the EU going to reach the poverty SDG?
What aspect of poverty is doing the worst?
Which countries are the worst performers?
What groups of people are most at risk?
Where is the data for my chart?
Margareth
42 years old business analyst
Consulting for small businesses umbrella organisation
'Where is the data? I need the data!'
'A visualisation is worth a 1000 words'
Alan
63 year old professor
Research: 'UK SMEs trade after Brexit'
'Wait, let me get my reading glasses'
'Right mouse click... How do I do that again?'
Where do small enterprises trade the most?
How much do small enterprises import on average?
And in total?
How do enterprises behave on import vs export? Intra-EU vs extra-EU?
Â
Is Brittish trade very different from other countries trade?
Will SME's suffer more from Brexit than big companies?
Â
Â
Takeaways
Chart Design
Â
Add a clear and descriptive chart title
Make text run horizontally
Respect hierarchies in the data visually
Be consistent
Avoid log scales, or be very clear about it
Don't cut scales for bars and time scales
Use appropriate chart type
Always show units
Add some absolute numbers to indicate scale
Targets are often the story
Takeaways
CHART INTEGRATION
Â
Element of the page a lot of people look at first
Charts should be self sufficient: have a title, units, keys
Don't bury the lead, support it with a visualisation
Ensure good interplay between text and charts
When you use visual language ('clusters', 'outliers', 'big gap', ...), think about making a visualisation
Offer data link in all views
Why data should be visualised
and how
II
Why visualisation?
- Interpreting numbers becomes easier
- Trends become clearer
- Let users contextualise themselves
- Message is supported and amplified
- Augments reliability
- Makes long text breath
- Replaces lengthy paragraphs full of numbers
- Easy sharing of message
- Can be aesthetically pleasing
- Draws attention
Chart integration
Take a Stats Explained article
With the different users and their possible questions in mind, answer the following questions:
- If you would have to pick 1 chart to go with the text, which one would it be? Why?
- What charts are missing from the article in your view? Are there claims or explanations in the text that lend itself well to be visualised?
- Look for 'Figure 1', 'Figure 2', ... in the text. What does the text say about what you see on the chart? Is that clearly visible in the chart?
- Do the charts tell a story? Or are they just meant for looking up values or countries?
When not to use visualisation
Â
When you want to offer quick lookup of values. Use a table for that
Â
When you don't have a clear goal for the chart. Ask yourself: 'For whom and why am I making this visualisation? What questions does the chart answer?'
Â
When you have too little data
Chart design
How to choose a chart type?
How to design a chart?
Introducing:
The Datavisualization Checklist
Pitfalls
&
more design tips
III
Seen so far:
Horizontal text
Descriptive titles
Units
Meaningful colors
Sorting
Don't cut scales
What else?
Choose an appropriate aspect ratio
One chart, one message
No double axes
Don't cut bars
Guide your reader: where does he/she need to look?
Label directly
Axes, axes labels and grids should be supportive, in the background
Be careful with stacked charts
Data density shouldn't be too low nor too high
Charts should tell something
Respect and show the data structure and hierarchy
Experiment, but guide the reader
Correlations are powerfull
but dangerous
Tell the story, with the title, annotations and colors
Think about fonts
Sketch & draft
Â
Get feedback
Â
Adapt
Makeover
IV
Take a Statistics Explained article
Think about how visualization could improve the article.
Consider:
- The datavisualisation checklist
- The pitfalls in data visualisation
- Appropriate chart types
- Chart integration, text-visualisation interplay
- Consistency (colors, forms, ...)
- Telling a story with visualisations
- Different user profiles and the questions they might have
Thank you
Communication of statistics through visualisations
By maartenzam
Communication of statistics through visualisations
- 3,544