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?

 

Is Brittish trade very different from other countries trade?

Will SME's suffer more from Brexit than big companies?

 

 

Takeaways

Chart Design

 

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

 

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?

  1. Interpreting numbers becomes easier
  2. Trends become clearer
  3. Let users contextualise themselves
  4. Message is supported and amplified
  5. Augments reliability
  6. Makes long text breath
  7. Replaces lengthy paragraphs full of numbers
  8. Easy sharing of message
  9. Can be aesthetically pleasing
  10. 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

 

When you want to offer quick lookup of values. Use a table for that

 

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?'

 

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

 

Get feedback

 

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

  • 1,003
Loading comments...

More from maartenzam