Consider the context

• Who is your audience?
  Familiarity with the topic
• What are common practices in your field?
  Established plot types, colors, …
• Where do you present your figure?
  Different contexts require different designs

Contexts in science

Don’t hide data behind summaries

Don’t hide data behind summaries

Don’t hide data behind summaries

Bar graphs hide a lot of information about the data.

Don’t hide data behind summaries

Same bar plot - different data & statistical test results

Alternatives to bar plots

Bar plots only show mean ± SE/SD.

Alternatives to bar plots

A box plot is already better (shows more of the distribution)

Alternatives to bar plots

A box plot is already better (shows more of the distribution)

Alternatives to bar plots

Add raw data points to increase the information content of the plot

Alternatives to bar plots

Raincloud plots show raw data, summary stats, and distribution

Principle of proportional ink

Sizes of shaded areas should be proportional to the data values they represent

Principle of proportional ink

Other plot types don’t have to start at 0.

Choose the right chart type

There are so many chart types - and cool tools to explore them

Different channels - different accuracy

• Accuracy of judgement decreases from left to right
• More accurate judgements vs. more generic judgements

Different channels - different accuracy

Focus on the core message

• The readers attention is limited: be concise
• Think about the main message you want to convey
  • Which variables do you need?
  • What can you omit?

Arrange your plot

Arrange your plot so that it’s easy to extract the main message

Choose a good plot type

Different plot types tell different stories

Keep it simple

Don’t overcomplicate your figures and bury your message

What is the main message here?

Keep it simple

Message: Life expectancy and GDP differences in the world

Reading a figure is a timely experience

• We look at elements step by step before we come back to understand the figure as a whole
• Put yourself in the readers shoes
  • What will they look at first, second, etc.?
  • How many steps does it take to understand all the elements?

Consider the trip

A story about the GDP China and India

Rotate your plot

Reading labels upside down is a neck rotation - very annoying

Highlight the main message

Use highlighting and de-emphasizing

Highlight the main message

Highlight focus contries, de-emphasize all others

Highlight the main message

Highlight focus contries, de-emphasize all others

Order your data

Order categories consciously not automatically.

Order your data

Order categories consciously not automatically.

The importance of differences

The importance of differences

Use differences to communicate not to decorate

Declutter your figure

• Try to maximize the data/ink ratio
• This is to an extent a matter of taste
• Remove redundant figure elements
  Excessive grid lines, boxes, duplicate text …
• But keep elements important for reading

Declutter your graphs

Element size

• Make sure your elements are big enough
  Text size, Linewidth, Point size
• Depends on the context

Contrast

Make sure that the contrast is high enough

• Use tools to check contrast, e.g. 
  https://snook.ca/technical/colour_contrast

Color

Use logical/intuitive colors

Color

Choose colorblind friendly palettes (if in doubt: test!).

Add redundancy

Redundancy increases the chance that everyone can see the difference!

Summary

1. Consider the context
2. Make your data transparent
3. Choose the right chart type
4. Focus on the core message
5. Consider the trip
6. Less is more
7. Make it accessible

Next lecture

Topic t.b.a.

📅 15th May 🕒 3 - 4 p.m. 📍 Webex

🔔 Subscribe to the mailing list

📧 For topic suggestions and/or feedback send me an email

Thank you for your attention :)

Questions?

References