Information design
Key question
How do you translate values into shapes?
Every visualisation translates values into visual attributes: position, length, area, colour. This translation is known as encoding. Not every encoding is interpreted with the same level of accuracy.
Learning objective: You can name visual variables, assess their accuracy, and use colour purposefully and accessibly.
The form determines how closely it is read.
5.1 / Visual Variables
Seven ways to show differences.
In 1967, Jacques Bertin described visual variables that can be used to represent differences:
01
Position
Where something is located.
02
Size
Length, area, volume.
03
Form
Circle, square, triangle.
04
Brightness
Light or dark.
05
Shade
Red, Blue, Green.
06
Orientation
Spatial orientation.
07
Texture
Pattern and grain.
Some variables are suitable for quantities, such as position and size, whilst others are primarily suited to categories, such as shape and colour hue.
Quantities need size, categories need distinctions.
5.2 / Accuracy
Lengths beat areas.
In 1984, William Cleveland and Robert McGill investigated how accurately people read different encodings. Positions on a common scale and lengths are compared most accurately, while angles, areas, and colour saturation perform significantly worse.
This is why bar charts are often more accurate than pie charts or bubbles. For important comparisons, it is worth choosing the most precise encoding.
For exact comparisons: position and length.
5.3 / Colour in Data
Colour scales follow the data.
There are three types of colour scales for data:
SEQUENTIAL
From little to much
A colour gradient from light to dark, such as for population density.
DIVERGENT
From the centre outwards
Two colours with a neutral centre, for example for variances.
CATEGORICAL
Different groups
Easily distinguishable colours with no hierarchy, such as for political parties.
A proportion of people, especially men, have red-green colour blindness. Colour scales should therefore also be distinguishable by brightness, and important differences should additionally be indicated by labelling or shape (Design Psychology, Section 4.1).
The colour scale corresponds to the type of data.
Exercise
Test an encoding
Task: Compare how accurately different encodings are read.
STEP 1
Choose five values, for example population figures for five places.
STEP 2
Display them as bars, as circles, and as colour intensity.
STEP 3
Have three people estimate the values.
STEP 4
Compare the variances and choose the most accurate representation.
Evaluation: Do your results confirm the ranking of Cleveland and McGill?