CHARTPOLICEChart Police

Crime taxonomy · Maps and colour

Misleading colour scale

Colour scales built to exaggerate, invert or invent differences.

What it looks like

Class breaks chosen to blow up tiny differences, a diverging palette centred on an arbitrary value, a saturated or truncated colour range, a reversed scale, a rainbow for ordered values, or legend colours that do not match the map.

Example

Example (Misleading colour scale): Every state sits between 4.8% and 5.2%, but class breaks 0.02 points wide paint a crisis.RIGGED COLOURS
The crime. Every state sits between 4.8% and 5.2%, but class breaks 0.02 points wide paint a crisis.
The same data shown honestly: A sequential scale from 0% to 10% shows what the data says: nearly uniform.FIXED
The fix. A sequential scale from 0% to 10% shows what the data says: nearly uniform.

Why it is bad

How to fix it

  1. Use a sequential palette (light to dark) for ordered values, and diverging only around a meaningful midpoint.
  2. Choose class breaks from the data's real spread (quantiles, round numbers) and state them in the legend.
  3. Avoid rainbow palettes for ordered data; they are also hard to read for colour-blind readers.
When it is not a crime. Fine-grained breaks are honest when small differences genuinely matter and the legend makes the range obvious.