Software packages make producing graphs quick and easy. But they can also make it quick and easy to produce a bad graph.
- are at best confusing, and at worst misleading
- can lead to the wrong actions, or lack of action
- give data nerds a bad rep.
- are intuitive and require little reader effort
- are visually appealing
- contain data that best communicate the overall ‘story’
- make the data ‘shine’
- uphold principles of scientific publishing.
The main aim of a good graph is to show the relationships between data. Graphs can be used to:
- compare the magnitude of data values or measurements
- show changes over time or across a distribution
- demonstrate how 2 or more variables are associated.
To be successful, graphs should:
- have 1 message per visual
- be clear, and typically simple
- be based on scientific principles, so that visuals
- are accurate
- are unbiased
- are accessible
- are consistent
- draw on visual design practices to best communicate the evidence.
See Which graph should I use? for more tips on creating good graphs.