Graphs can quickly convey relationships and trends between and among data in ways that a data table or text narrative struggles to achieve.
Each type of graph has a purpose that it is best suited for.
Some have purposes they should never be used for.
Editors who know these principles can more effectively ensure that the graphs suit their intended purpose and fit scientific and mathematical standards for data communication.
This is a round-up of advice and resources explaining the best practices (and some pitfalls) for data presented in graph form.
Pitfalls of Graphs
These are the most common errors made in creating graphs. They can even put your motives into question because they manipulate perceptions so deeply. Avoid these:
- Exaggerated scale (affects perception of line slope / change over time)
- Scale doesn’t start at zero (affects perception of scale of change)
- Type doesn’t suit data (implies relationships that don’t exist)
Types of Graphs and Their Best Uses
| Type of Graph | Best Uses |
Line | Best Uses: Changes over time (or any continual measure, such as size)
Poor Use: (example) Many people’s test results |
Pie / circle | Best Uses: Compare parts of a whole
Poor Use: Individual percentages or unrelated data, totals more than 100% |
Bar / histogram | Best Uses: Compare different groups
Poor Use: Changes over time |
| X-Y / scatter
 | Best Uses: Pattern in relationship between variables, with discrete data points
Poor Use: Percentages |
Area | Best Uses: Compare changes in related groups within a category
Poor Use: Unrelated categories |
Variables
Dependent and independent variables and why the difference matters.
Resources
Editing by Design, Jan White, pp. 159–162, 168.
What each type of graph is used for.
Statistics Canada’s guidelines for graphs.
Graphics Cheat Sheet by Billion Dollar Graphics, interactive or PDF.
See chapters 16 & 17 in Geoff Hart’s Writing for Science Journals
Sources of samples: line graph author’s original, pie chart, bar graph, scatter plot, area chart
