3 Basic rules and quick tips

The choices made to visualize data impact how people view, comprehend and therefore react to the underlying data.

A series of simple rules can be followed to ensure that data is viewed in the most accurate and impactful way possible.

All data visualization requires defining both your objective and your audience. Therefore, start with the following questions (From Quick tip for visualizing data, Centre for Humdata):

• What am I trying to say? What you are trying to say should help you choose how to visualize the data.

• Who am I trying to say it to? Your audience – and how they may interact with data – will help you focus and simplify the design of your visualizations.

Before starting, make sure you understand your data to know what can and cannot (or should not) be visualized. As stated above, reviewing the metadata and data dictionary can help you understand the data included in your dataset.

Other key rules:

• Most importantly, data visualization must accurately convey the data, not misleading or distorting the audience.

• Data visualizations should answer 1-question. Doing so while allow quick and accurate review that does not leave basic patterns up to interpretation. It should never take more than a minute for the reader to understand the purpose of the graph, chart or table.

For example, in the graph below, the results clearly show a response to the question: “What is the family situation of the child beneficiaries of Program X?”

Terre des hommes & CartONG

• Ensure your charts, graphs and tables are simple and readable (think about the KISS principle: Keep it Simple, Stupid!). Cluttering graphs with too much information or multiple variables can overload the reader, thereby degrading the value of the visualization. Some potential key considerations to keep graphs and charts as visually straightforward as possible are as follows:
  • Don’t use 3D graphs – they can distract and distort proportions
  • Reduce colors, and use standard color schemes when possible (GREEN = good and RED = bad)
  • Remove special effects, like shading
  • Lighten labels, and remove any redundant labels
  • Reduce, light or remove grid-lines
  • Remove borders
  • Directly label data points or bars (if possible use) “remote” data collection mechanisms to limit the frequency and number of contacts between individuals.

Darkhorse Analytics

For more information and visual examples, please review content developed by Terre des Hommes and CartONG, on basic tips for data visualization.