Part 1 of my Journal article covered the 4 layers to understand well before designing your visualizations. Those are: know your data, know your message, know your audience and know your options. In part 2, my aim was to cover the science aspect of data visualization, which entails understanding how the brain stores and process visual information.
Cognitive load is essential to understand when designing visuals. There are ways to decrease the load and, therefore, maximize data processing and understanding. Some charts are better than others when trying to convey the greatest number of ideas with the least ink.
It is because of how the brain process visuals that many data visualization designers criticize the use of pie charts in representing data. The brain is not very good at comparing the size of angles, and because there is no scale, reading accurate values is difficult. As you add more segments and colors, the problem gets worse. Labels can be hard to fit, especially on smaller segments, so legends are often required. Aside from the principles explained in my Journal article, here are few more tips in designing pie charts:
- Visualize no more than 5 categories per chart.
- Avoid using multiple pie charts for comparison.
- Do not use a dotted empty pie chart to represent null.
- Avoid using pie charts to represent metrics such as low, medium and high.
- Make sure all data does indeed add up to 100%.
- Order slices correctly: large to small, starting at 12 o’clock, going clockwise.
And regardless of the chart selected, the use of too many visual elements (such as 3-D) can cause an overload and completely distract the viewer from the point meant to be communicated. Those extra useless elements are called chart junk, a term which is further explained in my Journal article.
Read Karina Korpela’s recent Journal article:
“The Art of Data Visualization: A Gift or a Skill?, Part 2,” ISACA Journal, volume 2, 2016.