Larissa Schuermyer

In a data-driven world, expressing visualization is crucial to driving the narrative and understanding of an analysis. Yet, many times the visual representations of data seem arbitrary – picking a chart or graph sometimes feels like a game of roulette. If we use data to support our business decisions, we should also use it to support visualization decisions. Fortunately, several academic papers have been published on the human brain’s ability to process different types of visual elements.

One paper, entitled Graphical Perception: Theory, Experimentation, and Application to the Development of Graphical Methods, written by William Cleveland and Robert McGill, outlines a few basic perceptual tasks involved in reading charts and how well we can perform these tasks. Cleveland and McGill discovered that “position judgments were 1.96 times as accurate as angle judgments.” One perceptual task required in reading a pie chart is judging angle. A bar chart, which requires judging position relative to a common scale, is a better alternative. Let’s examine this in the context of our response when we forget someone’s name:

While the main takeaway in the charts above may be the same (we wait for some else to say their name), judging relative proportions is easier with a bar chart. As such, the time it takes to read and interpret the data and make more detailed comparisons is faster, allowing greater understanding among audiences. Some analysts may opt for a donut chart, what about you?

One quick option to show visual comparisons is to add visualization to an Excel report via conditional formatting and data bars. This aligns with our ability to most accurately judge position relative to a common scale. Of course, a chart works here as well:



Visualizations should be kept as simple and effective as possible without overwhelming the audience. Just because there is a fancy color or chart doesn’t mean you need to use it – does it really provide additional value for the reader?

Aside from the visual element of a chart, adding data labels further increases the ability of a viewer to takeaway insight or make comparisons. Let’s add those to the previous bar chart:

Now we get even more readability. Finally, don’t forget to pick appropriate colors (usually brand colors, which should be available in a brand guide).

These guidelines are only the tip of the iceberg. Other gold nuggets can be found in Cleveland and McGill’s paper, and other papers that have been published more recently.

Three Ideas for You:

  1. The human brain is better at accurately judging position along a common scale than angle, which is used in reading pie charts. Keep this in mind when creating reports and dashboards.
  2. Don’t pick visualizations arbitrarily. Do some research into what works in which situations by searching for papers, such as the one cited here. A few other research papers can be found here and here.
  3. There is still a lot to be discovered about how to best visualize data. Don’t be afraid to play around with visualizations, compare them side by side, and let your intuition guide you to the most effective one.

Larissa Schuermyer