Design Best Practices for Data Visualizations

Apr 01, 2018

Data visualization is now common in most business settings, but that doesn’t mean that organizations are necessarily using it effectively. In this video, you will learn about the design best practices for data visualization.

While data visualization is now common in most business settings, that doesn't mean that organizations are necessarily using it effectively. For example, most visualizations are created on and for desktop screens without consideration of their effectiveness on different sized screens. While these designs are often useful on smaller tablet or smartphone screens, they don't take into account that much, if not most sharing of these visualizations is done using large screens in conference or common areas. The needs of these audiences for sharing, discussing, and interacting with the stories using the data, are too often not taken into consideration. And that's just one example.

Properly used, data visualization applies the tenants of Gestalt theory, which refers to how the brain interprets what to visualize to effectively communicate statistical information. So, it's important to understand how our brains see graphics. Consider icons and symbols many people encounter daily like the trash can icon on a computer screen. While it may seem abstract, this represents the very real task of deleting something.

When dealing with data, it's important to take abstract concepts represented by the data and turn these into visualizations that are more tangible and intelligible. Gestalt theory centers on the Law of Pragnanz, German for pithiness which means being substantial yet concise. Symbols like the screen trash can must make immediate sense without requiring further contemplation or explanation. They should be specific enough to convey meaning not allowing for confusion.

According to the Law of Pragnanz, the human brain favors simplicity. We process basic visual patterns much faster than complex ones. To identify the year's top performer by means of a bar chart, for example, if arranged in alphabetical order, the relevant data would not be immediately apparent. However, if arranged from the highest to the lowest score, the data is simple to interpret. When it comes to data visualization, the data needs to be arranged in a logical fashion.

Further, your audience will drive what is logical and simple. Subject matter experts in the field your data addresses will be able to navigate complex visualizations that will baffle general audiences. Who you are telling your data story to must be considered when thinking through questions of simplicity and complexity.

Then there's the Gestalt Law of Continuity, which posits that the human brain naturally groups items that are in line with each other. Take a top performer bar chart, in an alphabetically arranged chart, the bars create a jagged line with sudden ups and downs. Whereas in a chart organized by score, there's a continuous line, making it more immediately intelligible.

The Gestalt Law of Similarity suggests we automatically group objects of similar shape, color, size, etc. Assigning a different color to each bar in a bar chart for instance makes the data simpler to follow. Both of these suggest the importance, as well, of color choice. Don't automatically rely on the tendency to use color as visual keys. That is, red for negative, green for positive, yellow for caution, and so on. Avoid using especially bright colors for data points, as their relative intensity may give a false impression of significance or distinction that's unintended. Likewise, use color consistently across different visualizations related to the same overall topic, or showing the same data in different contexts.

Of course, we also interpret objects placed close to each other as a group according to the Gestalt Law of Proximity. Consider now a bar chart displaying sales for four regions across four quarters. When grouped by quarter, figuring out the sales per region across all four quarters proves difficult, but when grouped by region, it's much easier. When using data visualization, grouping should support a chart's objective.

And finally, always use a legend or key in a chart containing multiple variables. The legend acts as an explanation of the variables used in the chart as they appear. It provides a quick guide on how to identify the data from each variable, but consider the position, clarity, and detail of the legend. When placed or labeled incorrectly, a legend can cause confusion which results in misinterpreted data. And the purpose of all of the choices made in creating data visualizations is to clarify, making the story the data tells as accurate and succinct as possible.