Understanding data visualization biases

And it doesn’t matter what language you write your code in. At the end of the day, code is a way to communicate business solutions.

This being said, it’s important to have a business mindset when you’re developing code, as you’re solving a problem for the company in some way or another.

Part of the “business solving” process is communication of data. One of the most common methods of being able to communicate data to key in on a solution is by leveraging visualizations such as graphs and charts.

Just like any other form of communication, biases will be introduced when developing and displaying visualizations, both visual and cognitive.

Understanding these biases is crucial for creating effective and trustworthy data visualizations.

By accounting for these biases in the design process, we can ensure that our visualizations convey the intended message accurately and facilitate better decision-making based on the data.

Understanding visual biases

Visual biases come from how our brains naturally perceive and process visual information. Certain visual cues like color, size, and position can unintentionally draw our attention in specific ways.

For example, the color red is generally perceived as an alert or warning signal, triggering our brain's ability to quickly focus on it. A darker, richer shade of red can serve this "alert" purpose well without being as jarring or distracting as a lighter shade of red.

Understanding cognitive biases

Cognitive biases are mental shortcuts our brains take based on prior experiences and assumptions. These can lead us to misinterpret or see patterns in data that aren't really there.

A great example is the classic “correlation equals causation”. If (for some reason) you’re showing an increased trend in ice cream sales and an increased trend in sunburn, you can subconsciously give a message that “the more ice cream sales I make, the more people get sunburn”, which is clearly not the case.

Ice cream sales don’t cause sunburn, but it’s correlated because maybe they’re standing out in line out in the sun?

Overcoming these biases

Of course, I can dive extremely deep into specifics of each of these biases (such as selection bias, cherry-picking, scale manipulation, etc). However, that’s out of the scope for this newsletter.

For simplicity, follow these 3 tips to be able to overcome visual and cognitive biases:

• Maintain simplicity and clarity. Avoid overly complex or cluttered visualizations that can overwhelm the viewer's cognitive load and introduce biases. Strive for a clean, straightforward design that prioritizes the most essential information.

• Provide appropriate context and guidance. Include clear labels, legends, and explanations that help viewers accurately interpret the visualizations. Contextual information can mitigate biases by grounding the viewer's understanding and framing the data appropriately.

• Foster objective and transparent design. Be mindful of potential biases during the visualization design process. Adopt an objective and transparent approach, avoiding distortions, exaggerations, or visual elements that could unintentionally mislead or influence the viewer's perception.