Tortas and Data

Data is a powerful tool. Can I manipulate it? Does it manipulate me?

Data visualization is part of our daily lives. But how we read these charts, graphs, infographics, heat maps, etc. depends on a number of things. Studies have shown that cognitive bias and data visualization literacy play a big role in how we interpret them.

Data visualization literacy is how well a person can read a chart or graph. Cognitive bias is a subconscious flaw in a person’s judgment and decision-making.

Today, I will eat a delicious torta ahogada for lunch. It’s a sandwich drowned in salsa and a local specialty of Guadalajara, Mexico.

On the corner near my house are 3 torta ahogada vendors. But which one is best?

Today, Vendor 1 sold 79 tortas, Vendor 2 sold 77, and Vendor 3 sold 76. In a bar chart, it could look like this:

They seem pretty neck-and-neck. Let’s take the same data and make some slight changes to the Y-axis and… voilà!

Changing the lower limit and unit has distorted the previous chart. Now, Torta 1 looks significantly more popular, no?

One study found that attraction bias causes a viewer to perceive 2 choices more positively when a third choice seems irrelevant. In our chart above, Torta 3 looks pretty irrelevant - nearly bottoming out the chart. By contrast, this makes Torta 1 look even better.

Let's turn our attention to the good ol' pie chart. Is this chart below helpful at all?

Let’s make it 3D and rotate it.

Do the pieces look different sizes in these two charts? Maybe Torta 3 is the best. The 3D adds no additional information and, in fact, distorts the piece sizes.

You’d be surprised at how often these simple examples happen. Data visualization is the first step in using data to tell a story. For public policy, storytelling is critical to advocacy. For nonprofits, it can justify their existence.

As for my lunch, I chose the torta vendor with no line. It was delicious.

How does data visualization influence you?

Follow GraphCrimes on Twitter for hilarious, real examples of data visualization gone wrong.