As data scientists, we can use data visualizations to tell a story. But just as storytellers can be biased, our data visualizations can reinforce our own or our audience’s biases and give a biased narrative. Here’s one example:
Jorrit Schapp, who posted the above chart, highlighted a cultural bias in this chart. A non-US citizen unaccustomed to American party colors might see the red in this chart in a negative light. He saw the chart as biased against the Bush Administration. As an American, I see this chart as biased against the Obama Administration. Even though the deficit generally trends up under the Bush Administration and down under the Obama Administration, the text in the chart has a different focus. It emphasizes that under Bush the deficit never went above $1 trillion and under Obama it never went below $1 trillion. Indeed, this chart was created in response to another chart that highlighted the deficit trends under each administration. Here is the original chart taken from the blog Political Math:
These charts show how from the same data, you can tell completely different stories and confirm completely different biases. As data scientists, it’s important that we be aware of what narrative biases we may be reinforcing in how we choose to create our data visualizations. That includes our own biases as well as those of our audience.