Amelia Wattenberger is a frontend developer and designer focused on data visualisation. Amelia: “I love the fact that visualization is so open-ended – we have names for specific types of charts, but there are so many types of charts that have as-of-yet never been used”!
What is the power of a good visualization? In what way can a visualization help to understand or to communicate information?
One of my favorite anecdotes is the story of John Snow and the Cholera epidemic. In 1854, the city of Westminster in London had a horrible epidemic of Cholera. People were dying left and right, and to make things worse, experts were conflicted on how people were getting sick. One major theory was that Cholera was caused by miasmata: particles that travel through the air. But John Snow, an English physician, believed in germ theory: it is caused by injesting infected food or water. So how could he prove his theory?
John Snow ended up creating a map and placing a dot over every reported Cholera case. He also marked the location of water pumps, and noted how the dots clustered around specific pumps. Most notably, he found that almost all of the deaths from a specific outbreak took place a short distance away from a water pump on Broad Street. While Snow had written multiple essays on his theory, his dot plot maps communicate his message more quickly and clearly – to even those with limited experience with diseases.
I have read that someone has to understand a visualization within seconds, others say visualizations should leave something to explore. What is your opinion on this?
I definitely disagree! Perhaps a small supporting chart at the side of an article will work better if it’s immediately understandable, but there are many different contexts for visualizations. For example, a visualization of stock activity that helps stock brokers make quick, informed decisions could definitely justify a more complex interface. At first glance, a broker needs to take more than a few seconds to learn how to read the visualization, but that initial investment more than pays off over time.
Even journalistic articles can justify more “exploratory” visualizations. I know I have spent the greater part of an hour exploring more detailed visuals, diving deep into a topic that I would otherwise have skipped over.
What is the first most important question one should ask before starting visualizing the data?
I follow a very goal-oriented approach to visualizations, most likely stemming from my background in user-experience design. The most important thing for me to nail down at the start is: what are the goals of this visualization, who are the targeted “readers”, and what do they need to get out of it? My answer to these questions serves as a North star for creating the visualization, to ensure that I’m not just creating for the fun of it.
What’s the biggest mistake often made in visualizations?
One approach that really frustrates me is when people say “what kind of chart should I make? A line chart, or a scatter plot, maybe?” I love the fact that visualization is so open-ended – we have names for specific types of charts, but there are so many types of charts that have as-of-yet never been used! I prefer to identify what kind of data I’m working with, then play with different visual representations of each metric.
For example, recently I played around with collision-detected pie charts on a US map. While it might not be everyone’s cup of tea, it’s an interesting, novel format that really highlights anomalies that would otherwise be undetectable.
What is the most exciting aspect of your training?
The training first teaches the fundamentals of data visualization development, then the fundamentals of design. The most exciting part comes next: all that learning comes together and we start working through creating our own, novel visualizations! Throughout, we’ll cover general steps that trainees take with them and can use forever – to keep them focused and confident in their approach to new projects.