On day two of the Digital Transformation Conference, Cindy Xiong spoke about how to effectively tell a story digitally.
Xiong, an assistant professor at University of Massachusetts, Amherst, talked about designing visual data and how easy it is to mislead people with poor digital design choices.
“Poorly chosen visualizations can leave important patterns obscured and misunderstood,” she said.
For example, Xiong displayed a chart depicting the number of COVID-19 cases reported over two weeks in five counties in Georgia. The chart’s visualization displayed a misleading downward trend.
The X axis values were sorted by size, not by time, so at first glance, it looks like cases are consistently decreasing.
“The Department of Health in Georgia intended for this visualization to help you notice which day had the most number of cases and which day had the least number of cases. But it wasn’t very effective,” Xiong said. “It’s so intuitive for us to notice this downward trend.”
She suggested that a way to fix this would be to sort the data by county and have the X axis ordered by time, making the message of the chart much more obvious.
However, while it can be easy to point out the flaws in visualizations, Xiong said it’s complicated to design one.
Here’s how visualizations are typically designed:
First, researchers collect the data and data scientists analyze it, generating insights based on what the researchers collected.
Next, the visualization itself must be designed before it can reach the end user or information consumer. According to Xiong, that involves a “series of design decisions.”
“What I’ve spent the past six years researching on, is the space of how design decisions map to viewer interpretations,” she said.
Using models, data scientists and researchers can predict how viewers will interpret a visualization. The way it is designed and displayed can make or break the way viewers look at it.
Specific design choices, such as choosing a bar chart versus a pie chart, will also change how people view and compare the data. Additionally, design choices can change people’s gut reaction to data, and the decisions they make from the data.
Xiong also provided solutions and design tips to make data visualizations simple and clear.
She suggested decluttering the chart by avoiding excessive labeling and gridlines, as well as eliminating unnecessary backgrounds and text colours.
Xiong added that by highlighting or bolding the key takeaways in a visualization, the “story” will stand out even more by focusing the reader on the most important part.
While it can be easy to just throw together a chart and share it in this digital age, it’s important to make sure the design of it is well thought out to deliver the right message.
