Clear data visualization makes it easier for readers to understand and interpret data
Charts should be anchored at 0 on Y axis.
Be aware of misleading data in charts with two axes
You can better inform and influence your audience by using colors and layouts
Jeffrey D. Camm holds the Inmar Presidential Chair for Analytics and is Senior Associate Dean of Business Analytics Programs, Wake Forest University School of Business. This article is based upon his presentation at the Cengage session of the 2021 Decision Sciences Institute conference.
We live in a data-driven world. Understanding data visualization is crucial to understand modern economics. Statista.com predicts that the global data consumption will increase from 79 zettabytes to 181 zettabytes by 2025. The internet of things and cell phones, as well as social networks, and sensors, generate huge amounts of data. These data are then used in a variety ways to market products or services and inform and/or influence public opinion.
We expect data to inform us before we make any decisions, whether it’s about what investments we should make or which college we should attend. Like many other phenomena like how we shop or work, the COVID-19 epidemic has increased our dependence on data. There are many charts, graphs, and maps showing COVID infection rates. Publications like USA Today, The Wall Street Journal, and The Economist heavily rely on data visualization. What does this mean for the skills required to thrive in a data-driven society?
Data Visualization: Understanding Data as a Consumer of Information
It is crucial that you are able to recognize the best data visualization practices so that you can avoid making mistakes and avoiding being misled. The best advice is to pay close attention to the axes in a chart.
It is okay to start an axis at a different zero point than zero, for example. When displaying the change in a series or data over time, it is okay to start an axis at something other than zero. However, there are times when this can be misleading or used to make changes seem more prominent. Take a look at Figure 1. Which chart (a) or b) elicits a stronger reaction from you to the 2020 drop in sales? Most people would choose chart (a). Chart (b) gives a better view of what’s happening. Always make sure to check the axes before you start seeing things. Chart authors should anchor column and bar charts at zero in order to present data in the most transparent manner.
Figure 1. Comparing two Column Charts. Source: Jeffrey D. Camm
Charts with multiple axes should be treated similarly. Take Figure 2. This chart was created by the Kansas Department of Health and Environment. The graph compares daily COVID-19 incidences in counties with mask mandates (orange) and those without mask mandates (blue).
Figure 2. Figure 2. Source: Engledowl & Weiland
How do you interpret this graph at first glance? The positive rate drops when a place has to implement mask mandates, compared to when there isn’t. That’s good. It appears that COVID levels in counties with mask mandates are lower than those without. However, this is not true.
You can see that there are two axes. The orange mask on the left and the blue mask on the right. Here’s a question to ask: “Why are they using 2 axes?” This is often justified when the graphs you are plotting have very different scales (example: millions of dollars or percentages). The scale of the two graphs is not so different.
Figure 3. Figure 3. Source: Steven Strogatz on Twitter
Figure 3 shows both graphs on the same horizontal axis. Although mask mandates appear to decrease the positivity rate overall, counties without a mask rate have a lower overall positivity rate. If you want to live in a safer county, you should move to one without a mask mandate. How is that possible? Perhaps low positivity rates in rural counties are less likely to need mask mandates than counties with higher positivity.
Although the second chart shows more of what was going on, it may not convey the message as well. Before jumping to conclusions, always verify the scales and assignment of multiple axes.
Understanding Data Visualization and Influencer
You should have an intended audience and a message that informs, provokes, and influences that audience when creating a chart. Data visualization has some basic best practices.