To maximize the benefit of visualization, you need to focus on the task you are aiming to achieve. The visualization approaches for monitoring are different to those for exploring, and different again from explaining – as I outlined in my post about MEME (Monitor, Explore, Manage, and Explain). In this post I will look at using visualization for Explaining, showing how you can use visualization approaches to help ensure your audiences understand your message.
It’s for your audience, not for you!
Perhaps the most important point about creating materials that are intended to explain and/or persuade somebody is that they need to be designed with the audience in mind. When creating a visualization for exploring the data, your preferences matter most. But when using visualization to explain something to an audience, your preferences are a secondary matter. This means that you should find out as much as possible about what works for your audience, and if you can’t find out detailed information, use estimates of what works for typical audiences.
Know what the message is
To design a visualization that explains your message, you need to be clear about what that message is. Explain is the endpoint of a process, a process where you have understood a question, explored the data, and found the answer. At the Explain stage, you are not trying to share everything you know, you are seeking to answer a question, share a message, and cause a specific outcome to happen. The more somebody looks at your visualization, the more the reader should find they are drawn inexorably towards your message.
A great example of visualization is provided by the FT (Financial Times) Covid website.
The visualization highlights three waves of COVID deaths and the fact that the current wave is declining. Other key strengths include: 1) Using labels in the chart, rather than a legend, 2) relatively few numbers (the message is the pattern, not the exact numbers), 3) using colour groups to represent geographies, and 4) adding markers to identify the three waves.
It is about truth, not precision
Too many data people focus on accuracy, at the expense of getting the truth across. The popular meme below highlights the trade-off between precision and message.
The ‘truth’ is that the person under the weight needs to move – however, too many data people are more like the person on the left, focusing on the precision.
The visualization needs to be readable
This point comes back to being audience focused. The fonts need to be large enough, the colour palette needs to work for people who are colour blind, the alt-tags need to work for people who use a screen reader, labels need to be attached to the right places, and the choice of visualisation needs to convey the message that is in the data.
Part of making your visualization readable is ensuring that your graphics do not deceive the eye. For example, types of visualisation that do NOT convey the message are some pie charts, especially three-dimensional pie charts, as in the example below.
In this example, sector D looks larger than sector C (even though they are the same value), and sector D looks larger than sector E (even though E represents a value 5 percentage points larger than D). By contrast, a simple stacked bar chart (like the one below), represents the sections fairly.
Arrange the data to make the message clearer
Too often we see data reported in some arbitrary sequence, for example in alphabetical order, or in questionnaire order. The message in the data can be made much easier for the audience to see if the visualization is arranged to make the message clearer. In the chart below, data about deaths per million and cases per million is arranged in the order it was downloaded ie. alphabetically.
In the second chart, the data have been sorted by descending number of deaths per million. In this version, the message in the data becomes clearer. The message, BTW, is that during this period of time, some countries (eg. UK, Italy, Sweden & France) were doing much less testing than others, resulting in a very different ratio between the number of cases and the number of deaths.
Simplify the details to highlight the message
The message is not the detail; indeed, the detail can often obscure the message. A good visualization usually reduces the amount of information that is being communicated to strengthen the message.
Nate Silver’s politics and sports commentary site FiveThirtyFive.com is well known for its use of visualization. In the example below, the visualization from FiveThirtyFive in 2016 looks at two key drivers of support for Hillary Clinton from white voters, in the US Presidential election.
The visualization shows that 77% of white voters who never attend religious services and who have a degree voted for Hillary Clinton. By contrast, only 23% of white voters who attend religious services weekly and have no college education voted for Hilary Clinton.
Elements of simplification that help make this visualization effective at explaining the message are: 1) restricting the story to two factors (religion and college), 2) using a symmetrical split of three levels for religion and college, 3) using colour coding and shading to help those familiar with party coding in the USA (Republican red and Democrat blue), 4) avoiding decimal places in the numbers, to aid clarity.
Learning from history
The visualization below does not look so impressive in 2021, but we need to recognize that it was created in the 1850s by Florence Nightingale.
A Royal Commission into deaths in the Crimean War had discovered that most people died because of poor medical care and from diseases, not from deaths on the battlefield. This visualization was effective at improving conditions for the soldiers. Although we would draw it differently today, we can see the two key points of a great visualization for explaining here:
- A single, clear message (deal with non-battlefield deaths)
- Simplified – just three categories of death, no numbers, one block per month
- Readable – the chart is easy to grasp and does not require special training or explanation.
In summary
When creating a visualization (or a set of visualizations) to explain your message ensure that you know what the message is, remove anything that is not communicating the message, and make it readable.
You might also be interested in
Here are some other great resources on the topic of visualization:
- MEME - votre clé pour créer de grandes visualisations
- Utiliser la visualisation pour le suivi - une partie de l'approche MEME
- Using Visualisation for Exploring Data – Part of the MEME Approach
- Utiliser la visualisation pour gérer - une partie de l'approche MEME
- Webinar 27 July - Making your data dance | The 4 Pillars of Data Visualisation | APAC
- Webinar 27 July - Making your data dance | The 4 Pillars of Data Visualisation | Europe
