So you have acquired some data. But what does it mean and who is it for? If you want to change someone’s mind or behavior, presenting it through a visual medium is often the fastest way to get them to connect with the information. Cue data visualization.
Data visualization involves engaging our audience in learning something new through a visual medium-- seeing something in a light they’ve never seen before. The process of creating a compelling visual com can be broken down into a few linear steps of decision making. Once we understand some important principles about how to effectively communicate data, we cannot only tell a great story but educate and inspire change in the process.
In her book, Storytelling with Data: A Data Visualization Guide for Business Professionals, Cole Nussbamer Knaflic lays out a set of 6 lessons in the art of data visualization. Here, we demonstrate the process using Excel using this Stack Overflow Developer Survey, 2017 data because who knew there were so many ways to customize visualizations with this well-known program?
1. Understand the context
Data can be overwhelming. It requires a high level of exploratory analysis before it can be disseminated in any concise way. So that’s our job-- knowing the content and breaking it down into the most important elements necessary for our audience to digest. There’s a lot more going on under the hood but no one else needs to know that.
When setting foot on the data visualization journey, it’s too easy to ask, “What is the most important information in this data?” Let’s avoid this question initially, however, as it elicits no clear answer itself and is highly dependent on answering a few preceding questions:
Who is my audience?
What is my relationship with the audience? Do they trust me as an expert or must I establish credibility?
*What do they need know? To do? *
What actions might be taken as a result of understanding this data?
What conversations need to be started?
How will I communicate the information?
What data do I have access to that will support the story and compel my audience to take action?
The more specific the answers to these questions are, the more successful the end result will be.
Once you are clear on these answers, it is time to start considering how the explanatory analysis will go. What is the Big Idea? What background information is necessary to provide and what can be left out for greater clarity?
At this point, Excel is still a tool on the shelf. Before we can generate a visual representation, we must get specific on the experience we need to create for our audience.
2) Choose an appropriate visual display
Depending on the type of content, we can make an intentional decision about which type of visual most effectively conveys our message.
12 most commonly used visuals:
- Simple text (numbers to communicate themselves)
- Table (interact with verbal system, not good for presentations or quick scanning)
- Heatmap (convey relative magnitude, color saturation to reduce cognitive load)
- Scatterplot (relationship between two things)
- Line (plot continuous data, single or multiple series)
- Slopegraph (two time periods for comparison, eg feedback over time)
- Vertical Bar (always zero baseline, leverage white space between,
- Horizontal bar (go-to for categorical data, easy to read L-R)
- Stacked vertical (compare totals across categories)
- Stacked horizontal (visualizing portions on a scale, eg Likert scale data)
- Waterfall (pull apart stacked to focus on one at a time)
- Square area (rarely appropriate, eg interview breakdown- screenings - offers)
Note: Knaflic recommends avoiding pie charts. Comparing angles, arcs and areas is highly error prone and there are more effective ways to communicate comparison using a simple bar chart or line graph.
In order to demonstrate the use of Excel with the principles outlined in the book, we will use a horizontal bar graph:
Now that we have chosen a visual, let’s begin telling the story.
3) Remove the clutter
"You know you have achieved perfection, not when you have nothing more to add but when you have nothing to take away."
When our audience looks at the visual, the goal is to require the shortest possible amount of time to figure out how to read the data so the majority of mental effort can be spent understanding what the data means. In other words, we must reduce the cognitive load. The more elements we add to the visual, the greater the cognitive load. The greater the cognitive load, the lower the chances are our audience will be drawn to engage with the data.
There are several simple ways to do this:
1. Remove borders
The human brain will naturally think of the visual elements as part of a group given the surrounding whitespace.
2. Remove gridlines
Gridlines threaten to compete with the data itself. If you feel they are necessary to effectively process the information, fade them to a light gray to avoid distraction.
3. Remove data markers or labels
The data here is already shown visually with bars that the eye naturally follows. While this doesn’t render markers completely useless, it does diminish their need.
4. Clean up axis labels
Ask yourself a few questions when cleaning up the axis labels. Do we need trailing zeros? Is it possible to abbreviate?
5. Label data directly
What does each bar represent? Is a legend absolutely necessary? Ultimately, a legend adds up to a greater cognitive load. With graphs such as this, we can simply label the bars individually that follow the natural flow of the path.
6. Use color to your advantage
Which parts are related? Use color to your advantage by using the same color for related bars and labels. The human brain will naturally associate the two.
Now that we’ve cleaned up a bit, draw your attention to the remaining white space. Now that the unnecessary elements have been removed, get comfortable with the newfound visual freedom. Beautiful, isn’t it?
Our goal is to provide the viewer with visual order so the data does the talking without complexity stepping on its toes.