Florian Ramseger is a product specialist at software company Tableau. This year he came to the World Resources Forum, an international conference that takes place annually in Davos, Switzerland, and brings together people with a mission from research institutes, think tanks and universities, as well as policymakers and businessmen.
Standing in front of the audience at his master class titled “Telling Your Story Through Data,” he shared an anecdotal story from his past as an economist. Once he tried to build a linear regression model using data, but none of his attempts to fit the data into a good model were successful. When nothing worked, he went to the people from the statistics department and showed them his data. In response, he received a question: “Why do you fit your data into a linear model if there is clearly nonlinear correlation?”
Looking at the data is crucial for researchers, Ramseger said. However it is not the only problem that we have with data now. Participants in the master class agreed that they often experience difficulties in communicating their research outcomes to the public. They find it especially hard to communicate results to the nonscientific community.
Fortunately, data journalism provides useful tools that researchers can use to examine their data and communicate the results in a digestible manner. Here are the main tips researchers should follow to succeed:
1. Your data should tell a story.
Every story has three major parts: context, struggle and resolution.
Context should make readers familiar with the concepts and objects that you are researching. Here, researchers should try to link new information to existing knowledge. One way to do this is to use infographics to set up the context, as in a Süddeutsche Zeitung article (see below) that visualizes the difference between the number of delegates at a G-7 meeting and the number of policemen involved in the event as the reader scrolls through. The article first sets up seven figures representing the G-7 delegates and then compares them to the number of policemen.
Visual context usually works for simple and well-known objects. In a more complicated context, textual introduction will work well.
2. Make the struggle in your story easy to understand.
In the struggle part, researchers should illustrate the differences and contrasts in their data. Data visualization serves as a great tool to clarify and highlight these differences. Following are some useful tips.
First, try to use horizontal bar charts to illustrate differences in data. They are much easier to understand than, say, pie charts, as the latter require readers to make more complicated visual comparisons, such as those between the angles of the pie’s various sections. The second key element here is to always choose horizontal over vertical headers, as readers should not need to rotate their head at a 90-degree angle to read it. And finally, do not forget to use different colors for readers to understand the differences between each bar.
Second, do not be afraid to show differences in differences.
A good way to do this is to present two lines of data showing different trends, and leave a space between the lines to make it easier for users to focus on the gap between them, as in the example below. Using different colors together, leaving space between the lines, simplifies the process of seeing and grasping the differences in the data. But do not go too far into it, and do not overload the graphic with too much information, even if it looks pretty.
Third, do not be afraid to show difference in differences in differences.
But do not use 3-D for that. 3-D maps are very complicated to follow and understand. One solution may be to provide small illustrative graphs for each time period next to each other. This helps readers see trends over days, weeks and years. Using colors is still very helpful to illustrate the differences from Point 1 within each of the small pictures.
3. Ask yourself: What do you want the reader to take away?
Once this question is answered, make it the resolution of the story. Do not leave the reader without some culmination, as otherwise your information will look just like raw data and will not leave a message.
Just telling a fact or showing the picture is not enough to create good data visualization. “For me, good visualizations are not necessarily the ones that are the most pretty and not those that use correct tools, though it helps,” Ramseger said. “The really good ones are those that touch you and tell you the story.”
The example below, illustrating gun-related murders in the U.S., is not a scientific piece, but it clearly sets up the context, shows the differences and leaves a final message that readers should take with them.
4. Something to bear in mind about data journalism
Keep in mind that in some cases, it may be dangerous to rely too much on infographics. “If we are going to start reducing the word counting in the article and only use infographics, we are going to lose the big picture,” says Jocelyn Blériot of the Ellen MacArthur Foundation. To avoid this, each infographic should be combined with a thoroughly researched story, he adds.
Another problem with data journalism is that sometimes people still have a hard time understanding the visual part. Claude Fussler of the CO2 Forum said that in his village in France, many people do not look at data visualizations because their education level does not allow them to understand infographics easily. This is another reason why researchers should be careful when selecting data visualization tools and should prefer those that simplify the understanding of data, such as the ones mentioned here.
After all, “if you do it right, you can use it for any topic—except, maybe, love,” Ramseger said, laughing.