Storytelling through data visualisations
There is a stupefying amount of data out there and the aim of any researcher, analyst or manager should be to make it accessible and useful. By what means can data be digested and translated into knowledge? One way is via data visualisation.
The public and decision-makers do not want data. What they want are insights and explanations, which are – rest assured – derived from data and internalised into knowledge. There are three main reasons why data visualisation is popular. It is a tool to communicate complex ideas and relationships more effectively to peers, but also to decision-makers, and the broader public. Second, it is a way to represent and explore a data set. It is also a powerful tool to explain data and tell stories.
Visualisations are more efficient than reading for instance a table with data. “A picture is worth a thousand words”, as the saying goes. Processing charts and graphics appears to be ideal for most people given the author knows what they are doing. It is visually instinctive and creates an almost immediate reaction. You can see below how much easier the trend or anomalies in numbers appear compared to the table in a simple example with a visualisation called sparklines, a tiny word-like chart (see figure 1).
Sparklines in Excel 2010 or later are graphs that fit in one cell and give you information about the data.
Separating useful from useless information
Secondly, visualisations can be used to explore a dataset. The story, insights or irregularities of a dataset are often not immediately available. Therefore, charts and graphics are created to understand and learn about what the data is telling us. It can also be used to engage the reader.
An example of using this method is the Open Data Barometer (http://opendatabarometer.org), a leading ranking of open data use across the world (see figure 2). Next to a report and an executive summary you can visually explore the dataset yourself, for example by looking at the indicator “land ownership”. The site also publishes a detailed methodology, which is crucial to establish trust, transparency and inspiration for people to use the work as a stick in policy discussions – “look how badly we’re doing in the ranking” – to improve data releases.
Open Data Barometer
Next to a report and an executive summary you can visually explore the dataset yourself, for example by looking at the indicator “land ownership”. The site also publishes a detailed methodology, which is crucial to establish trust, transparency and inspiration for people to use the work as a stick in policy discussions – “look how badly we’re doing in the ranking” – to improve data releases.
Thirdly, visualisations allow us to tell a story and, thus, explain or support an argument. This is the most difficult, because it requires skills from several disciplines. The best way to tackle it, is to create a team with a mix of skills, for example, a strong collaboration between a statistician, a developer and a communication expert. Gro Intelligence, a software start-up based in New York City and Nairobi, demonstrate this approach with data related to agriculture and food production. For example, the map shown in figure 2 in conjunction with a time series feature can be a simple yet powerful visualisation to show changes over time. In this case, this visualisation shows the impact of Tanzania’s corn production after the drought in 2016.
Figure 2: Tanzania’s corn production after the 2016 drought
Source: Gro Intelligence, slide from the GODAN Summit 2016
Similarly, dashboards like Gro Intelligence’s that made one of the US corn market are useful in obtaining an overview of the most important data in several visualisations what normally could be a complex set of indicators. A thought-through dashboard, therefore, allows a user to understand information and use it to make decisions or explain the insights to others. The key in both examples is separating useful from useless information and being astutely aware of what needs to be explained in more detail. An exemplar dashboard, graphic or narrative visualisation picks up at the right point and guides the reader.
Becoming an expert in data visualisations
A good way to start is learning from experts: look at examples in newspapers, leading blogs such as http://flowingdata.com/ and read seminal books such as the series by Edward Tufte. He will teach you about the data-ink ratio, a reminder to be minimalist and use ink mainly for data and not for lines or boxes around the chart. The rest is about practice, with a liberal mind and peer exchange of ideas. For instance, it is always a good idea to peer test a graph or a data chart to make sure that those whom it is intended for can interpret it correctly, and that shapes and colours are not distracting or culturally hard to read.
For many the solution is a spreadsheet, such as Microsoft Excel. That does not mean you are stuck with the default graphics; you can find some nice templates for popular charts on this website http://labs.juiceanalytics.com/chartchooser. For some tasks a simple spreadsheet is ideal – but it carries the disadvantages that it is error-prone and hard to reproduce.
Commercial software such as Tableau is user-friendly and often offers various ways to customise it for your uses. However, it comes at a sometimes hefty price and there is a risk that you are stuck in a “gilded cage”, a system full of features but without the freedom to leave.
Open source software, e.g. the statistical language R (https://www.r-project.org), or programming languages such as D3 (https://d3js.org), offer flexibility, powerful capabilities and an outstanding support from an enthusiastic community. However, they have a steep learning curve and can be frustrating because at the beginning simple tasks can take a long time.
If you have a blog or create content for a website, give Datawrapper a try. It is a state-of-the-art tool that produces simple graphics with a touch of interactivity and yield results that are leading practice. Often simple charts are more than enough (see figure 3). As an added bonus the software is open source, which allows for more unrestricted usage
Figure 3: Open source visualisation software from Datawrapper
In summary, data visualisations have come a long way, are used from sport statistics to pension reports, and are driven by rapid progress in available and accessible software. Edward Tufte may say “above all else show the data” and I supplement this by saying “and be clear why”. Few people are great composers, but most people enjoy music. The same goes for data visualisations. Using graphics and charts is a first step in democratising information from data. Big challenges as hunger can only be eradicated when we avoid bias, empower people and translate the insights into action.
Datawrapper is an open source tool helping everyone to create simple, correct and embeddable charts in minutes.