This article was written by Steven Hubbard, Ph.D. and Data Scientist at New American Economy

The world we live in today is made up of billions of small data points. We measure everything from air pollution to biodiversity and the average height, weight and happiness of humans. To make sense of all this, data visualisations have emerged as a powerful tool, but also one that is often overlooked in the public sector.

Data visualisation graphics, dashboards, and interactives have become important tools for government and health officials in communicating important information about the spread and impact of the Covid-19 public health crisis. Effective data visualisation tools can quickly summarize important information for many different audiences and has recently been used to deliver accurate reports on the spread of the coronavirus.

It’s important to understand your data — and the story you want to tell with the data you have.

But simply having data is not enough. Creating effective data visualisations requires many different types of skills including storytelling, data analysis, and design. This article outlines seven tips to build effective data visuals that will give your constituents clear and accurate information.

Data visualisation is a form of storytelling

Before beginning to design a data visualisation, it’s important to understand your data and the story you want to tell with the data you have. Some of the best data visualisations create clear, simple stories that can quickly engage a user’s interest. Similar to a journalist, a professional designing a data visualisation graphic must ask several basic questions. Who is your audience? How can you tell the story from their perspective? What are the most important points you want your users to walk away with? How can data be used to get your point across? Why would they be interested in the data? How adept are your users in understanding your data and analysis?

Transparency is critical

All data visualisations start with a raw data source, which needs to be cleaned and analyzed to tell the story that you want to convey. There are a lot of steps involved in this process, from first acquiring your data to providing a story that is accurate and insightful for your user. I won’t go in detail with all those steps here other than to say that while you are preparing the data, it’s essential to note the steps and methods you used to filter and interpret the data.

When working with data and visuals that represent the social world, time becomes an important variable to communicate.

These steps will be important to include in a methodology section or in the footnotes of your visualisation. For instance, you need to be transparent about how you handle missing data and the assumptions you use in filtering the information. Data visualisation is a powerful tool that can provide clear and accurate information that informs the user to make better decisions. Data visualisation can also be used as a tool to manipulate and mislead the public. Being transparent about your data source and the steps you take to prepare the visualisation is key.

Find the best visual to use

The classic bar chart is, of course, the traditional method of visualizing data. There are many newer methods that might be more effective in communicating your data. When I start designing data visualisation projects, I will often start with the Data Viz Project to provide a dictionary of different types of visuals that could be used to communicate data comparisons, correlations, distributions, and geospatial outcomes.

In addition to this resource, I like to start by drawing my ideas on paper. These drawings do not need to be neat or clear to others at this point. It’s an opportunity to visually layout the story and to think creatively of different ways of visualizing your data. I try to create at least three good different options. Also, if I have several visualisations in one project, I will use these drawings to create the storyboard of the project I’m completing.

Use your design elements wisely

As a designer of data visualisation, you have to consider many different design elements (e.g., colour, size, shape, opacity, texture, position, orientation). French Cartographer, Jacques Bertin, outlined that maps (and other graphic representations of data) are created on a plane with a series of three properties: points, lines, and polygons. You typically show values and differences in your data by changing the design variables of these three properties like changing colour, position, opacity, size, etc. For instance, the colours used in a visual may have an important impact on how your users interpret your data. The use of warm colours like red or orange often indicates loss or danger. Other colours like blue or green can indicate growth or positive outcomes.

Constructive use of data visualisation serves the purpose of communicating an idea or story that drives action.

Carefully choose appropriate colours that will reinforce the story you are telling rather than distract. I have found the online resource Color Hunt to be a useful tool in deciding colour choices for a visualisation. Also, an in-house or consulting graphic designer can be an invaluable resource in helping you make decisions about your design elements and create engaging visuals without clutter or confusing components.

Time is an important variable

Working in the public sector means that your data will involve people and their interactions with their communities and the environment. When working with data and visuals that represent the social world, time becomes an important variable to communicate. If the visual communicates data over a period of time, the visual needs to be clear about how the data has changed over time. For instance, in western cultures, time is usually represented from left to right usually along a horizontal axis. There are other ways to effectively communicate time as well, but it’s important those methods are clear to your user. If the data only represents one period of time, it’s also important to clearly communicate that point in time and provide further context about why that point in time was chosen.

Communicate error

Any statistical analysis of data involves an element of error. For instance, when calculating an average, you can also calculate the standard deviation and error of the average. Error is an important piece of information for the user to know and it leads to transparency. Statistical error is usually communicated by footnote. However, to be more transparent, error can also be communicated visually by changing opacity around a line or point, or an interactive data visualisation can provide standard error statistics through a tooltip.

There are a wealth of free tools to choose from

There are many free online tools that public officials can use to visualize data. Google Charts and Tableau Public are popular online resources that have a library of different methods to visualize your data. You can also create static data visualisation with Microsoft and Adobe products. If you are looking at developing data visualisations that provide more creative and interactive options, you will want to look into open source packages that require more experience in coding. Data science analysis tools like Python and R both provide different packages like ggplot, plotly, highcharter, Matplotlip, and Bokeh that can be used to design effective visualisations. For interactive data visualisation websites, I highly recommend D3.js and P5.js as exceptional data storytelling tools. However, both of these programs have a steep learning curve in Javascript and HTML/CSS.

Constructive use of data visualisation serves the purpose of communicating an idea or story that drives action. When I worked as a Zolberg Fellow with the International Rescue Committee in Jordan, our team needed to communicate the results of several focus groups on the healthcare concerns of Syrian refugees in a short amount of time. In addition to a full report, I used data visualisation and machine learning to both provide a visual summary of the results and allow the user to explore further to read quotes from the refugee participants on their experience with the healthcare system. The visualisation provided a means to quickly present results and explain how the data was analyzed. — Steven Hubbard

👋 You can create a post like this one! Share your thoughts with a community of public servants. Learn more.

(Picture credit: Unsplash)