Visualize the data

In the previous section on pre-process data, data visualization was used to explore data. This section focuses on visualizing data to interpret results based on analysis questions and to show insights to key internal stakeholders. As policymakers, you are not expected to visualize data yourself. Instead, this section provides you with guidance on how to best work with data analysts to produce compelling and insightful visuals.

How to get started

If you are new to data visualization, we invite you to take a few moments to get inspired by what other governments and organizations have done. You may want to reflect on what you like about these visualizations and what you would do differently.

  • Dashboards by Jabar Digital Service, Indonesia
  • Indices of Deprivation, UK
  • Plastic Waste Pollution Globally

How to implement

Visualize data to interpret

By visualizing data, you can gain a deeper understanding of the data. However, there is also a risk of over-investing in visualization without benefiting from significant additional insights. Before you start asking your team to visualize data, it will often be helpful to discuss with them:

  • What are our analysis questions, i.e. what would we like to find out (see section on “Define the problem”)?
  • What characteristics are we particularly interested in, for example gender, location, development over time?
  • What is the scope of the visualization and who will look at it? Do you need a couple of simple graphs for yourself vs. a more comprehensive dashboard accessible also to others?

As with many projects, you may want to consider to “start small” and work iteratively. Meaning, first create 3-5 graphs and then discuss if/what additional visualizations may be helpful.

Hiring and technical considerations

In case you do not yet have an expert on your team, who could support with the data visualization, you may wonder: what profile and skills should I look for?

Usually, the following profiles come with a certain level of data visualization skills: data scientist, data analyst, business intelligence analyst and data visualization engineers/specialists. In terms of skills, check out the section on data capabilities.

To be able to conduct the analysis, the technical expert requires access to a visualization software as well as to sufficient computing power to process the data. In many cases, common software such as Microsoft Excel already fulfils the needs of the technical expert. Other software that is commonly used and more specifically tailored to the purpose of visualizing data is Microsoft Power BI, Tableau and Google Analytics.

Tip: If you are looking for short-term support with data visualization, an alternative path to hiring an expert is to work with volunteers. One such volunteer group is Viz for Social Good. In addition, local universities may offer student projects for data visualization work.

Provide effective instructions to a data analyst

Once the data expert is on board and there is clarification on the analysis question and thus the purpose of the visualization, the data analyst will need to identify the appropriate visualization type. While this is largely the responsibility of the expert, it may be helpful to understand the different, most common options.

Bar charts utilize horizontal or vertical bars to visually compare different categories of data, with the bar lengths representing the magnitude of the values.

Find examples here.

Especially useful for:

  • Categorical data (e.g., age group, gender, education level)
  • To show distribution and compare values (e.g. income by educational level)

Tip: Bar charts are often a good choice!

Line graphs depict data points connected by straight lines to illustrate trends and variations over a specified period or continuous range.

Find examples here.

Especially useful for:

  • Time series data (e.g. weather data, heart rate monitoring, industry forecasts)
  • To understand trends and patterns (e.g., rainfall development over time)

Scatter plots display individual data points plotted on a two-dimensional graph to show the relationship between two variables.

Find examples here.

Especially useful for:

  • Different types of data
  • To show relationships between variables and represent how variables influence each other and identify data patterns

Pie charts display data as sectors of a circle, where each sector's angle is proportional to the percentage or proportion it represents in the whole dataset.

Find examples here.

Especially useful for:

  • Categorical data (e.g., age group, gender, education level)
  • To show how a quantity or percentage is distributed (e.g., number of people per age group)

Tip: Many people find bar charts easier to understand than pie charts.

Box & whisker plots visualize the spread and skewness of data by showing the median, quartiles, and potential outliers through a box and whiskers.

Find examples here.

Especially useful for:

  • Different types of data
  • To present the minimum, maximum, median, first quartile and third quartile of a dataset, see the overall shape of your dataset and identify outliers

Tip: Plots help you understand and communicate confidence intervals and, as such, how (un)certain your analysis may be.

Heatmaps use gradients of colors to represent the intensity or frequency of data values. Choropleth maps use different shades or colors within geographic regions to represent data values or densities, typically in a map format.

Find examples here.

Especially useful for:‍

  • Different types of data
  • To identify noteworthy variations in data as well as point out patterns and trends (e.g. DiCRA)

Here are two helpful resources to identify the right data visualization type:

Good practices for guiding a data expert

To further effectively guide the data expert creating the visualizations, consider the following good practices:

  • Keep it simple: Remove unnecessary elements, labels or decorations that don’t contribute to the understanding of the data. Use white space effectively to guide the audience's attention and avoid overwhelming them with information.
  • Check clarity of titles: Write titles for each visualization. The title should ideally be the one sentence summary of the main conclusion of the data being shown not a description of the data. For example: “Income levels peak in the age group of 35–55-year-olds” instead of “Income levels by age group”.
  • Double check scales: Ensure that the scales are appropriate and accurately represent the data. Choose scales that provide a meaningful representation of the data without distorting or exaggerating the information.
  • Use appropriate color palettes: Choose color palettes that are visually appealing and support effective communication. Ensure that the colors you use are distinguishable, especially for color-blind individuals. Consider using color gradients to represent quantitative values or categorical color schemes to differentiate different groups or categories.
  • Check the file name: Name the file of the visualization as accurately as possible. Avoid creating files called “chart.png”.
Tip: Consider doing a very simple user-testing by asking a colleague outside of the team if they understand the visualization.

How do I know I have successfully visualized my data?

Example: Dashboards - how to do it right?

Example of a less-ideal dashboard

Figure 1: Example of a less-ideal dashboard I Source: Young & Kitchin
  • Unclear flow of information (i.e., the reader doesn’t know where to start)
  • A lot of information of limited relevance to most readers
  • Many different colors, some without clear purpose
  • Many different visualization types and images
  • No legends

Example of a stronger dashboard

Figure 2: Example of a stronger dashboard I Source: OCHA
  • Clear layout and structure with good use of white spaces
  • Easy-to-understand visualizations and icons without adding too much information
  • Consistent and simple color scheme and font style
  • Easily readable titles, descriptions and legends for each visualization

What’s next?

Now that you have a good understanding of your data, let’s turn that data into policy.

Case Downloads

Related Use Cases

Mexico (Women in Business)
Enabling Women's Economic Empowerment: An AI driven approach to Gender Equality
Learn More
Moldova (Refugee Mobility)
Tackling Moldova’s energy vulnerability using an online compensation system
Learn More