00:00:00 Introduction

Hello everyone and welcome to today's webinar, "Best practices for data visualization in monitoring and evaluation." My name is Fay Koutsioumaris. I'm working with the ActivityInfo team, where among others, I work on producing various guides and documentation for the platform. Today, together with Mr. Maarten-Jan Kallen, the Managing Director of BeDataDriven, and Mr. Alex Bertram, the Technical Director of BeDataDriven and founder of ActivityInfo, we'll be hosting this webinar, broadcasting from The Hague in the Netherlands. We're excited to see such a big interest from all over the world; more than 1,500 people registered for this webinar. So, thank you all very much for joining us.

Before we start, I would like to share some housekeeping rules for everyone. Your microphone is muted, and you should all be able to see the shared screen. The webinar is being recorded, and you will receive the recording after the webinar. It will also be available on our website in a few days' time, where you can find recordings of previous webinars on M&E and on using the software ActivityInfo. During the webinar, you can send your questions via the Q&A section found in the toolbar at the bottom of the screen. This makes it easier for us to keep track of all the questions.

I would like to introduce today's speaker, Maarten-Jan Kallen. Mr. Maarten-Jan Kallen is the Managing Director of BeDataDriven, the company which owns and operates ActivityInfo. He has been working for the company since 2013 and has been closely involved in the evolution of the software. He has a PhD in applied mathematics from the Delft University of Technology in the Netherlands, and he has advised companies and governments in the areas of risk analysis and decision theory for more than 15 years. Also together with us today is Ogun, our colleague, who is working as a software engineer on the platform.

00:02:23 Key takeaways

Welcome everyone. As I said, my name is Maarten-Jan Kallen. I'm a part of the ActivityInfo team and today I wanted to introduce you to a couple of best practices for data visualization in monitoring and evaluation. My background is mathematics, so I've spent quite a bit of my career creating data visualizations. This webinar shouldn't be seen as a complete reference of "how-to" for monitoring and evaluation applications, but I want to give you a few key takeaways.

The first one is that you should put thought and care into how you visualize your data. Before you embark on creating the visualization, think about three aspects which we will cover in this webinar: your audience, the purpose of your visualization, and the medium that you want to use. Remember, there's no wrong or right. A visualization is essentially a creative product, but I'll give you some do's and don'ts and some examples during this webinar. These examples will help you become more critical of visualizations that you see or that are presented to you in project reports or infographics.

I have an overview in three questions: Who is the audience for your visualization? What is the purpose of your presentation? And what medium would you use for your visualization? Those three questions will help you shape your visualization, and you should ask those questions before you get to the question of which type of graphic, colors, or icons to use.

00:04:52 Who is the audience?

The first question is: who is the audience of your visualization? Your audience really shapes your visualization. Think of individuals like you and me, partner organizations, donors, a management team, or researchers. These are typical examples of different types of audiences, and it is good to keep these in mind as this can guide you in selecting the type of visualization and its complexity.

My first best practice is to use the right level of detail. Don't add more information or detail than your audience can digest in the time that is available to them. For example, looking at a graphic from The Economist, it is very pretty with lots of colors and detail. It contains scales, text, and subtle graphs. This is a complex graph designed for people to take their time and digest the information. It draws the attention of the reader, contains long labels, and highlights specific parts.

In contrast, looking at an example from a Save the Children annual report, the chart is designed for the casual reader. It contains a lot less detail and is very simple. It is meant to draw the reader's attention by its use of colors but is kept simple on purpose. It's not meant to grab the reader's attention for a long period of time. Underlying detailed data can be added in an annex to the report.

00:08:18 What is the purpose?

The second question is: what is the purpose of your visualization? The purpose of your message shapes your visualization. Potential purposes include marketing or fundraising, accountability towards donors, reporting on project progress, or lobbying to influence decision-makers. These purposes will help you choose the correct graphic and what type of information to include.

A typical visualization in Monitoring and Evaluation is a graphic to demonstrate a funding gap. The purpose of visualizing such a gap is fundraising—presenting the fact that you need more financial support. You might also use this in a progress report to flag that the project is underfunded and at risk of failure. Finally, you might use this information to lobby politicians or decision-makers to ensure governments follow through with their promises.

An example from a UNICEF situation report shows a donut chart representing a funding gap. The red part is large, representing the gap, while other sections show funds received or carried forward. Another example from The New York Times uses a treemap to show spending categories. It displays the relative size of different categories and visually demonstrates the need to shrink a budget. If you want to present a funding gap, a treemap can show requested funds next to funds received with the correct relative size.

Another common visualization in M&E is progress, such as realized versus targeted values. The purpose could be accountability, showing donors that funds are being put to use, or reporting, stating that a certain percentage of targets have been met. An example from a UNHCR dashboard uses a meter or gauge chart to present progress towards target values. The darkness of the color indicates how close the indicator is to being 100% realized.

Match the chart to its purpose. A static view, like a meter, is great for accountability to show where you are right now. However, it doesn't tell you how you got to that value. If you want to show progress over time, a line chart with a time axis is more appropriate. It can show if progress was made early on and stalled, or if it started slow and gained traction recently. Another option is a bullet chart, which presents progress towards targets in a slightly different way, often using icons to mark status from previous periods.

00:19:13 Choosing the medium

The third aspect to consider is the medium: where will your visualization live? The most common types are print (reports, articles, posters) and web (dashboards, interactive infographics, blogs). Each medium gives you possibilities but also restrictions. Choose the medium that allows you to best present your story.

Dashboards provide benefits for presenting data because they are generally interactive. You can click on elements to open up more information, and they can be updated continuously. Dashboards are great to share with donor and partner organizations because they can drill down to select their data of interest. Interactive maps are a great example of dashboards, allowing the audience to see where activities are taking place and their impact.

A best practice here is to acknowledge a partner's effort. When you ask partners to provide data, it is often a duplicate effort for them. A great way to acknowledge this is to allow them to filter their data in your dashboards. This allows them to see why they are providing reports and share that information with their own donors. For example, a dashboard of the refugee response in Iraq includes a separate section specifically for partners to drill down into information linked just to them.

00:24:32 Chart types and colors

Once you have considered the audience, purpose, and medium, you can decide on chart types and styling. There is no right or wrong, but certain charts work well for certain cases. For comparing values, use bar charts, pie charts, or treemaps. For progress, use dashboard meters or bullet charts. For geographic values, use maps. For time-based values, line charts or bar charts work well.

Choose colors wisely. Be aware that color can have certain connotations for different groups of people. For example, red can imply danger or something negative. In a perception survey graphic, using red for female respondents might be unfortunate if it unintentionally conveys a negative association. Think carefully about what colors may invoke with your audience. Often, you can choose colors that match your corporate style.

For maps or presenting categorical or sequential data, use resources like ColorBrewer to choose a suitable color scheme. You can choose the number of classes and select a scheme that is color-blind safe.

00:28:32 Do's and don'ts

I want to show you some examples of do's and don'ts. Let's look at the donut chart of the funding gap again. The labeling in the original example was a bit off, and the use of red conveyed urgency. I recreated the chart using Google Sheets to improve the labeling, adding lines connecting labels to slices and including percentages. However, small slices might not show values automatically.

A more styled version created in Inkscape allows for different choices. I used light gray to indicate missing funds, more narrative labels with colors linking to the slices, and pulled out sections for clarity. To do this, you can create a base chart in a spreadsheet program, save it as a vector format (PDF or SVG), and then open it in a graphics editing program like Inkscape to change colors and add labels.

A crucial best practice is: do not make numbers look better than they are. Include all relevant data. In the funding gap example, a table in the report showed "other resources used," which were excluded from the donut chart. Adding this data changes the visualization significantly, showing a smaller funding gap. It is generally good to put all relevant data in your chart.

Provide a link to the source data and provide your data in an easy-to-use format. Also, use proportional scales. Cut-off scales distort relative sizes. For example, a Fox News chart on Obamacare enrollment cut off the y-axis, making the enrollment look lower than it was. Starting the y-axis at zero shows the relative sizes more accurately. Similarly, in the NGO annual report example, resizing slices of the donut chart to make management costs look smaller is misleading.

00:37:55 Q&A session

Alex Bertram: Thank you so much, Maarten-Jan. We have a number of questions. Carlos asks, "Is visualization a manner to avoid long narratives? To what extent does visualization complement narratives?"

Maarten-Jan Kallen: It certainly is a way to avoid long narratives. You are allowing your audience to create their own narratives of the data and form their own opinions. Geographic data, for example, is cumbersome to explain in text. However, don't underestimate the power of tables; many bar charts can be replaced with a table which is much more informative.

Alex Bertram: Lulu asks, "Can more than one chart be used for the same purpose?"

Maarten-Jan Kallen: Certainly. Visualization is a creative process. Different charts can provide different angles on the information. You could present both a meter and a line chart if you have the space and your audience has the time to digest it.

Alex Bertram: Mazhar Hassan asks, "When exactly do you use a table and when do you use a graph?"

Maarten-Jan Kallen: Visualization draws the attention of the reader and breaks up large blocks of text. It allows you to use corporate colors and gives cachet to reports. However, if you have too many categories, a bar chart becomes too large and hard to read. In that case, a table in an annex may be more useful.

Alex Bertram: Josh Bowlin asks, "Should y-axis labels on charts with percentages always be zero to 100%? Sometimes charts automatically reduce the axis, but I was told this is misleading."

Maarten-Jan Kallen: I would say it is quite dangerous to start percentages at anything other than zero. I would certainly avoid doing that.

Alex Bertram: Saburi asks, "Is it good practice to vary the type of visualization for one type of information that may be in a different slide?"

Maarten-Jan Kallen: Yes, certainly. If the analysis is complex, viewing the data from different angles can be helpful. Visualization can help you understand what is going on, such as looking for correlations.

Alex Bertram: We have many questions about software. While we work for ActivityInfo, which offers built-in visualization, it is less about the software and more about how you apply the options. Whether using Excel, Power BI, or Inkscape, the key is paying attention to how you use those options. Maarten-Jan, can you elaborate on the Inkscape process?

Maarten-Jan Kallen: Inkscape is open-source and free. The process involves generating a chart in a tool like Google Sheets or R, saving it as a vector file, and opening it in a graphics editor like Inkscape or Adobe Illustrator. This allows you to change colors and add labels, though these tools have a learning curve.

Alex Bertram: Babacar asks, "Is it better to use counts or percentages to visualize data?"

Maarten-Jan Kallen: That is an interesting question. When looking at counts on a map, you often just see population density—high numbers in populated areas. A better approach is often to look at numbers per thousand inhabitants to create a level playing field. Percentages might be a better option in some cases than absolute values to assess relative size.

Alex Bertram: Meseret asks about choropleth maps and Sankey charts.

Alex Bertram: Regarding choropleth maps, the ColorBrewer tool is a great resource. It helps distinguish between qualitative scales (for categories) and sequential scales (for values like rates). It also handles divergent scales (e.g., budget surplus vs. deficit). It is important to remember that human perception is limited; people generally can't distinguish more than seven to ten colors, especially those who are colorblind.

Alex Bertram: Malaya asks, "For simple data, such as actual versus target, is it not better to use a narrative or a table?"

Maarten-Jan Kallen: Absolutely. If you are simply listing realized and target values, a table is perfect. Many people are compelled to use attractive charts, but if you are just reporting information, a table is often sufficient.

Alex Bertram: Hanne suggests mentioning the importance of taking audience background into consideration.

Maarten-Jan Kallen: Researchers have much more skill in reading complex visualizations. If you present those to the general public, they may have difficulty reading them. If your purpose is to inform the general population, it is better to use simple icons and numbers.

Alex Bertram: Abu Hafs asks, "Should we attach or mention our sources for visualization?"

Maarten-Jan Kallen: When dealing with researchers, donors, or management, it is well worth mentioning the source. In ActivityInfo dashboards, we usually mention that the data is taken from the platform.

Alex Bertram: We had questions about the Iraq dashboard. It was made using Power BI. Dashboards are great for communicating with partners and donors who need to drill down into data. However, for a general audience just interested in what is happening, these charts might be too complex and use too much specific terminology.

Alex Bertram: We have come to the end of the hour. Thank you everyone for joining and being active in the chat. Please take a look at ActivityInfo if you have a chance. We hope to see you at the next webinar on Regular Expressions. Thank you, Maarten-Jan.

Maarten-Jan Kallen: Thank you, Alex. Thank you everyone for joining. Bye.