Skip to content

Data visualisations for M&E reports: visualising survey outcomes

In a previous article, I described three examples of data visualisations for types of outcomes and impacts that are often reported in Monitoring & Evaluation (M&E) reports. Visualising data in M&E reports can help to better understand the data, can increase engagement of stakeholders, and improves the overall readability of the M&E report.

Here I will share some data visualisations specifically for survey data since surveys are often conducted to monitor or evaluate a project. I provide some examples of charts that can be used for this purpose, but of course which type of visualisation best fits depends on the type of data and the information you want to convey. Generally, it is a good practice not to try to put too much information into one graph or figure and to include a clear descriptive title for all data visualisations. For survey data, having a conclusion of the data as a descriptive title can help ensure that the data and visualisations are not misinterpreted. I also recommend to stick to one colour or colour palette throughout the M&E report so that it does not distract from the information you want to convey.

Survey data can be difficult to visualise, because of the length of surveys. It may be a good idea to select certain, more important or more relevant questions from the survey and only visualise those. When visualising survey data, it is also important to use different types of graphs for different survey results. For instance, questions with a binary answer option may best be visualised using a donut graph, while questions with a Likert rating scale may best be visualised using a stacked bar graph. When comparing different group’s answers on the same questions, it may be best to use a bar graph.

On a more general note, whenever presenting data, it is important to include information about how many data points were collected and a description of the methods used to collect the data. If the sample size is very small it can be misleading to show percentages in data visualisations. Similarly, cropping an axis can be misleading and should be avoided. It may also be necessary to provide a breakdown of specific groups within the data, for instance women and men or different ethnic or age groups.

As in my previous article on data visualisations for M&E reports, all graphs below have been made using Python, Matplotlib and Seaborn. All data is fictitious.

Visualising binary questions or the breakdown of a category: donut graph

Surveys often contain binary questions, for example Yes/No questions. These can be visualised in different ways, for example using a stacked bar graph (see figure 4 below) or a donut graph. Donut graphs are very useful when visualising a small number of categories, typically not more than 5 or 6.

The first donut graph below shows the breakdown of which training(s) project beneficiaries participated in. The program had 1200 program beneficiaries and provided six different trainings. Program beneficiaries could participate in one or more different trainings.

Figure 1. The percentage of program beneficiaries participating in each of the trainings.

The second donut graph (figure 2) shows the same as the first donut graph, but also includes a breakdown of female and male participants in each of the trainings.

Figure 2. Breakdown of female and male participants in each of the trainings.

Visualising differences between groups: stacked bar graph

When analysing survey data in an M&E report, it is common that two or more groups are compared to each other. A stacked bar graph is a nice and easily understandable way to present comparisons between groups on certain outcomes, for instance agreement with statements or satisfaction ratings.

In figure 3 below, you see a stacked bar graph that shows the agreement with the statement “Women and men are equally good leaders” across different sectors. The graph nicely shows the differences in agreement with the statement between different groups, in this case employees from different sectors. The mean line indicates the mean of the agreement ratings.

Figure 3. Agreement with the statement that women and men are equally good leaders differs across different sectors (in %).

Figure 4 also shows a comparison between groups, but in a slightly different way. The outcome in this example is employment status. The figure shows who among those who completed a certain training found employment within three months after completing the training and who did not. Figure 4 is also an example of how to visualize binary data (in this case Employed/Unemployed). Since there are different groups (participants of different trainings) and the emphasis here is more on the comparison between groups, a stacked bar graph is a good visualisation.

Figure 4. Among those who received training, here is who found employment within 3 months after completing the training and who did not (in %).

These were some data visualisations for survey data that are commonly reported in M&E reports. It is important to choose the right chart type to ensure that the visualisation helps to effectively communicate the data in the M&E report. I will provide additional data visualisations for M&E in a future article.